Professional Documents
Culture Documents
It's Thanksgiving time, and I'm surely thankful for the free open source software I use. But going
open source always seemed counter-intuitive to me. Why would a company invest time, money
and development resources to create valuable intellectual property and then throw it out to
everyone to use for free as they see fit?
Especially if they see fit to use your open source software against you in competitive products.
That's reportedly what happened with mobile dev company RoboVM.
"We have seen competitors actively exploiting our good faith by using our open source code to
compete with us directly in commercial products," read a recent Google Groups post from
RoboVM exec Mario Zechner about the company's decision to end its open source efforts.
Henric Mller, CEO, wrote his own post on the matter. "We have received no notable external
contributions to the components that represent the core of our product," he wrote, which "means
that neither RoboVM customers nor us, as the maintainers of RoboVM, have realized any
benefits to sharing the products source under such liberal terms." (You can read more details
here.)
Which got me to thinking. Why isn't this happening to more companies? You always hear about
the many benefits of open source, but exactly how are companies benefitting from creating
products for free?
Some researchers recently tackled the question, publishing an official academic study examining
the "Motivation, values, and work design as drivers of participation in the R open source project
for statistical computing." But it's full of research-ese nonsense like "The data are analyzed using
item response models and subsequent generalized linear models, showing that the most
important determinants for participation are a hybrid form of motivation and the social
characteristics of the work design. Other factors are found to have less impact or influence only
specific aspects of participation."
I don't know what that means and I'm not paying for the report to try to find out. Why can't they
talk in plain English? Why can't they open source their research?
On Quora, where they talk in plain English, a question asked "How rewarding is open source for
a programmer's career?" One answer included this: "The reason I 'open source' my code, or at
least talk about it at length, is -- I find it rewarding that I am lessening the suffering of other
developers and that others may go on to discover new things thanks to work I have done."
Another answer said: "Many of us program for the sheer love of programming. The money is just
gravy."
How is that helping the company? I can see the obvious things: community goodwill, greater
exposure and branding, loss leaders for profitable products and so on. But how much does that
help the bottom line, and how do you even measure any financial impact?
Open source champion journalist Steven J. Vaughan-Nichols shed some light on the matter in a
recent Redmond Magazine cover story, "How Open Source Is Shaping Microsoft's Future." He
quoted Allison Randl, president of the Open Source Initiative, as saying companies need to go
open source or get left behind in the software development game.
"Microsoft knows this and is aware it can no longer take a 'not invented here' attitude any more,"
Vaughan-Nichols wrote. "Microsoft's new management realizes the old proprietary software
business model that served it well for so many decades has been milked as much as it can be.
Hence, Microsoft has been moving to a service-oriented business model and, at the same time,
the company is moving to open source software to power those services."
I don't know. It all sounds good, but I still don't see a clear profit motive. So I went looking for
more answers.
"So, they should be using their open-source code in their own commercial products," Vaughan-
Nichols continued. "They should be spending their marketing dollars on explaining why their
commercial adaptation is better than their competition's, expanding their support services. If
someone is actually ripping off their code and putting it into a proprietary product, it's time to
turn to the lawyers. Depending on the license they're using, there's almost always a way to throw
water on someone who's simply stealing code."
An article on readwrite.com offered "5 Reasons Your Company Should Open Source More
Code," boiling down to: great advertising; a force multiplier (if others work on your project);
attract talent; best technical interview possible for companies that are hiring; retaining talent.
Another Quora question asked "How would companies make money if all software was open
source?" One answer read:
Ads: Twitter, is built using open source code. Also signing up for it is free. In spite of
that, it is valued at $18bn+. And 85 percent of its revenue comes from advertisements.
Hardware: Android, one of the most widely used mobile operating system is open source.
But Android also needs hardware to run on. Individual hardware making companies mod
android according to their needs (and also pack in a few proprietary software sometimes)
and ship their devices.
Services: Netflix is open source too. But it does charge a sum of money for its services.
Also Wordpress, being open source, charges for domain names and hosting spaces.
Value added contents: While open source codes are free to download and use, it is also
possible to charge for premium content. Like Wordpress gives you a set of themes to
work with, but it gives you the option to buy themes too!
Divio recently explained "Why we support open-source software." "It's not always clear to
people why companies like Divio -- businesses that need to make a profit -- would want to give
money away to open-source communities and projects," the article states. "Giving money away,
after all, is not the first thing one would put on a business plan."
Part of the reason Divio does it is community goodwill, along with return on investment that
might not be immediate and might not be obvious, like long-term community investments. "The
fact that many companies -- successful, profitable, sensible ones, like Divio -- do make those
long-term investments in the community is an indication that this is a wise way for them to spend
their money," the article states.
Wired recently wrote that "Open Source Is Going Even More Open -- Because It Has To." "Why
are so many companies giving away their intellectual property?" the article asks. "It's not
happening for altruistic reasons."
Rather, Wired said in discussion about Google's decision to give away rights to the Kubernetes
cloud computing system, "Companies like Google want others to use their open source software
since it can help drive the use of online services, like Google's cloud computing tools. They want
others to contribute code to this software too. But increasingly, others don't want to use or
contribute to projects unless they're independently managed."
So you get the idea. A lot of "soft" reasoning, but not much hard data on how open sourcing your
software contributes to the bottom line.
"Levine says the conventional open source business model is flawed: Open source companies
that charge for maintenance, support, warranties and indemnities for an application or operating
system that is available for free simply can't generate enough revenue," the article states.
So I'm back to square one. Divio says the fact that successful companies are doing it proves its
value. A venture capitalist says it can't succeed. Microsoft is doing it in a big way. Apple isn't
doing it much at all. Red Hat is living it. What's the answer? Where's the data?
You tell me. Comment here or drop me a line. Why are for-profit companies open sourcing their
software? I'm crowd-sourcing the open so
4 reasons companies say yes to open source
Open source isn't just about saving money -- enterprises are
adopting it to develop applications faster, with higher quality
components.
By Howard Baldwin
| Follow
When individual developers think of open source, they think "free." And with good cause: Who
in their right mind wouldn't be interested in technology that they can get at no cost and use with
few licensing restrictions?
When companies think of open source, these days they think "business agility," a quality they
increasingly value above all others in the fast-changing marketplace.
The ability to create new applications quickly, reliably and economically is drawing businesses
big and small to open source and emboldening them to use it for ever-larger projects, IT
practitioners say.
Which is likely why open source's popularity is booming (with a few holdouts). According to the
Forrester Research report "Development Landscape: 2013," 76% of developers have used open-
source technology "at some level," says Jeffrey Hammond, a Forrester analyst specializing in
application development and delivery.
Here are four key reasons why organizations of all sizes are taking open source seriously.
Cost savings may be only part of open source's allure, but it's still a big part, no matter what size
the organization. "How can Netflix charge as little as $8 per month for its service?" Hammond
asks. "Because everything is built on open-source software. They focused on content, not
building an operating system or a testing framework."
"It's like using Spring, JBoss or Drupal for content management," Hammond says. "Companies
get the 'Lego blocks' for free, so they can spend their time and resources building what they want
in particular." Enterprises have always customized packaged software such as ERP applications,
except now, with open source, that customization is less expensive.
Indeed, in some cases, open source is helping to bring back custom development of applications,
an option that has decreased in popularity in the past 10 years or so as the use of commercial
applications and software-as-a-service offerings gained ground.
Taking on some custom development work in order to save money appealed to Carestream
Health. A $2.5 billion provider of dental and medical imaging systems with 8,000 employees,
Rochester, N.Y.-based Carestream wanted to consolidate the data from its worldwide
manufacturing facilities into a single product life-cycle management (PLM) application to reduce
new product development and manufacturing time by routing information more efficiently.
"We wanted consistent management of product-related information across our global company
footprint," says David G. Sherburne, director of global R&D effectiveness and engineering IT at
Carestream. "With a modern platform in place that could be built upon into the future, we were
expecting a 5% productivity gain through the integration of existing point solutions and the
elimination of manual process step
Carestream chose Aras, an Andover, Mass.-based PLM vendor that uses an open-source model to
encourage its customers to develop and share new components with one another. Aras had "the
best functionality for a reasonable total cost," says Sherburne. "It didn't have some of the
functionality we needed, so we knew we'd have to do some extra development, but when we
completed that, we knew we could deploy it globally from a fixed-cost perspective."
There were no upfront capital licenses costs, which allowed Carestream to move forward without
having to purchase and inventory licenses. "The subscription model allowed us to enter into the
PLM project and focus on proper implementation," says Sherburne. "It provided a fixed-cost
platform that can be enhanced over time and scaled to allow more collaborative access without
continued cost outlays."
As projected, Carestream came out ahead: Its ongoing costs for approximately 1,500 users (1,000
internal, 500 suppliers) when the software is fully deployed are at "the low end of six figures,"
says Sherburne, as opposed to "millions of dollars upfront" for a packaged application, not
including ongoing maintenance, he says.
Big businesses aren't the only organizations that benefit from open source's cost structure. The
economics mean that smaller entities with niche software requirements can get what they need in
a cost-effective package.
Over the course of several years, Wiss evaluated a variety of healthcare-oriented software
packages, but none offered the specific functionality she was after. Few SaaS applications met
her needs because of the uniqueness of her specialty, and those that might have were too
expensive, she says.
Wiss finally decided to grow her own, turning to an open-source developer whose one-time fee
was about the same as the cost of one year of access to some of the SaaS offerings she'd looked
at. "I was concerned because I didn't speak 'computer' well enough to tell someone what I
wanted. But [the consultant] said to forget what he did, and just tell him the way I work," Wiss
relates. (See "Open Source Bails Out Small Business" for details.)
Open source fans have long contended that the methodology produces better software. Their
reasoning: If code is flawed, the developer community can identify and address the problem
quickly, where a single coder might plod on unawares, at least for a while.
There is a consequential benefit [to open source] from both a reliability and a financial
perspective. Peter Richards, managing director of global banking,
Bank of America
That quality appeals to Bank of America. "We have a broader range of choice when it comes to
high-quality software," says Peter Richards, the bank's managing director of global banking in
New York. "There is a consequential benefit from both a reliability and a financial perspective."
Asked if it's surprising for such a large company to use open source, Richards cites Linux's path
to widespread acceptance in commercial organizations. In the beginning, he explains, enterprises
worried that Linux was a hobbyist's operating system, not one that a big corporation could
depend on. But over the years, "the number of people who support Linux through peer review
have made it into one of the better operating systems for corporations," Richards says.
"The quality of open-source code for development comes because of the number of people who
are able to contribute, review and test it," he asserts. "That means it's a solid piece of code." That
development structure also ties back into cost: "If you had to pay for that yourself, you'd end up
with enormous costs because you'd have to do testing and code review yourself," Richards adds.
"That's one of [open source's] big advantages: quality at a reasonable cost."
Not to be confused with agile development, business agility is the ability to react to marketplace
demands quickly. Open source provides this to developers and businesses alike by speeding up
the pace of software development.
Ron Pitt, the developer who worked with Development Is Child's Play's Wiss, is a partner with
software consultancy LevelHead Solutions in Poway, Calif. If he needs new code for a project,
he downloads it in minutes rather than developing it himself. "Sure, some of it's buggy, but I'd
rather spend 15 minutes debugging it than writing it from scratch in 15 hours," Pitt says.
Businesses likewise benefit from open source's ability to let them react quickly. For one thing,
companies that use open software code aren't tied to vendors' timelines for commercial
application upgrades. "If you have to wait for vendors to make the changes you want, it affects
the pace at which your company can innovate," says Mike Milinkovich, executive director of
Ottawa-based Eclipse, an open-source community for individuals and organizations focused on
tools originally launched by IBM for Java.
Madhu Nutakki, vice president of digital presence at Oakland, Calif.-based healthcare provider
Kaiser Permanente, concurs that open source brings value in the form of flexibility.
[Open source] gives us more flexibility when we release updates more frequently. Madhu
Nutakki, VP, Kaiser Permanente
Kaiser Permanente has been using the GitHub source code control system since 2011. "It was
built by developers for releasing code in an expedited way. It gives us more flexibility when we
release updates more frequently," says Nutakki. (Note that while GitHub also works with
proprietary development tools, Kaiser uses it primarily for open source deployment.)
"We started using GitHub because our paradigm changed to a faster release model," Nutakki
explains. The healthcare provider's increasing push into mobile means that it's now serving
customers who have higher expectations for frequent updates. "We used to build large
applications with a release cycle of every six months. Over the last two years, we do releases
more quickly -- monthly, quarterly and even faster," he says. "With other products we were
using, it took much longer to do a build. With GitHub, it takes an hour."
Forrester analyst Hammond confirms that open source's speed advantage is making it more
popular in enterprise IT development. "If you ask a developer how they're going to handle a
specific project, they can respond that they don't have to buy specialized hardware, because they
can run it on Linux. They can use an open-source development framework, and they can develop
what someone needs specifically."
Open source also brings a lot of "elasticity" to the process of spinning up new resources,
Hammond says. "You don't have to ask 'Do I have a license?' or 'Do I have to buy more
software?'" he says. That's why there's a high correlation between cloud-based and open-source
software, he points out -- both provide a scalability and flexibility that companies haven't had in
the past.
Another, perhaps unsung, benefit to using open-source tools, and thereby reducing dependence
on a single or multiple vendors, is that the open-source option may reduce business risk.
Milinkovich notes that when the company developing TOPCASED, a development tool for
embedded systems, was acquired, "the developer stopped working on it." So the companies that
used it and loved it, notably Airbus, banded together to fund other developers to continue
supporting it.
Vendors come and go, and commercial priorities change, whereas a community's focus is more
constant. "The openness and transparency of open source mitigates a lot of risk," says
Milinkovich. "Whether a company is big or small, it'll stop developing code if it's no longer
commercially viable, and you no longer have access to the source code and repositories. If you
can actually get a vibrant community built up around your code, it's much more resilient than a
strictly commercial enterprise."
Gerald Pfeiffer, director of product management for Nuremberg-based SUSE, which offers
enterprise Linux, believes that open source is thriving for all these reasons.
"People are reaping cost benefits by using open source, but that's not the No. 1 priority. It's also
the avoidance of lock-in, the ability to customize, the ability to have a better feel of what you're
paying for. It's the combination of all that," Pfeiffer says. "You're sharing development costs with
other people, so you get more diversity and more independence than from a single vendor."
She used Google Calendar so that if one parent cancelled an appointment, other parents could
quickly see newly available slots -- but she also had to synchronize that calendar with a paper-
based calendar that the therapists used. For the sake of efficiency and accuracy, she needed an
electronic application.
She investigated options that addressed billing but not scheduling, accommodated sole
practitioners instead of multiple practitioners, or focused on tracking medical issues not germane
to occupational therapy. Some software was customizable but not user-friendly. Wiss marveled
that, even in the midst of Silicon Valley, "I couldn't find something I liked at a price I could
afford."
Finally, Wiss was introduced to Ron Pitt, a Poway, Calif.-based consultant. He understood her
frustration. "When you have a small business like hers, it's hard to commit to thousands of
dollars upfront and then monthly when your income fluctuates," says Pitt. He agreed to custom-
build an application for Wiss using Linux, Apache, MySQL, PHP, and the NetBeans IDE. The
cost: $5,000 plus a few hundred dollars for hosting and backup each month, about the same as an
annual fee for a SaaS application.
Pitt retains the rights to the code so he can create another application for another occupational
therapist if he wants. He says he was able to charge just $5,000 because the code is "free,
modular and the tools are robust. It's good, solid software engineering."
Frequent contributor Howard Baldwin last wrote for Computerworld about how to get a job in
financial IT.
This article, "4 Reasons Companies Say Yes to Open Source," was originally published on
Computerworld.com.
How much more? (Almost) everything, to quote GitHub co-founder Tom Preston-Werner. The
post is a few years old, but judging from how the industry continues to treat software like
something to be hidden, not nearly enough people have read it.
Reasons To Open Up
If it sounds like business model suicide to open source (almost) everything, its not. At least,
not for the 99.999% of companies that sell services, not software. For this overwhelming
majority, Preston-Werner offers several reasons to open source code:
1. Open sourcing code is great advertising for you and your company
translat[ing] into goodwill for [your company] and more superfans than ever
before;
2. If your code is popular enough to attract outside contributions, you will have
created a force multiplier that helps you get more work done faster and
cheaper. More users means more use cases being explored which means
more robust code;
3. Smart people like to hang out with other smart people. Smart developers like
to hang out with smart code. When you open source useful code, you attract
talent;
4. If youre hiring, the best technical interview possible is the one you dont
have to do because the candidate is already kicking on one of your open
source projects; and
5. Once youve hired all those great people through their contributions,
dedication to open source code is an amazingly effective way to retain that
talent. Lets face it, great developers can take their pick of jobs right now.
These same developers know the value of coding in the open and will want to
build up a portfolio of projects they can show off to their friends and potential
future employers.
Keep in mind that open source really only helps you with a particular kind of audience. Chances
are, open source wont appeal to your end customers, whatever your product happens to be. But
thats OK, because open source is really just a way to engage the people who will help you build
your product.
Self-Selection Bias
Still not convinced? Take a look at Facebook, the company Ive called the worlds largest open
source company. In an excellent Fast Company article, Facebook open source chief James Pearce
touts the benefits it derives from open source, most notably recruiting:
It turns out that large percentages of our engineers will have known about our open-source
projects before they will have joined and they will say that it contributed positively to their
decision to join the company. Its a great window into the world of the sorts of problems that
we solve, and of course were hoping there are world-class engineers around the world who
would relish those kinds of opportunities and when they see the problems were solving will feel
the urge to take a look.
Ive written before about the importance of open source to recruiting the industrys best
engineers. Much of todays best software is written in the open, be it Hadoop or Spark or
MongoDB or Android. Its how companies engage with developers long before they engage with
end-customers.
Its also how companies learn from their peers, as Box did from Facebook.
Its why Netflix bothers to write technical blog posts that only a geek could love. (There are a
couple of things you should think about though before trying to score 100 billion records in
Pig.)
Its why your company needs to stop hoarding software that offers no competitive advantage and
start sharing. It will help you attract and retain talent which, in turn, will help your company
better serve end customers.
Thats how open source helps. And its why you should give (almost) all of it away. Now.
2) Economy. With increased competition and current state of economy, it makes economical
sense for even large companies to shift some of the software engineering efforts (development,
testing, localization, etc.) - and their costs - to external "workforce".
6) Flexible licensing. Therefore, companies balance their liking of / belief in open source (or
desire to achieve competitive advantage, or both) with profit-oriented activities, which might be
in parallel to open source involvement (having commercial and open source products in
portfolio) or include open source products in their portfolio as commercial offerings (under
double or commercial license).
7) Flexible cost structures. In addition, do not forget that today pre-sales consulting as well as
post-sales services and technical support represent a large share of potential revenue sources.
"Losing" money by providing some open software priced free or less expensive than competition
is frequently much less than amount of money the companies might earn by charging for these
services.
8) Liking (believing in) open source approach. Certainly, many companies, especially
software engineering people, working there, believe in open source approach. Many members of
their teams grew up in open source (hacker) culture and they just cannot stop this "addiction".
Moreover, many people rightfully feel that, since they and their companies frequently use others'
open source software - results of other people's efforts, they should "return the favor". The
reciprocity is an important component of open source culture, be it in a form of participating in
open source software production, or helping each other with questions or problems via forums,
mailing lists or other means.
1.7k Views View Upvotes
10 Reasons Open Source Is Good for Business
With the many business and government organizations that now use open source software such
as Linux, it's becoming increasingly clear that price is not the only advantage such software
holds. If it were, companies that adopted it during the Great Recession would surely have
switched back to the expensive proprietary stuff as soon as conditions began to ease, and that's
clearly not the case.
Rather, free and open source software (FOSS) holds numerous other compelling advantages for
businesses, some of them even more valuable than the software's low price. Need a few
examples? Let's start counting.
1. Security
It's hard to think of a better testament to the superior security of open source software than the
recent discovery by Coverity of a number of defects in the Android kernel. What's so
encouraging about this discovery, as I noted the other day, is that the only reason it was possible
is that the kernel code is open to public view.
Android may not be fully open source, but the example is still a perfect illustration of what's
known as "Linus' Law," named for Linus Torvalds, the creator of Linux. According to that
maxim, "Given enough eyeballs, all bugs are shallow." What that means is that the more people
who can see and test a set of code, the more likely any flaws will be caught and fixed quickly. It's
essentially the polar opposite of the "security through obscurity" argument used so often to
justify the use of expensive proprietary products, in other words.
Does the absence of such flaw reports about the code of the iPhone or Windows mean that such
products are more secure? Far from it--quite the opposite, you might even say.
All it means is that those products are closed from public view, so no one outside the companies
that own them has the faintest clue how many bugs they contain. And there's no way the limited
set of developers and testers within those companies can test their products as well as the
worldwide community constantly scrutinizing FOSS can.
Bugs in open source software also tend to get fixed immediately, as in the case of the Linux
kernel exploit uncovered not long ago.
In the proprietary world? Not so much. Microsoft, for example, typically takes weeks if not
months to patch vulnerabilities such as the recently discovered Internet Explorer zero-day flaw.
Good luck to all the businesses using it in the meantime.
2. Quality
Which is more likely to be better: a software package created by a
handful of developers, or a software package created by thousands of developers? Just as there
are countless developers and users working to improve the security of open source software, so
are there just as many innovating new features and enhancements to those products.
In general, open source software gets closest to what users want because those users can have a
hand in making it so. It's not a matter of the vendor giving users what it thinks they want--users
and developers make what they want, and they make it well. At least one recent study has shown,
in fact, that technical superiority is typically the primary reason enterprises choose open source
software.
3. Customizability
Along similar lines, business users can take a piece of open source software and tweak it to suit
their needs. Since the code is open, it's simply a matter of modifying it to add the functionality
they want. Don't try that with proprietary software!
4. Freedom
When businesses turn to open source software, they free themselves from the severe vendor
lock-in that can afflict users of proprietary packages. Customers of such vendors are at the mercy
of the vendor's vision, requirements, dictates, prices, priorities and timetable, and that limits what
they can do with the products they're paying for.
With FOSS, on the other hand, users are in control to make their own decisions and to do what
they want with the software. They also have a worldwide community of developers and users at
their disposal for help with that.
5. Flexibility
When your business uses proprietary software such as Microsoft Windows and Office, you are
on a treadmill that requires you to keep upgrading both software and hardware ad infinitum.
Open source software, on the other hand, is typically much less resource-intensive, meaning that
you can run it well even on older hardware. It's up to you--not some vendor--to decide when it's
time to upgrade.
6. Interoperability
Open source software is much better at adhering to open standards than proprietary software is.
If you value interoperability with other businesses, computers and users, and don't want to be
limited by proprietary data formats, open source software is definitely the way to go.
25% off 100-pack iCloth Screen Cleaning Wipes for Electronics - Deal Alert
46% off Foval 150PSI Car Digital Tire Pressure Air Gauge - Deal Alert
7. Auditability
With closed source software, you have nothing but the vendor's claims telling you that they're
keeping the software secure and adhering to standards, for example. It's basically a leap of faith.
The visibility of the code behind open source software, however, means you can see for yourself
and be confident.
8. Support Options
Open source software is generally free, and so is a world of support through the vibrant
communities surrounding each piece of software. Most every Linux distribution, for instance,
has an online community with excellent documentation, forums, mailing lists, forges, wikis,
newsgroups and even live support chat.
For businesses that want extra assurance, there are now paid support options on most open
source packages at prices that still fall far below what most proprietary vendors will charge.
Providers of commercial support for open source software tend to be more responsive, too, since
support is where their revenue is focused.
9. Cost
Between the purchase price of the software itself, the exorbitant cost of mandatory virus
protection, support charges, ongoing upgrade expenses and the costs associated with being
locked in, proprietary software takes more out of your business than you probably even realize.
And for what? You can get better quality at a fraction of the price.
If you're considering using open source software, it will typically cost you nothing to try it out
first. This is partly due to the software's free price, and partly due to the existence of LiveCDs
and Live USBs for many Linux distributions, for example. No commitment required until you're
sure.
None of this is to say, of course, that your business should necessarily use open source software
for everything. But with all the many benefits it holds, you'd be remiss not to consider it
seriously.
To comment on this article and other PCWorld content, visit our Facebook page or our Twitter
feed.
Why Big companies are embracing open source?
Himanshuz.chd | Aug 15 2012 | Visits (11522)
inShare
Tweet
Gone are those days when open source software were used only by start-ups and
small corporates. Today big companies are using and promoting open source
software. Google, the internet search giant uses Linux(modified to cater its own
needs) for most of its servers. Also Google is funding open source projects like
chromium which forms the backbone of a popular chrome web browser from google.
So we see that one of the biggest IT company is both using and contributing to open
source. This change in outlook of big corporates towards the open source has not
come overnight. There has been a long struggle for open source to make everyone
believe that it is as good as (or even better) than its proprietary counterparts. Here
in this article, we will discuss various reasons why big companies are embracing
open source.
The cost factor
Yes, this is one of the most important factor that attract not only the small
companies or start-up's but also the big corporates these days. The question here is
that who would like to spend extra money if the same (or even better) software is
available free of cost (or at a marginal cost). For example, if we take the example of
Microsoft OS. One has to buy the OS for a good amount of money then over that
one again has to buy good anti-virus and even after that the system does not seem
much secure. Also, the closed source software come with a per CPU license and
obviously, each license costs. Now imagine multi processor machines where only
one CPU is dedicated for a licensed software, companies still have to pay the license
fee for all the CPUs attached. This is sheer wastage of money. So, what's the point
of spending so much money there when open source alternatives can provide the
same or even better functionality and that too for no or very little cost? So we see
that the cost factor is one of the major attractions of OSS.
Security
In today's digital world when servers and websites of top notch technology
companies are repeatedly in news for getting hacked, security has become the
prime concern while selecting a software. The open source software that were once
discarded as being less secure are being preferred today on security aspect and
what more, many of the OSS(open source software) are being considered as
technically superior to their proprietary counterparts. Many surveys and statistics
have shown a considerable increase in use of OSS in areas where security has been
a prime concern.
Does that mean OSS are near to perfect? No, absolutely not. The main reason for
them to be more secure is the number of contributors. If you compare a team of 50
talented programmers with a team of a 1000 talented programmers, then whom do
you think will be able to produce a better quality product? Yes, you are right. The
number of resource contributing to a product makes a difference and in case of
popular OSS, the contributors are definitely very high.
The other reason for being secure is the code being open source. Anybody can
identify a bug or loophole and can quickly fix it. The open source code combined
with the fact that large number of contributing developers prepares an ecosystem
where bugs are fixed at a quicker speed than they are exploited. This results in a
secure software.
The big corporates now know that there is little security transparency while using
proprietary software as compared to OSS. More on this aspect is covered under my
article on open source security.
Technical superiority
This is something that the OSS speak on the top of their voice and in the face of
closed source software. I don't mean to say that proprietary software are of inferior
quality but when compared with their popular OSS counterpart, they somehow loose
their charm in terms quality. This win is attributed to large number of open source
contributors that give their heart and soul (and of course mind ) to their
contribution. Also, the same users when act as reviewers, spot and fix most of the
bugs in the software. This definitely increases the quality of product.
This again is one of the very useful aspect of an OSS. Anybody using it can modify
the source code as per the needs and usage. Google has already benefited from this
aspect by modifying the Linux kernel for its Android OS. In fact there are various
applications that run on Android which require some tweaks in Android source code.
So we see that, you don't have to fall back to the vendor for any and every change
that you want in the software. This saves time as well as cost. On the same lines, if
the source code for a software is available, any organization can quickly set up a
small team to audit and review its code to make sure that there is nothing
unwanted or buggy in the code. Though there is support available for most of the
popular OSS but big corporates always like to have the code verified by their own
professionals. So we see that the OSS provides this kind of platform.
Support
Usually for any popular OSS, there is a sea of support available on internet. There
are mailing lists, wikis, experienced users, official support staff, updates, patches
etc available very easily. So this large area of support provides a quick platform to
get your queries resolved within no time. This is in wide contrast with the closed
source software for which there is usually a single small team available doing all the
maintenance and support task. So we see that in this area too, the OSS has a upper
hand now.
To conclude, we see that OSS stand ahead of their closed source counterparts in all
the above aspects.
If you want, you can also go through a nice little research paper on Organizational
Adoption of Open Source Software that I found on internet.
Whenever a piece of source code for any software from Microsoft gets leaked, there is a flood of
warnings from security consultants that this can have huge negative impacts on the security of
that particular software. These types of warnings mostly lead (or rather mislead) to doubts that if
a small piece of code can cause security compromises then what about the open source software
for which the whole source code is available to any and everyone. A big question arises that how
do pro open source people advocate that its even more secure than proprietary software. Here in
this article, I will focus on some of the important aspects regarding the security of open source
software.
Please note that all the views expressed here are entirely of my own.
I usually get absorbed in this debate during tea breaks in my office. Most of the people argue that
if the whole source code is available then a cracker knows the loopholes and the security of that
software can easily be compromised. I agree to their point but quickly I point out that not
everybody out there is a cracker or a person with negative intentions. There are certainly more
people that support the cause of open source and contribute to it in a positive way. There are
more eyes that are protecting open source by identifying the loopholes and quickly correcting
them. So we see that the most powerful security asset of an open source software is the fact that
it is open source.
On the contrary, the belief that a closed source software is secure is not true. The live and biggest
example of this is the Microsoft family of operating systems. Everybody knows that these guys
do not release the source code but still we have huge number of bugs and security compromises
for Microsoft family of OS. Some people may argue that since it is very popular so it becomes
the prime target of crackers. Though there is nothing wrong in their statement but these people
try to deviate from the real issue. The real issue is not the number of compromises but the real
issue is that despite of being closed source, there ARE security vulnerabilities in closed source
software too and these vulnerabilities ARE being exploited in real world.
This is another very important aspect that businesses are concerned while dealing with open
source software. There is always a concern as to the quality of maintenance of an open source
software from a vendor. I believe this concern is valid but up to a certain extent only. Why I
agree to it is because there are still certain open source software that are managed by small team
or even a single person. So these software are still not preferred by corporates and businesses.
Now, why I said that I agree to it up to a certain extent only is because there is a long list of open
source software that are being used in real world by big businesses for real time work. Look at
Linux, Apache etc these are giving severe competition to their custom counterparts on every
aspect. The updates are provided at regular intervals and patches to critical bugs are provided
even in a few hours. For those who doubt the patch/update speed and capability of open source, I
would like to give the example of the Ping of Death. This bug resulted in a crash on almost every
operating system but the amazing part is that the patch to this bug was released only in couple of
hours on Linux. This can be attributed to the number of contributors to Linux. So we see that
though there are certain open source software that are still not mature to be used in real world but
we cannot paint every software with the same brush.
If we audit the custom or the closed source software on the same line, then we see that though
there is a dedicated team for a custom software but updates and patches are not as fast. Mostly
the updates and patches are not released within hours or days (as with many of the popular open
source software).
Other theories
There are some popular theories and rumours that state run agencies use open source software to
inject loopholes but I argue that why cant they use their influence on custom software vendors
for the same. There can be one or two instances where open source developers have publicly
accepted some wrong doings but as with everything, there is always a darker side and who
grantees that custom software is all pure and clean. At least within open source software one has
the privilege to review and verify the complete source code and clear any security doubts. Can
you do the same with closed source software when what you get is a bunch of binaries? What if
one day you find that the vendor of the custom OS that you use has deliberately introduced some
vulnerabilities so that the anti virus companies can survive and can mint money from you? :-)
So you see that bluntly saying that open source software is insecure is wrong. As open source
software is doing excellent in real world today.
Open source software (OSS), unlike proprietary software, is software that keeps the code open so
IT professionals can alter, improve, and distribute it. Although it has been around since relatively
early in the history of computers, in the past several years OSS has truly taken off, in what some
might see as a surprising example of a successful communal collaboration.
One of the biggest advantages of OSS is that as long as you have someone with the know-how,
you can completely customize the software to your needs. Because of the monies saved on
licensing fees, using OSS is beneficial even if you don't do any customization; however, if you
do customize, you'll need someone on hand who knows how to use open source software.
In addition, most proprietary software includes free tech support from the company, while OSS
companies charge for those services. But this fee is countered by the fact that the code is open
source, making it easier for your IT team to learn how to use it.
Another reason that people think they need too much support while using OSS is that this type of
software has a reputation for being relatively inaccessible to the average user. Unlike proprietary
software, which is tested extensively for user-friendliness, OSS software is often written by and
for computer professionals. Choosing your software wisely helps eliminate this potential issue.
You can negotiate with proprietary software companies about indemnification for intellectual
property (IP) infringement, but that's not possible with OSS companies. Plus, the rules that
govern IP for OSS are complex. However, you can counter this issue by purchasing
indemnification insurance through a third-party vendor like OpenLogic. In either situation,
reading the terms of the license helps mitigate the risk of committing IP infringement.
Often without strong central management, the OSS community must identify and provide
solutions for errors with the software. This leads some to worry that problems will not be fixed,
as compared to traditional software, which has centralized management and a dedicated team of
developers to fix any issues.
In reality, the opposite is true. Repeatedly, issues with OSS are quickly fixed thanks to the work
of the community members, whereas with a proprietary software company, users need to wait for
the release of the next software update to fix a bug.
This is perhaps the biggest misgiving that people express regarding OSS: Since the code is open,
any opportunist can identify and exploit the program through hacking and viruses. Proprietary
software companies, on the other hand, have team members dedicated to ensuring the security of
their software.
Some risk is associated with using any software, and the overall risk associated with OSS is not
higher than with any other type of software. While it's true that anyone can look at and
potentially exploit the code, it's also true that anyone can look at the code to identify potential
causes of security breaches and address them immediately. What's more, as long as someone is
on your team who knows how to use open source software, you can examine the software before
using it, and thus determine the level of risk associated with using it.
The concerns that people have about OSS are not completely unfounded, but each concern can
be mitigated with an understanding of the software in question. In many cases, using this type of
software helps companies save money while also getting a product that is better suited to their
needs. Once your company learns how to use open source software - and how to mitigate some
of the risks associated with it - you, like many others, may reap great benefits.
Originally posted on the OpenLogic Enterprise OSS blog. Reposted using Creative Commons.
Open source audits for IP issues
The adoption of open source software has helped every team deliver value faster but the variety,
volume, and risks have also grown, leading organizations to look inward to meet legal and
compliance obligations and eliminate security risks. The OpenLogic application audit service
analyzes internally-developed software for both known and unknown open source packages to
create complete bill of materials (BOM) and bill of licenses for open source components, helping
you understand the scope and depth of usage.
Once scanned, Rogue Wave aggregates the information and creates comprehensive reports to
give companies the information they need to make informed decisions about distribution,
security issues, and legal considerations.
Going one-step further, the Compliance Checklist service provides the critical information that
organizations need to fulfill their open source license compliance requirements. This is essential
for companies that distribute software or products containing software and need to assure
license compliance. The Compliance Checklist service does just that - it provides the framework
to verify OSS license compliance. With this service, you avoid customer objections and potential
litigation.
During an M&A transaction, its critical for both sellers and buyers to understand the IP profile
of any software assets, including information about any open source software that may be
included in proprietary code. Buyers can leverage the M&A Open Source Audit service to
identify intellectual property conflicts and ensure that they understand any embedded open
source licenses and obligations. Sellers can use the M&A Open Source Audit service to provide
accurate disclosures and ensure mergers and acquisitions move forward smoothly and without
asset devaluation.
After almost three decades of development, open source software has firmly crossed over into
mainstream use. Companies understand the unique value derived from software developed
through open communities and are welcoming its use in mission critical settings throughout the
enterprise.
Companies that adopt open source are in a prime position to contribute back to the open source
communities on which they depend. For example, most of the Linux kernel is developed and
maintained by employees from companies like IBM and RedHat. However, corporate culture in
many companies (and particularly in small businesses) tends to lean strictly toward consumption
of open source and away from contribution. For example, in a recent survey of the Liferay
community we discovered almost 75% of companies that responded do not reward or encourage
open source contribution.
Anecdotal evidence from our community suggests there are two key reasons why companies do
not actively engage in the open source communities on which they depend. First, companies
believe time spent contributing to open source is time spent away from contributing to the
companys goals. Secondly, there is a fear that open source participation means giving away
competitive intellectual property.
These concerns quickly evaporate in the unique atmosphere created by collaborative open source
software, and heres why:
After responding to the boss' complaint about the website being slow and implementing the
grand supplier chain portal redesign, computing staff do not have a lot of spare time for what
employers may consider "pet projects" that must be done "off the clock."
What these companies fail to realize is that open source participation is a form of free on-the-job
training. Sure, it takes time to fix a bug or implement a new feature, test it, prepare a code
commit, and work with the core development team to incorporate the change into the codebase.
The good news is that in doing these tasks, employees are learning about sound software
engineering practices, quality assurance, leadership, communication skills, and teamwork (often
across geographic and cultural boundaries)a classic win-win for the company and its staff.
In addition, each successful contribution is one less thing that the company has to support during
future upgrades. Finally, as staff develop expertise through participation, they become much
more efficient in using the software, and reduce the cost of incremental feature development and
future maintenance. Companies can sell that expertise through value added services related to the
project.
News Flash: Your competitors would rather fail than acknowledge your expertise and follow in
your footsteps using your intellectual property. They dont want your IP because its worthless,
and they are sure of it. So, contributing that new feature that is so important to your business is
likely going to create zero buzz at your competitors HQ, yet still give you the benefits discussed
above.
In addition, the open source project as a whole will better understand your business, and is likely
to find innovative improvements youve never considered.
Of course, a clear company policy about participating in and contributing to open source is a
must, particularly if the company is related to the public sector. These should be developed with
the core goals of the company in mind, but a little common sense and consideration of the
indirect benefits of contribution can go a long way toward attracting and retaining passionate
staff that are willing to take the extra step to give a little back.
Companies decide whether to enter into business with other companies based on more than just
the bottom line. Forming a deep relationship with the open source communities on which you
depend sends a strong signal to your current and potential customers that you are willing to
invest in the mutual benefits of open source, and youd be just as likely to do so with them.
Highly skilled job seekers also participate heavily in open source projects and seek out
companies with an active presence as potential employers, so it pays to become ingrained in core
communities associated with your business.
So how can companies start making their marks in open source communities? Companies dont
have to go all-in on open source contributions (though theres certainly nothing wrong with
that!). Traditional means such as bug reports, forums, patches, features, documentation, and
translations are great. But there are other significant ways in which companies can contribute:
It doesnt take a lot of effort, but encouraging a culture of open source participation will have
lasting positive effects on the company, its staff, and the open source communities to which it
contributes.
Comments:
I totally agree on this article. More and more companies are beginning to look a open source to
know what to do with their products. Also companies look if employee participates in Open
Source projects to hire. In this way they will know if future employees are committed and self-
empowered.
Nobody wants your intellectual property: This is the most correct point of all this article. Myself
are the contributor of two projects GaiaEHR an (Electronic Medical Record) and Matcha an (bi-
directional microORM for Sencha) this two projects are Open Sourced, and we don't even look
at other proprietary products to tell us what to do with ours, most often we take ideas from the
community (potential users). This is the obvious path to choose, some times we analyze those
ideas and make a choice in what it will be implemented and what not. This path will always
surpass the proprietary ones, and the reason is logical they have a vision and, often that vision is
shortened by this centric vision by the one that give them.
Open source participation impresses potential customers and employees: And yes, it impress a lot
and it's also a great PR (Public Relations).
Even so, this is NATURE! Humans are meant to be creative and create new things. Don't you
think?
There are more ways to do one thing, right?
Great article, I share the same sentiments as Gino Rivera on this subject.
I am part the development team at Clinovo behind ClinCapture, the open-source Electronic
Data Capture system, the system is forked from OpenClinica. Our system is positioned as an
alternative to costly proprietary systems or inefficient error prone paper-based solutions for
conducting clinical trials. We see the open-sourcing of ClinCapture as one of the best ways to
improve the overall quality of the code base due to the vast number of people reviewing and
working on it.
Our community members all have the ability (and are encouraged) to submit code. The patches
are reviewed by fellow developers, who then commit the code to the repository if they feel that it
is of the required standard and does not break existing functionality. Anyone can request features,
log tickets for bugs or suggest enhancements and all committed code is organized into a
milestone, which is usually time-boxed to two week culminating into a bi-weekly release
schedule. Our community is also geared towards users, as Gino states, we take our ideas from
our community of users and potential users.
In the case of ClinCapture, I would hesitate to call it 'free on the job training'. Moreover, we
attract many developers who may wish to use the system for their current company and are
suggesting changes, improving features or fixing bugs. In this sense, they save their employer
money while still contributing to the greater good of the system.
On a side note, if anyone is interested in taking a look at our community, feel free to do so by
visiting: http://www.clinovo.com/clincapture/community
One choice is to buy from commercial vendors. To help select among commercial
vendors I have been offered documentation, peer referrals, advertising, and dog 'n
pony visits from those vendors. I've also been offered Starbucks gift cards, cameras,
luxury suite boxes for major league baseball games, and 3 day golf junkets to
Arizona. I've always declined this second set of offers. I love both baseball and
playing golf but I don't see what they have to do with my decision for how to spend
Netflix's money.
We do utilize some commercial software but there is often the alternative choice of
utilizing open source software, preferably open source software that implements an
open standard. Open source software projects often originate as a labor of love by
software developers who are tired of seeing a shared problem solved over and over
again in one off solutions, or perhaps they realize that they can offer a more simple
and elegant alternative to a commercial product. The great thing about a good open
source project that solves a shared challenge is that it develops it's own momentum
and it is sustained for a long time by a virtuous cycle of continuous improvement. At
Netflix we jumped on for the ride a long time ago and we have benefited
enormously from the virtuous cycles of actively evolving open source projects. We
benefit from the continuous improvements provided by the community of
contributors outside of Netflix. We also benefit by contributing back the changes we
make to the projects. By sharing our bug fixes and new features back out into the
community, the community then in turn continues to improve upon bug fixes and
new features that originated at Netflix and then we complete the cycle by bring
those improvements back into Netflix.
Netflix is hiring. If this blog resonates for you then I hope you'll visit
http://jobs.netflix.com and apply for one of our open positions.
Thanks,
Kevin
Nothing good is free: How Linux and open source
companies make money
We all know how popular and helpful Linux and open source products are, but since most of
them are available for free, how do the companies that produce them make any money to pay
their bills? As it turns out, lots of ways.
By David Gewirtz for DIY-IT | November 18, 2016 -- 13:57 GMT (19:27 IST) | Topic: SMBs
Last week's article on Linux for older hardware set a new record for Linux-related articles, in
that I did not get even one threatening letter. I did, however, get a bunch of email asking
business-related questions about Linux and open source.
A reader named Hermann sent me a comment that serves as a good starting point: "I don't see
how these Linux and open source companies can afford to make good software and still make it
free."
Hermann has a point, but it turns out there are a lot of ways to make money from free software.
Let's start with the issue of demand. The more specialized a type of software is, the fewer users
there will be. Generally, the fewer users, the smaller the market opportunity. The smaller the
market opportunity, the fewer the number of companies that will invest in developing
applications of that type.
David Gewirtz was a director of the Software Entrepreneurs' Forum, an executive at Symantec,
had the title of "Godfather" at Apple, and ran an award-winning tech company for 20 years. Now,
he's here to help you start your own software business. Buckle up!
Read More
There's actually a bit of a bell curve in the demand/developer ratio for commercial products.
Usage areas with very few users have very few developers who are willing to invest. But
applications like office suites, which are dominated by a few incredibly powerful players, also
have very few companies creating them (although they tend to spawn substantial aftermarket
ecosystems of add-on products).
In the middle, there are a great many commercial software companies developing software for a
wide variety of needs. In some cases, smaller total available market means lower revenue. But
many software companies compete quite well in small markets. Small markets with specialized
needs also foster very profitable vertical development opportunities, with companies that sell
expensive solutions complete with deep customization and support services.
In a sense, then, Hermann is right. A well-run, profitable software company must develop for a
market that can support its investment. After all, the company has rent, salaries, benefits, and
cost of doing business to recoup, as well as investors who demand a reasonable return on their
investment (ROI).
But open-source software is fueled by a variety of economic models. Many open-source projects
were originally created because the developers themselves needed a solution for their own use,
and then chose to release the code as open source.
This can happen when an individual or a company doesn't really want to be in the software
business but wants to provide value to others for free. Sometimes, they hope that by releasing
their project to open source, they can multiply their development resources for free, without
having to hire more coders.
There are other economic models that drive open software, including community editions and
freemium. In general (these terms can often blur together), a community model is one where a
company builds a product that it then releases as open source. That becomes the community
edition, a version of the software where the company doesn't invest any support resources, and
it's up to the users to provide ongoing support.
Where a community edition model company makes its money is on installation and support, and
sometimes on additional premium features. Customers who don't want to do all the hard work
themselves pay the company for support, training, maintenance, and any other service-related
opportunity the company can come up with.
One example of a company with a community edition is SugarCRM. It was founded in 2004.
Back then, it strongly promoted its community edition. The marketing benefit to it was that a
community edition is, in a lot of ways, a free trial. A lot of users will test a free download. A
subset of those users will convert to paying customers.
Today, while the community edition of SugarCRM still exists, you'd be hard pressed to find a
download link by simply navigating to its main site. I had to Google "SugarCRM community
download" before I found the right page.
A variation on this model is the freemium model. This can take a number of forms, ranging from
a downloadable free product with a so-called "pro" upgrade, to a core free product with
commercial add-ons, to a vendor-hosted, maintained SaaS version.
One company that mixes these approaches is Odoo, which used to be known as OpenERP. Odoo
has a free downloadable community edition, (which you can install on your own server), a per-
user enterprise edition (which can also be installed on a server), and a hosted version (which
customers also purchase by the seat and the month). [Corrected spelling of company Odoo, not
Oodo. Oopso.]
This approach allows Odoo to reach users who might also contribute back into the product but
don't want to pay for it, users who want to manage their own servers but have additional support
services, and users who simply want a SaaS solution.
Linux companies like RedHat and Canonical, the company behind the incredibly popular Ubuntu
Linux distro, also make much of their money from professional support services as well. If you
think about it, software used to be a one-time sale (with some upgrades), but professional
services are an ongoing annuity. That's a big part of why companies like Microsoft and Adobe
moved to subscription programs. Those annuities are quite profitable.
Another way open source companies make their money is with add-ons. One of the most
successful of these is WooCommerce. At the time it was acquired by Automattic in 2015, it
powered over 25 percent of all ecommerce stores. The base product was a free download that
was a plugin for WordPress (itself an open-source project with some commercial components).
But the way WooThemes (the parent company of WooCommerce) made its money was in selling
many different add-ons to the base online store product.
If you think about it, this is how almost all money-making apps in mobile app stores make
money. Most of the apps aren't open source, but they offer a starter app that's a free download,
and then offer in-app sales, which fuel their revenue stream. Given how much money some in-
app purchases make for their vendors, especially on popular smartphone games, you know it's a
viable revenue model.
WooCommerce also shows another way open source projects can make money. By building up a
huge user base, it created value. Automattic saw that value and paid a reported $30 million to
acquire the company and its 55 employees.
While we're on the subject of Automattic, let's discuss its business model. Automattic makes
much of its money through hosting services, where they offer a customized SaaS version of
WordPress for free, but sell add-on services. These range from allowing the use of a custom
domain name all the way up through managed hosting for very high-profile (and high revenue)
websites.
In this article, we've barely scratched the surface of how Linux and open source companies make
their money. Wikipedia lists sixteen different business models in its article on the topic.
There are many models for open source profitability. There are also, undeniably, many
challenges. Mostly, I'd like you to take away one encouraging and important point: there actually
are lots of ways to make money in software. Even though the old, 1990s model of selling
software on disks and in packages through retail stores is thankfully long dead, the modern
software industry is vibrant, creative, and exploding with possibility.
You can follow my day-to-day project updates on social media. Be sure to follow me on Twitter
at @DavidGewirtz, on Facebook at Facebook.com/DavidGewirtz, on Instagram at
Instagram.com/DavidGewirtz, and on YouTube at YouTube.com/DavidGewirtzTV.
One company that mixes these approaches is Odoo, which used to be known as OpenERP. Odoo
has a free downloadable community edition, (which you can install on your own server), a per-
user enterprise edition (which can also be installed on a server), and a hosted version (which
customers also purchase by the seat and the month). [Corrected spelling of company Odoo, not
Oodo. Oopso.]
This approach allows Odoo to reach users who might also contribute back into the product but
don't want to pay for it, users who want to manage their own servers but have additional support
services, and users who simply want a SaaS solution.
Linux companies like RedHat and Canonical, the company behind the incredibly popular Ubuntu
Linux distro, also make much of their money from professional support services as well. If you
think about it, software used to be a one-time sale (with some upgrades), but professional
services are an ongoing annuity. That's a big part of why companies like Microsoft and Adobe
moved to subscription programs. Those annuities are quite profitable.
Another way open source companies make their money is with add-ons. One of the most
successful of these is WooCommerce. At the time it was acquired by Automattic in 2015, it
powered over 25 percent of all ecommerce stores. The base product was a free download that
was a plugin for WordPress (itself an open-source project with some commercial components).
But the way WooThemes (the parent company of WooCommerce) made its money was in selling
many different add-ons to the base online store product.
If you think about it, this is how almost all money-making apps in mobile app stores make
money. Most of the apps aren't open source, but they offer a starter app that's a free download,
and then offer in-app sales, which fuel their revenue stream. Given how much money some in-
app purchases make for their vendors, especially on popular smartphone games, you know it's a
viable revenue model.
WooCommerce also shows another way open source projects can make money. By building up a
huge user base, it created value. Automattic saw that value and paid a reported $30 million to
acquire the company and its 55 employees.
While we're on the subject of Automattic, let's discuss its business model. Automattic makes
much of its money through hosting services, where they offer a customized SaaS version of
WordPress for free, but sell add-on services. These range from allowing the use of a custom
domain name all the way up through managed hosting for very high-profile (and high revenue)
websites.
In this article, we've barely scratched the surface of how Linux and open source companies make
their money. Wikipedia lists sixteen different business models in its article on the topic.
There are many models for open source profitability. There are also, undeniably, many
challenges. Mostly, I'd like you to take away one encouraging and important point: there actually
are lots of ways to make money in software. Even though the old, 1990s model of selling
software on disks and in packages through retail stores is thankfully long dead, the modern
software industry is vibrant, creative, and exploding with possibility.
You can follow my day-to-day project updates on social media. Be sure to follow me on Twitter
at @DavidGewirtz, on Facebook at Facebook.com/DavidGewirtz, on Instagram at
Instagram.com/DavidGewirtz, and on YouTube at YouTube.com/DavidGewirtzTV.
Business models for open-source software
From Wikipedia, the free encyclopedia
Contents
1 Funding
2 Challenges
3 Approaches
o 3.1 Dual-licensing
5 Examples
6 See also
7 References
8 Further reading
Funding
Much unlike proprietary off-the-shelf software that come with restrictive licenses, open-source
software is distributed freely, through the web and in physical media. Because creators cannot
require each user to pay a license fee to fund development this way, a number of alternative
development funding models have emerged.
There also exist stipends to support the development of open source software, such as Google's
Summer of Code.[2]
Another approach to funding is to provide the software freely, but sell licenses to proprietary
add-ons such as data libraries. For instance, an open-source CAD program may require parts
libraries which are sold on a subscription or flat-fee basis. Open-source software can also
promote the sale of specialized hardware that it interoperates with, some example cases being the
Asterisk telephony software developed by PC-telephony hardware manufacturer Digium and the
Robot Operating System (ROS) robotics platform by Willow Garage and Stanford AI Labs.
Many open source software projects have begun as research projects within universities, as
personal projects of students or professors, or as tools to aid scientific research. The influence of
universities and research institutions on open-source shows in the number of projects named
after their host institutions, such as BSD Unix, CMU Common Lisp, or the NCSA HTTPd which
evolved into Apache.
Companies may employ developers to work on open-source projects that are useful to the
company's infrastructure: in this case, it is developed not as a product to be sold but as a sort of
shared public utility. A local bug-fix or solution to a software problem, written by a developer
either at a company's request or to make his/her own job easier, can be released as an open-
source contribution without costing the company anything.[3] A larger project such as the Linux
kernel may have contributors from dozens of companies which use and depend upon it, as well
as hobbyist and research developers.
A new funding approach for open-source projects is crowdfunding, organized over web
platforms like Kickstarter, Indiegogo, or Bountysource.[4]
Challenges
This article or section may need to be cleaned up. It has been
merged from Commercial open-source applications.
Open-source software can be sold and used in general commercially. Also, commercial open-
source applications are a part of the software industry for some time.[1][5] While
commercialization or funding of open-source software projects is possible, it is considered
challenging.[6]
Since several open-source licenses stipulate that authors of derivative works must distribute them
under an open-source (copyleft) license, ISVs and VARs have to develop new legal and technical
mechanisms to foster their commercial goals,[7] as many traditional mechanisms are not directly
applicable anymore.
Traditional business wisdom suggests that a company's methods, assets, and intellectual
properties should remain concealed from market competitors (trade secret) as long as possible to
maximize the profitable commercialization time of a new product.[8] Open-source software
development minimizes the effectiveness of this tactic; development of the product is usually
performed in view of the public, allowing competing projects or clones to incorporate new
features or improvements as soon as the public code repository is updated, as permitted by most
open-source licenses. Also in the computer hardware domain, a hardware producer who provides
free and open software drivers reveals the knowledge about hardware implementation details to
competitors, who might use this knowledge to catch up.
Therefore, there is considerable debate about whether vendors can make a sustainable business
from an open-source strategy. In terms of a traditional software company, this is probably the
wrong question to ask. Looking at the landscape of open source applications, many of the larger
ones are sponsored (and largely written) by system companies such as IBM who may not have an
objective of software license revenues. Other software companies, such as Oracle and Google,
have sponsored or delivered significant open-source code bases. These firms' motivation tends to
be more strategic, in the sense that they are trying to change the rules of a marketplace and
reduce the influence of vendors such as Microsoft. Smaller vendors doing open-source work may
be less concerned with immediate revenue growth than developing a large and loyal community,
which may be the basis of a corporate valuation at merger time.
Approaches
A variety of open-source compatible business approaches have gained prominence in recent
years[according to whom?]; notable examples include dual licensing, software as a service, not charging
for the software but for services, freemium, donation-based funding, and crowdfunding.
There are several different types of business models for making profit using open-source
software (OSS) or funding the creation. Below are existing and legal commercial business
approaches in context of open-source software and open-source licenses.[7] The acceptance of
these approaches varies; some of these approaches are recommended (like selling services),
others are accepted, while still others are considered controversial or even unethical by the open-
source community. The underlying objective of these business models is to harness the size and
international scope of the open-source community (typically more than an order of magnitude
larger than what would be achieved with closed-source models) for a sustainable commercial
venture.[citation needed] The vast majority of commercial open-source companies experience a
conversion ratio (as measured by the percentage of downloaders who buy something) well below
1%, so low-cost and highly-scalable marketing and sales functions are key to these firms'
profitability.[citation needed]
Dual-licensing
Dual licensing offers the software under an open-source license but also under separate
proprietary license terms. The proprietary version can be sold to finance the continued
development of the free open-source version.[9] Customers can be attracted to a no-cost and open-
source edition, then be part of an up-sell to a commercial enterprise edition. Further, customers
will learn of open-source software in a company's portfolio and offerings but generate business
in other proprietary products and solutions, including commercial technical support contracts and
services. A popular example is Oracle's MySQL database which is dual-licensed under a
commercial proprietary license as also under the GPLv2.[10] Another example is the Sleepycat
License. Flask developer Armin Ronacher stated that the AGPLv3 was a "terrible success" as
"vehicle for dual commercial licensing" and noted that MongoDB, RethinkDB, OpenERP,
SugarCRM as well as WURFL utilizing the license for this purpose.[11]
The financial return of costs on open-source software can also come from selling services, such
as training, technical support, or consulting, rather than the software itself.[12][13]
Another possibility is offering open-source software in source code form only, while providing
executable binaries to paying customers only, offering the commercial service of compiling and
packaging of the software. Also, providing goods like physical installation media (e.g., DVDs)
can be a commercial service.
Open-source companies using this business model successfully are for instance RedHat and
IBM;[14] a more specialized example is that of Revolution Analytics.
Some open-source organizations such as the Mozilla Foundation[15] and the Wikimedia
Foundation[16] sell branded merchandise articles like t-shirts and coffee mugs. This can be also
seen as an additional service provided to the user community.
Selling subscriptions for online accounts and server access to customers is a way of making
profit based on open-source software. Also, combining desktop software with a service, called
software plus services. Providing cloud computing services or software as a service (SaaS)
without the release of the open-source software itself, neither in binary nor in source form,
conforms with most open-source licenses (with exception of the AGPL).
Because of its lack of software freedoms, Richard Stallman calls SaaS "inherently bad" while
acknowledging its legality.[22][23] The FSF called the server-side use-case without release of the
source-code the "ASP loophole in the GPLv2" and encourage therefore the use of the Affero
General Public License which plugged this hole in 2002.[24][25] In 2007 the FSF contemplated
including the special provision of AGPLv1 into GPLv3 but ultimately decided to keep the
licenses separate.[26]
Other financial situations include partnerships with other companies. Governments, universities,
companies, and non-governmental organizations may develop internally or hire a contractor for
custom in-house modifications, then release that code under an open-source license. Some
organizations support the development of open-source software by grants or stipends, like
Google's Summer of Code initiative founded in 2005.[2]
Voluntary donations
Larger donation campaigns also exist. In 2004 the Mozilla Foundation carried out a fundraising
campaign to support the launch of the Firefox 1.0 web browser. It placed a two-page ad in the
December 16 edition of the New York Times listing the names of the thousands who had donated.
[29][30]
The users of a particular software artifact may come together and pool money into an open-
source bounty for the implementation of a desired feature or functionality. Offering bounties as
funding has existed for some time. For instance, Bountysource is a web platform which has
offered this funding model for open source software since 2003.
Another bounty source is companies or foundations that set up bounty programs for implemented
features or bugfixes in open-source software relevant to them. For instance, Mozilla has been
paying and funding freelance open-source programmers for security bug hunting and fixing since
2004.[31][32][33]
Pre-order/crowdfunding/reverse-bounty model
A newer funding opportunity for open-source software projects is crowdfunding, which shares
similarities with the pre-order or Praenumeration business model, as well as the reverse bounty
model. It is typically organized over web platforms like Kickstarter,[34] Indiegogo,[35] or
Bountysource[4] (see also comparison of crowd funding services). An example is a successfully
funded Indiegogo campaign of Australian programmer Timothy Arceri, who offered for $2,500
to implement in two weeks an OpenGL 4.3 extension for the Mesa library.[35] Arceri delivered the
OpenGL extension code, which got merged into Mesa, and continued later his Mesa work with
successive crowdfunding campaigns.[36]
Advertising-supported software
In order to commercialize FOSS, many companies (including Google, Mozilla, and Canonical)
have moved towards an economic model of advertising-supported software. For instance, the
open-source application AdBlock Plus gets paid by Google for letting whitelisted Acceptable
Ads bypassing the browser ad remover.[37] As another example is SourceForge, an open-source
project service provider, has the revenue model of advertising banner sales on their website. In
2006, SourceForge reported quarterly takings of $6.5 million[38] and $23 million in 2009.[39]
Some companies sell proprietary but optional extensions, modules, plugins or add-ons to an
open-source software product. This can be a "license conform" approach with many open-source
licenses if done technically sufficiently carefully. For instance, mixing proprietary code and
open-source licensed code in statically linked libraries[40] or compiling all source code together in
a software product might violate open-source licenses, while keeping them separated by
interfaces and dynamic-link libraries might often adhere to license conform.
This approach is a variant of the freemium business model. The proprietary software may be
intended to let customers get more value out of their data, infrastructure, or platform, e.g.,
operate their infrastructure/platform more effectively and efficiently, manage it better, or secure
it better. Examples include the IBM proprietary Linux software, where IBM contributes to the
Linux open-source ecosystem, but it builds and delivers (to IBMs paying customers) database
software, middleware, and other software that runs on top of the open-source core. Other
examples of proprietary products built on open-source software include Red Hat Enterprise
Linux and Cloudera's Apache Hadoop-based software. Some companies appear to re-invest a
portion of their financial profits from the sale of proprietary software back into the open source
infrastructure.[41]
Some companies, such as Digium, sell proprietary but optional digital electronics hardware
controlled by an open-source software product.[42]
A variant of the approach above is the keeping of required data content (for instance a video
game's audio, graphic, and other art assets) of a software product proprietary while making the
software's source code open-source. While this approach is completely legitimate and compatible
with most open-source licenses, customers have to buy the content to have a complete and
working software product.[43] Restrictive licenses can then be applied on the content, which
prevents the redistribution or re-selling of the complete software product. Examples for open-
source developed software are Kot-in-Action Creative Artel video game Steel Storm, engine
GPLv2 licensed while the artwork is CC-BY-NC-SA 3.0 licensed,[44] and Frogatto & Friends
with an own developed open-source engine[45] and commercialization via the copyrighted game
assets[46] for iPhone, BlackBerry and MacOS.[47][48][49][50]
Other examples are Arx Fatalis (by Arkane Studios)[51] and Catacomb 3-D (by Flat Rock
Software)[52] with source code opened to the public delayed after release, while copyrighted
assets and binaries are still sold on gog.com as digital distribution.[53][54]
Doing so conforms with the FSF and Richard Stallman, who stated that for art or entertainment
the software freedoms are not required or important.[55]
The similar product bundling of an open-source software product with a proprietary hardware
part is called tivoization and legal with most open-source licenses except GPLv3, which
explicitly prohibits this use-case.[56]
If a software product uses only own software and open-source software under a permissive free
software licence, a company can re-license the resulting software product under a proprietary
license and sell the product without the source code or software freedoms.[57] For instance, Apple
Inc. is an avid user of this approach by using source code and software from open-source
projects. For example, the BSD Unix operating system kernel (under the BSD license) was used
in Apple's Mac PCs that were sold as proprietary products.[58]
The Free Software Foundation, on the other hand, is clearly against this practice.[60] The GNU
General Public License since version 2 has defined "source code" as "the preferred form of the
work for making modifications to it." This is intended to prevent the release of obfuscated source
code.[61]
Delayed open-sourcing
Some companies provide the latest version available only to paying customers. A vendor forks a
non-copyleft software project then adds closed-source additions to it and sells the resulting
software. After a fixed time period the patches are released back upstream under the same license
as the rest of the codebase. This business model is called version lagging or time delaying.[41][62]
For instance, the MariaDB Corporation created for business compatible "delayed open-sourcing"
the source-available Business source license (BSL)[63] which automatically relicenses after three
years to the FOSS GPL.[64][65] This approach garantuees for licensees that they have source code
access (e.g. for code audits), are not locked into a closed platform, or suffer from planned
obsolescence, while for the software developer a time-limited exclusive commercialization is
possible.[64]
However, this approach works only with own software or permissive licensed code parts, as
there is no copyleft FOSS license available which allows the time delayed opening of the source
code after distributing or selling of a software product.
Popular non-game software examples are the Netscape Communicator which was open-sourced
in 1998[72][73] and Sun Microsystems's office suite, StarOffice, which was released in October
2000 at its commercial end of life.[74] Both releases formed the basis of important open-source
projects, namely the Mozilla Firefox and OpenOffice.org/LibreOffice. However, Firefox
eventually gained a more-than-self-sustaining revenue model, so Firefox was not an example of a
commercial end-of-life release.
According to Yochai Benkler, the Berkman Professor for Entrepreneurial Legal Studies at
Harvard Law School, free software is the most visible part of a new economy of commons-based
peer production of information, knowledge, and culture. As examples, he cites a variety of FOSS
projects, including both free software and open source.[75]
This new economy is already under development. In order to commercialize FOSS, many
companies, Google being the most successful, are moving towards an economic model of
advertising-supported software. In such a model, the only way to increase revenue is to make the
advertising more valuable. Facebook has recently come under fire for using novel user tracking
methods to accomplish this.[76]
This new economy is not without alternatives. Apple's App Stores have proven very popular with
both users and developers. The Free Software Foundation considers Apple's App Stores to be
incompatible with its GPL and complained[77] that Apple was infringing on the GPL with its
iTunes terms of use. Rather than change those terms to comply with the GPL, Apple removed the
GPL-licensed products from its App Stores.[78] The authors of VLC, one of the GPL-licensed
programs at the center of those complaints, recently began the process to switch from the GPL to
the LGPL and MPL.[79][80]
Examples
Main article: List of commercial open-source applications and services
Much of the Internet runs on open-source software tools and utilities such as Linux, Apache,
MySQL, and PHP, known as the LAMP stack for web servers.[citation needed] Using open source
appeals to software developers for three main reasons: low or no cost, access to source code they
can tailor themselves, and a shared community that ensures a generally robust code base, with
quick fixes for new issues.
Despite doing much business in proprietary software, some companies like Oracle Corporation
and IBM participated in developing free and open-source software to deter from monopolies and
take a portion of market share for themselves. See Commercial open-source applications for the
list of current commercial open-source offerings. Netscape's actions were an example of this, and
thus Mozilla Firefox has become more popular, getting market share from Internet Explorer.[81]
Active Agenda is offered for free, but requires all extensions to be shared
back with the world community. The project sells a "Non-Reciprocal Private
License" to anyone interested in keeping module extensions private.
Adobe Systems offers Flex for free, while selling the Flash Builder IDE.
Canonical Ltd. offers Ubuntu for free, while they sell commercial technical
support contracts.
Francisco Burzi offers PHP-Nuke for free, but the latest version is offered
commercially.
Ingres is offered for free, but services and support are offered as a
subscription. The Ingres Icebreaker Appliance is also offered as a commercial
database appliance.
id Software releases their legacy game engines under the GPL, while
retaining proprietary ownership on their latest incarnation.
MySQL is offered for free, but with the enterprise version includes support
and additional features.
SUSE offers openSUSE for free through the openSUSE Project, while selling
SUSE Linux Enterprise (SLE).
Oracle - VirtualBox is free and open to anyone, but the VirtualBox extension
pack can only be used for free at home, thus requiring payment from
business users
Red Hat sells support subscriptions for Red Hat Enterprise Linux (RHEL) which
is an enterprise distribution periodically forked from the community-
developed Fedora.
Untangle provides its Lite Package for free, while selling its Standard and
Premium Packages by subscription
Zend Technologies offers Zend Server CE and Zend Framework for free, but
sells Zend Server with support and additional features.
Motivation, values, and work design as drivers of
participation in the R open source project for statistical
computing
1. Patrick Maira,1,
2. Eva Hofmannb,
3. Kathrin Gruberc,
4. Reinhold Hatzingerd,2,
6. Kurt Hornikf
Author Affiliations
1. Edited by Robert J. Tibshirani, Stanford University, Stanford, CA, and approved October
1, 2015 (received for review March 26, 2015)
1. Abstract
2. Full Text
4. Figures
5. SI
6. Metrics
7. Related Content
8. PDF
9. PDF + SI
Significance
Over the last years, the open-source environment R has become the most popular environment
for statistical computing and data analysis across many fields of research. The developer
community is highly active: Thousands of packages are available in the official Comprehensive
R Archive Network repository and more on developer platforms like GitHub or R-Forge. One
question that has not been studied yet is as follows: why do people contribute to the R
environment? What are the key motives that drive package authors? Do these developers have
specific personal value structures? Are some work environments more conducive to productivity
than others? This study is the first empirical study, to our knowledge, performed within the R
package author community that finds answers to these questions.
Next Section
Abstract
One of the cornerstones of the R system for statistical computing is the multitude of packages
contributed by numerous package authors. This amount of packages makes an extremely broad
range of statistical techniques and other quantitative methods freely available. Thus far, no
empirical study has investigated psychological factors that drive authors to participate in the R
project. This article presents a study of R package authors, collecting data on different types of
participation (number of packages, participation in mailing lists, participation in conferences),
three psychological scales (types of motivation, psychological values, and work design
characteristics), and various socio-demographic factors. The data are analyzed using item
response models and subsequent generalized linear models, showing that the most important
determinants for participation are a hybrid form of motivation and the social characteristics of
the work design. Other factors are found to have less impact or influence only specific aspects of
participation.
Schwartz values
motivation
work design
The story of the R environment for statistical computing (1) has been one of tremendous success.
Since it was first conceived (2), R has been attracting more and more users and contributors from
different fields where data analysis plays a major role. Fox (3) conducted a series of interviews
with members of the R Core Team to explore the social organization of R and to identify factors
crucial to its success.
The study presented here aims to examine why package authors participate in the R project. We
use scales on work design characteristics, personal values, and types of motivationbased on
theories from a general open-source software (OSS) perspectiveto learn about factors and
incentives that drive authors to develop R packages, as well as participate in R conferences and
mailing lists.
The overwhelming majority of R packages are released under open-source licenses, thereby
placing no restrictions on users and guaranteeing that these packages can become public goods
(4). Although from a traditional economic point of view, it appears to make no sense to give
away ones skills and efforts for free, thousands of highly skilled developers have organized into
communities like the Comprehensive R Archive Network (CRAN; CRAN.R-project.org/),
Bioconductor (5) (www.Bioconductor.org/), R-Forge (6) (R-Forge.R-project.org/), and GitHub
(https://github.com/) to contribute code and documentation to open-source R packages
distributed by these communities.
Studying software developers motivations and determinants for participating in OSS projects is
not a straightforward task. There are many internal and external factors that might potentially
play a role and, hence, have to be taken into account when one wishes to explain OSS
participation. Empirical findings in this research area are rather limited and partially ambiguous
(7). In this study, we apply models from item response theory (IRT) and generalized linear
models (GLMs) to data collected in a survey, conveyed on the popular platforms CRAN, R-
Forge, and Bioconductor.
Third, personal values can be important for understanding contributions to OSS projects. The
classic value theory in ref. 22 distinguishes 10 different values: benevolence, conformity,
tradition, security, power, achievement, hedonism, stimulation, self-direction, and universalism.
Oreg and Nov (23) determined the following three values to be relevant for OSS developments:
self-direction, power, and universalism (24). Self-direction type values (e.g., creativity, choosing
own goals, curiosity) are driven by independent thought and action. Thus, they are closely related
to forms of intrinsic motivation. Power type values (e.g., social power, social recognition,
authority) reflect abstract outcomes on an individuals achievements. These values do not refer to
the direct outcomes of any particular action, but to the status in social structure an individual is
able to derive from actions. Hence, they relate directly to forms of internalized extrinsic
motivation. Universalism type values (e.g., equality, wisdom, social justice) refer to action for
the welfare of all people and are derived from peoples awareness of the scarcity of resources.
They imply that individuals will consciously protect their own survival needs through the
acceptance and just treatment of anyone outside their group (22).
The online questionnaire for the package authors, provided as SI Results, included standard
socio-demographic variables, as well as more specific dichotomous work-related variables such
as whether respondents have a PhD degree, an education in statistics, are employed full time,
work in academia, and work as statisticians.
Based on the research results described above, three lines of possible psychometric incentives are
pursued: (i) hybrid forms of motivation, (ii) work design characteristics, and (iii) values. We
investigate to which extent these factors determine the degree of the authors participation in the
R project. The following subsections describe these variables and constructs included in our
study. Fig. 1 summarizes the latent structure of the psychometric scales we use and their relation
to the measures for participation.
In a new window
Download PPT
Fig. 1.
Psychometric constructs. Hybrid forms of motivation (25), work design characteristics (21), and
values (22) determining participation in the R project.
Degree of Participation.
Participation in OSS projects will primarily manifest itself in the form of code contributions. As
previous studies have shown, however, this is just one part of an underlying learning and
information process (17). A prominent example of other forms of contribution is the active
engagement in social media platforms such as mailing lists or blogs (9).
In the context of the R project, contributed code is typically conveniently organized in packages
and distributed via repositories such as CRAN or Bioconductor. This fact makes packages the
primary vehicle for communicating conceptual and computational tools related to R. Hence, the
number of R packages (co)developed by an individual author (cf. Fig. S1) can easily be
interpreted as the first, main variable of the extent of participation in the R project. As a second
indicator, we use active participation in R project mailing lists (R-help, R-devel, special interest
groups, ) as an indicator for engagement in social media. Finally, as third participation
indicator, we consider attending R conferences such as the annual useR! or the Directions in
Statistical Computing (DSC) meetings.
In a new window
Download PPT
Fig. S1.
Distribution of the number of packages the authors are involved in.
Psychometric Constructs.
As elaborated above, the classic distinction between intrinsic and extrinsic motivation is seen as
too rigid within our context. Reinholt (25) presents a concept that distinguishes between extreme
intrinsic motivation, well-internalized extrinsic motivation/moderated intrinsic motivation, and
extreme extrinsic motivation. Well-internalized extrinsic motivation and moderated intrinsic
motivation comprise hybrid types of intrinsic and extrinsic motivation. The corresponding scales
are based on this concept of motivation because it provides a nuanced and coherent
understanding of motivational types along a continuum of motivation. This framework also
accounts for potential interaction effects between intrinsic and extrinsic types of motivation. For
the intrinsic and extrinsic motivation subscales, 36 items are included in our questionnaire. Each
subscale (i.e., enjoyment based intrinsic motivation, self-reinforcement, obligation-based
motivation, integrated regulation, identification, introjection-based regulation, external
regulation) consists of four to eight items.
As suggested by previous studies (9, 10), the Work Design Questionnaire (WDQ) (21) is a
prominent tool to investigate work design characteristics. This work design model captures,
among others, the following three subscales: the effects of task characteristics (autonomy, task
variety, task significance, task identity, feedback from job), social characteristics (received and
initiated interdependence, feedback from others), and knowledge characteristics (job complexity,
information processing, problem solving, skill variety, specialization). In its original form, the
WDQ is composed of 77 items. Using the three subscales above reduces the questionnaire to 48
items. Note that WDQ items referring to work tasks in general were adapted to the work on R
packages.
Regarding personal values, we consider 3 of the 10 values of the Schwartz value scale (self-
direction, power, and universalism). All 19 items pertaining to these value subscales are included
in the questionnaire.
Research Questions.
Based on the theoretical extension of the concept of intrinsic and extrinsic motivation (25), we
hypothesize that extreme extrinsic motivation (comprising external regulation and introjection-
based regulation), extreme intrinsic motivation (stemming solely from enjoyment-based intrinsic
motivation), and well-internalized extrinsic motivation/moderated intrinsic motivation
(identification, obligation-based intrinsic motivation, self-reinforcement, and integrated
regulation) are positively related to the participation in the R project.
Regarding work design, it is expected that task characteristics (comprising autonomy, task
variety, task significance, task identity, and feedback from the job), knowledge characteristics
(including job complexity, information processing, problem solving, skill variety, and
specialization), and social characteristics (consisting of received and initiated interdependence
and feedback from others) are positively related to participation. The more positive these
characteristics are perceived, the more a package author should participate in R activities.
Finally, in line with earlier studies, it is hypothesized that the values self-direction and
universalism relate positively to participation, whereas power is expected to relate negatively.
Results
First, we look at the negative binomial regression with the number of packages an author has
(co)authored as the response variable (Table S1). The effect plots for the final model are given in
Fig. S2.
In this window
In a new window
Table S1.
Negative binomial GLM parameter estimates for number of packages
In a new window
Download PPT
Fig. S2.
Effect plots for negative-binomial regression on number of packages (for the variables selected
by stepwise regression).
The number of packages are positively influenced by hybrid and extrinsic motivation. Work
design is also an important determinant of the number of packages, with social characteristics
being positively associated and task characteristics being negatively associated. Thus, the higher
the intiated/received interdependence of an author and the more feeback he/she gets from the
community, the more packages he/she is involved in.
Conversely, the higher a package author scores on the task dimension, the lower the number of
packages (co)authored. In terms of the value scales, only power is found to be significantly
associated with the number of packages showing a negative effect. On the socio-demographic
side, the fact that a package author works full time and his/her field of work is statistics have a
significant effect.
The results for the logistic regression model of participation in mailing lists are given in Table S2
and the effect plots are shown in Fig. S3.
In this window
In a new window
Table S2.
Logistic GLM parameter estimates for participation in mailing lists
In a new window
Download PPT
Fig. S3.
Effect plots for logistic regression on lists participation (for the variables selected by stepwise
regression).
Finally, Table S3 presents the results of the logistic regression model for the binary response,
indicating participation of package authors in R conferences and workshops. The corresponding
effect plots are given in Fig. S4.
In this window
In a new window
Table S3.
Logistic GLM parameter estimates for participation in conferences
In a new window
Download PPT
Fig. S4.
Effect plots for logistic regression on conference participation (for the variables selected by
stepwise regression).
Regarding the motivational dimension, hybrid motivation is again found to be the most important
determinant. Its influence is again positive. In terms of work design, social characteristics are
significant with a positive impact on participation. Regarding values, universalism is significant
at 5% after stepwise selection. The only significant socio-demographic variable is the
occupational status: a full-time employment of a package author is a strong determinant to
participate in R conferences. None of other socio-demographic variables (except, to a certain
degree, statistics as the field of work that has a minor influence) has any impact on the model.
To summarize, the broad picture is very similar across all three participation responses (and
corresponding models), even if the details vary to a certain degree: hybrid motivation and social
characteristics are the most important determinants for higher levels of participation in the R
project. The picture for extrinsic and intrinsic motivation is less clear and varies over the
particular type of participation. Authors that score highly on the task characteristics scale
generally participate less, whereas knowledge characteristics do not play an important role.
Similarly, values are not found to be important drivers of participation as they rarely show up in
the selected models. The influence of the socio-demographic variables varies across the models:
full-time employment generally increases participation, whereas a job in academia somewhat
lowers it. Working in statistics has a positive effect on the number of packages and participation
in the conferences but a negative on participation in mailing lists. The remaining two variables
(having a PhD and an education in statistics, respectively) cannot be shown to have an impact on
participation in any of the models.
Discussion
This study has asked why R package authors participate in the R project for statistical
computing. A survey was conducted and the data were analyzed using IRT models and,
subsequently, GLMs (with SIMEX correction). In what follows, our findings are discussed in
more detail and related to the literature on participation in OSS projects.
In line with the literature (79), hybrid motivation a crucial determinant for participation,
whereas purely intrinsic and purely extrinsic forms of motivation are less important. These
findings are reflected by our regression results and conform well with the academic life cycle.
Various factors, including reputation, reciprocity, or social norms, can contribute to an
internalization of extrinsic motives. On the one hand, many academics do what they have to
do. On the other hand, they select tasks they enjoy doing that can also encompass activities such
as fun coding (8).
The influence of purely extrinsic motivation, which, in particular, includes monetary rewards (8),
varies across the participation variables. In part, this may be due to a strong rooting of the R
project in various academic communities. Although packages and conferences are by now
regarded as scientific contributions, mailing list contributions have no (direct) impact on
academic performance measures. This fact is somewhat substantiated by the positive (but not
significant) influence of intrinsic motivation on contribution to mailing lists. We note that
Bianchi et al. (16) found that contributions to electronic networks of practice are increased if
the contributors perceive that this enhances their reputation (i.e., a typical extrinsic motive).
Thus, participation in R mailing lists is apparently not perceived to do so. This situation might be
different in the more recently established question and answer websites such as Stack Exchange,
which work differently from classical mailing lists and explicitly try to capture the reputation of
its contributors.
Social work design characteristics reflect the fact that work is performed within a broader social
environment (21) where single individuals highly depend on each other. Our results show that
OSS projects provide high degrees of social dependency and feedback as theoretically
hypothesized in ref. 18. That social characteristics are such an important factor in our models is
not too surprising, given that we are interacting in a social media dominated environment and
social coding platforms are widely used (30). Psychological explanations for our results are the
following: first, interaction with persons perceived as important leads to reputation (self-esteem,
future job opportunities, etc.). Second, interaction with alike minded persons (i.e., interested in
solving statistical problems) might be a possibility to express oneself and enjoy social inclusion.
From a perspective that goes beyond work design characteristics, social aspects include social
recognition and identification. The R community seems to offer the opportunity for R developers
to identify with this highly valued group and feel a sense of belonging. It can be assumed that
they receive parts of their self-esteem by belonging to such a valued group (31) and are
especially motivated to contribute to this group. It would be interesting to study such general
social aspects of reputation gaining in a follow-up study.
Task characteristics are found to have a negative influence on participation which can be
explained as follows: if the work is organized around the development of an R package as the
central task (from development of code, via writing of manuals and vignettes to maintenance and
bug fixing), R authors appear to do that but are less involved in the development of further
packages or discussions on mailing lists. Or conversely, those authors who participate more and
develop several packages, do not appear to be driven be the task of R package development as
such but by the underlying knowledge characteristics involved.
Values.
Our results indicate that in the context of R packages there appears only little additional direct
effect of the valuesother than potential indirect effects through the types of motivation. There
are two notable exceptions: power is shown to have a clear negative effect on the number of
packages and universalism has a clear negative effect on conference participation.
The former reflects that package authors, for whom social power, wealth, social recognition, and
authority are important, produce fewer packages than their trait counterparts. The way the field
of applied and computational statistics has developed over the last years, R package
implementations have increased in scientific value. Thus, for a researcher, a corresponding
implementation has become an academic status symbol to the effect that they refer to themselves
as R package author even when involved in a single package only.
The latter shows that the higher a package author scores on the universalism dimension, the less
likely he or she is to attend meetings. A closer look at what is meant by universalism provides an
interesting interpretation of this result. According to Schwartz (22), attributes associated with
universalism include the following: a world of beauty, unity with nature, protecting the
environment, and inner harmony. These attributes are derived from an awareness of the scarcity
of resources. Thus, universalism implies a strong environmental attitude that may be
incompatible with carbon-intensive long distance travels to conferences.
Socio-Demographic Variables.
Full-time employment always has a positive impact on participation; significantly for the number
of packages and conference participation. This fact suggests that many contributions to the R
project are made as part of the job. For mailing lists the influence is weaker but, as already
argued above, such participation is typically not part of the job description. Additionally, there
may also be direct effects of full-time employment on conference participation (e.g., through
reimbursement of expenses).
Working in the field of statistics also has positive impact on the number of packages and
conference participations but clearly negative impact on mailing list participation. Although the
former is not surprising given that the R system is dedicated to statistics, the latter may not be
obvious. However, statisticians will typically have other ways of asking questions related to R
(e.g., colleagues within their department) and other ways of providing feedback about the
corresponding statistical methods (e.g., in forms of papers, books, or lectures). However, for R
authors and users coming from other domains (say, ecology, finance, or epidemiology), the R
mailing lists may be a more crucial means of obtaining information related to R. This supposition
overlaps with the findings in ref. 32, which showed that answers on the R mailing lists are
mainly given by a few central players feeling responsible for certain topics.
Interestingly, an academic background (i.e., having a PhD or a job in academia) does not lead to
more participation as hypothesized in ref. 14. In fact, it has almost no impact on any of the three
response variables.
Conclusions.
Our results show that growth of R-related projects is positively influenced by hybrid motivation,
whereas purely intrinsic or extrinsic motives are less important. Hence, this suggests that
extrinsic motives (such as monetary rewards or building reputation) can be important drivers but
need to be balanced by possibilities of internalizing them. However, given the ongoing
commercialization of the R ecosystem this aspect deserves reinvestigation in the future.
In conclusion, our results are important for institutions and individuals that want to stimulate
growth of R developments: they must provide a work environment and corresponding incentives
that foster a high amount of interdependence and feedback from others. Such collaborative
research strategies also include the encouragement to work on projects with researchers outside
the institution and the engagement in social coding platforms.
In total, we had 4,274 email addresses of R package authors. They were asked to fill out an
online questionnaire within the following 3 wk. The survey was conducted in May 2010 using
the online survey software Unipark. The platforms we used for the acquisition of the email
addresses were CRAN, R-Forge, and Bioconductor. In total we sent out 4,274 emails, of which
200 could not successfully be delivered (bounced). Note that if packages had multiple
authors, emails were sent out to those who provided an email address in the package description
file. In addition, in the email list we used, some package authors had multiple email addresses.
Therefore, the response rate below reflects a lower bound.
A total of 1,448 authors considered the questionnaire: 310 respondents quit immediately and 51
respondents scrolled through without answering. Altogether, a sample of 1,087 persons
remained, which leads to a response rate of at least 27%. This response rate is in line with related
OSS studies such as in refs. 10, 15, and 33. A total of 764 package authors completed the whole
questionnaire without skipping any of the items. From a statistical power point of view, this
sample size is sufficiently large to carry out all of our statistical analyses. The issue of possible
nonresponse bias is addressed and analyzed in detail in SI Text. Our results are representative for
R package authors who contributed to more than one package (see Fig. S5).
In a new window
Download PPT
Fig. S5.
Sample vs. population proportions for the number of packages. The percentages are based on the
conditional relative frequencies (i.e., conditional on authors with more than one packages).
Reproducibility Materials.
The following materials were submitted to fully reproduce the analysis in the article. The raw
data are stored in Dataset S1, along with the variable descriptions (Dataset S2). The code file in
Dataset S3 contains the R code for data preparation, IRT analysis, and all the GLM computations
(including regression tables and effect plots) presented in SI Text and SI Results. In addition, it
provides code to examine possible nonresponse bias.
SI Text
Nonresponse Bias.
Nonresponse bias is an issue that often occurs in email surveys, especially when the response
rate is not particularly large. The bias arises if the answers of survey respondents (the sample)
differ from the potential answers of those not in the sample. In this case, results cannot be
generalized to the whole population. There are several strategies that address nonresponse bias as
elaborated in classical texts such as refs. 34 and 35. The strategy we use to address potential
biases in our survey is that of comparing sample values of a variable with known values from the
population. The key variable within this context that we use is number of packages to which each
author contributed. Using this variable, we can determine the population values by extracting the
author names from the package description files and then computing the corresponding
frequencies.
In our sample, 31.15% of the authors contributed to one package only, whereas in the population,
we have 67.86%. These numbers indicate that one-package authors are underrepresented in our
sample. This fact is not surprising, because people who contributed only one package are likely
to have a lower committment to the R project and, therefore, are less likely to fill out such a
questionnaire. Let us examine the (conditional) relative frequencies of authors of two or more
packages. Note that we merge authors with 10 or more packages into a single category. Fig. S5
shows the percentages across the number of packages for the sample and the population. We see
that two-package authors are only slightly underrepresented in the sample, and for the remaining
ones, the sample and population proportions match closely. Therefore, our results are
representative for the subpopulation package authors who contributed to at least two packages.
Statistical Methodology.
The goal of our analysis is to determine the effect of hybrid forms of motivation, work design,
and values on participation in the R project, controlling for socio-demographic/work-related
factors. We thus start by computing the psychometric scores as depicted in Fig. 1 from the
corresponding questionnaire items using the two-parameter logistic model. Subsequently, these
scores are used as explanatory variables in GLM analysis for each of the three variables
measuring participation. To account for potential measurement errors when psychometric scores
enter the regression models, the estimators are corrected by the simulation-extrapolation
(SIMEX). The following subsections give a brief methodological background.
IRT analysis.
The latent constructs occurring in the scales for hybrid forms of motivation, work design, and
values are scored using IRT models. IRT is a family of latent variable models to score items and
persons on a single latent trait. Our IRT model of choice is the two-parameter logistic model (26)
defined as
P(Xvi)=exp[i(vi)]1+exp[i(vi)].
We perform our IRT analysis separately for each scale dimension (with the items assigned to the
dimensions correspondingly) using the R package ltm (36). Before fitting an IRT model, we
examined unidimensionality of each subscale using categorical principal component analysis,
implemented in the homals package (27). The following items showed a strong deviation from
unidimensionality: Packages are a byproduct of my empirical research. If I cannot find suitable
existing software to analyze my data, I develop software components myself and Packages are
a byproduct of my methodological research. If I develop/extend methods, I develop
accompanying software, e.g., for illustrations and simulations from the motivation scale
(extreme extrinsic motivation construct). In addition, The work on R packages requires that I
only do one task or activity at a time had to be removed from knowledge characteristics.
All subsequent Q1
fit statistics were not significant and, therefore, no additional items were eliminated. Note that
because we have a multiple testing problem, the level was corrected by dividing 0.05 by the
number of items per subscale. For our final item subsets, we compute the person parameters for
each of the nine traits. For subsequent analyses and tables, the resulting new variables are labeled
mextrinsic, mintrinsic, and mhybrid for the motivation scales; wtask, wsocial, and wknowledge
for the scales obtained from the WDQ; and vpower, vselfdirection, and vuniversalism for the
value scales.
g()=+X,
[S2]where is the mean of the participation response variable, g() represents the
corresponding link function, and is the matrix containing of person parameters
with corresponding regression coefficients . X is the matrix of socio-demographic
variables with corresponding regression coefficients
. For the count response number of packages, we fit a negative-binomial (NB) model, which, as
opposed to a regular Poisson regression, accounts for overdispersion. Binomial GLMs with
logistic link function are used for the binary responses capturing participation in mailing lists and
conferences, respectively.
SIMEX correction.
are subject to measurement error (ME) as they are obtained from IRT analysis. Due to this ME,
the ordinary GLM estimates are, in general, biased. To mitigate this problem, the
heteroskedasticity of the MEs needs to be taken into account.
Let
be the estimated regression coefficients. To get unbiased estimates in the presence of additive
MEs, we apply the SIMEX method proposed in ref. 29 after fitting the basic (naive) GLMs.
For our specific problem, we apply the jackknife variant of SIMEX (37), which is based on the
following idea: the starting point is the SE of the person parameters in construct c (c=1,,C
), which reflects the ME. The ME can be a single value for each construct c or a
vector of length n allowing for varying MEs across persons. In our analysis, we allow
for full ME heteroskedasticity (across constructs, across persons), which leads to the
.
Through ME-based jackknife resampling, the SIMEX approach simulates repeated
measurements. By refitting the model in each step, we get a new parameter vector
. SIMEX theory states that the mean of the parameter distribution resulting from
SI Results
Descriptive Data Analysis.
The first dependent variable measuring participation is the number of packages (co)developed by
an individual author. Its distribution is right skewed, has a mean of 2.9, a median of 2, maximum
of 33, and an SD of 3.45.
Fig. S1 shows the distribution of the number of packages. A few package authors stated that they
have been involved in zero packages. The reason for this could be that they contributed code to a
particular package, appear in the author list, but do not consider themselves being involved in the
development of this particular R package (e.g., authors that are active on R-forge only). The
other two dependent participation variables are binary, with 57.07% contributing to the R
mailing lists and 31.02% attending R conferences.
The items pertaining to the motivation, work design, and value scales are transformed to
psychometric scores using IRT analysis as described above. Our dichotomous work-related
variables give the following descriptive results: PhD degree (phd, yes: 71.47%), education in
statistics (statseduc, yes: 63.09%), employed full time (fulltime, yes: 85.21%), work in academia
(academia, yes: 60.47%), and work as statisticians (statswork, yes: 63.22%).
The following tables and plots show the results of the three GLMs. The first table refers to the
negative binomial regression with number of packages as response, the second table to the
logistic regression with participation in mailing lists as response, and the third table to the
logistic regression with participation in conferences as response. The effects plots depict the
effect structure based on the regression parameters for the predictors selected by stepwise
regression.
Questionnaire.
Dear R package author,
You have been selected as a potential participant in a survey about motivation for developing R
packages and participating in the R community more generally.
Filling in this questionnaire is voluntary and will take approximately 15 min to complete.
Your answers are anonymous and confidential and it will not be possible to identify your
individual responses when the data are analyzed and reported. The answers you provide serve the
improvement of the Comprehensive R Archive Network (CRAN) to offer developers and
maintainers of R an even more effective platform. They also are for research purposes and aim to
examine what motivates persons to participate actively in the development and maintenance of R
packages. You can withdraw your participation until you have completed the online
questionnaire and pressed the send-button at the end of the questionnaire. After this point, it is
not possible to withdraw your data as all responses are anonymous and individual responses
cannot be identified. The study data will be stored securely and only the project researchers will
have access to it. The overall results from the questionnaire will be used to undertake adaptations
in CRAN and will be included in academic publications, conference presentations and for
teaching purposes.
By filling in the questionnaire, you are providing your consent for your responses to be used in
the ways previously described.
If you have any queries regarding the study or its results, please contact us!
Below find a list of statements on your development of R packages. Please indicate whether you
agree or disagree with the following statements! Choose the option that slightly better represents
your position!
The development of R packages is arranged so that I can work on an entire package from
beginning to end.
The work on R packages itself is very significant and important in the broader scheme of things.
The work on R packages often involves dealing with problems that I have not encountered
before.
The results of my work on R packages are likely to significantly affect the lives of other people.
The work on R packages requires that I only do one task or activity at a time.
Other people in the R community provide information about the effectiveness (e.g., quality and
quantity) of my R package performance.
The work on R packages requires technical skills regarding package building and documentation.
The development of R packages provides me the chance to completely finish the pieces of work I
begin.
Unless my work on the R package gets done, other tasks cannot be completed.
The work on R packages requires me to keep track of more than one thing at a time.
Section 1.2.
Below find again a list of statements on your development of R packages. Please indicate
whether you agree or disagree with the following statements! Choose the option that slightly
better represents your position!
The work on R packages involves solving problems that have no obvious correct answer.
The activities while working on R packages are greatly affected by the work of other people. The
development of R packages involves completing a piece of work that has an obvious beginning
and end.
The tools, procedures, materials, and so forth used to develop R packages are highly specialized
in terms of purpose.
The development of R packages allows me to make decisions about what methods I use to
complete my work.
I receive a great deal of information from the R community about my R package performance.
The work on R packages allows me to make my own decisions about how to schedule my work.
The work on R packages depends on the work of many different people for its completion.
The work activities themselves provide direct and clear information about the effectiveness (e.g.,
quality and quantity) of my performance.
Section 2.
Find a list of values below. Please evaluate the importance (unimportant vs. important) of each
value as a guiding principle in your life! Choose the option that slightly better represents your
beliefs!
Section 3.1.
Find a list of statements on your development of R packages below. Please indicate whether you
agree or disagree with the following statements! Choose the option that slightly better represents
your position!
they are a byproduct of my empirical research. If I cannot find suitable existing software to
analyze my data, I develop software components myself.
it is important for my personal goals but for no apparent rewards, such as money, career
opportunities, etc.
it is part of my identity.
Section 3.2.
Again, find a list of statements on your development of R packages below. Please indicate
whether you agree or disagree with the following statements! Choose the option that slightly
better represents your position!
it is a joyful activity.
Section 4.
As student at a university
As academic at a university
Other
For how long have you been participating in the R community (in years)? ____________
Do you plan to continue to participate in the R community? Please indicate the extent to which
you think further participation probable on a scale from 1 to 5 (1 = Very unlikely, 5 = Very
likely)!
Do you use other statistical software packages than R? Multiple answers are acceptable.
Stata
SAS
S-PLUS
Minitab
Systat
EViews
MATLAB
Other
If you are working in a team coding R packages, how many people other than you work
approximately in this team? In case you are working alone, please fill in 0! ____________
Until now, in the development of how many R packages have you been involved? ____________
CRAN
Bioconductor
R-Forge ()
RForge ()
Other
If other, please specify! ____________
In case you have published manuscripts on your R packages, in which media have you published
them? Multiple answers are acceptable.
Computational Statistics
Other
Section 5.
Are you ?
Male
Female
High school
In which fields have you been educated? Multiple answers are acceptable.
Statistics
Social sciences
Life sciences
Other
Training/student
Retired
Not working
Student
Other
Statistics
Social sciences
Life sciences
Other
Abkhazia
Afghanistan
Albania
Algeria
Andorra
Previous SectionNext Section
Acknowledgments
The authors would like to commemorate with sorrow the death of Reinhold Hatzinger. His
contributions to this paper were invaluable and he will be fondly remembered by his friends and
colleagues. The authors would like to thank the editor Robert Tibshirani and two anonymous
referees for their careful and detailed comments.
Footnotes
1
To whom correspondence should be addressed. Email:
mair@fas.harvard.edu.
2
Deceased July 17, 2012.
Author contributions: E.H. and R.H. designed research; P.M., E.H., and K.H. performed
research; P.M., K.G., R.H., A.Z., and K.H. analyzed data; and P.M. and E.H. wrote the
paper.
Previous Section
References
1.
1. R Core Team
Google Scholar
2.
1. Ihaka R,
2. Gentleman RC
(1996) R: A language for data analysis and graphics. J Comput Graph Stat
5(3):299314
CrossRefGoogle Scholar
3.
1. Fox J
Google Scholar
4.
1. von Hippel E,
2. von Krogh G
5.
.
CrossRefMedlineGoogle Scholar
6.
1. Theussl S,
2. Zeileis A
Google Scholar
7.
1. Roberts JA,
2. Il-Horn H,
3. Sandra AS
CrossRefGoogle Scholar
8.
1. Lakhani KR,
2. Wolf RG
(2005) Why hackers do what they do: Understanding motivation and effort in
free/open source software projects. Perspectives on Free and Open Source
Software, eds Feller J, Fitzgerald B, Hissam S, Lakhani KR (MIT Press,
Cambridge, MA), pp 322
Google Scholar
9.
1. Shah SK
CrossRefGoogle Scholar
10.
1. Hertel G,
2. Niedner S,
3. Hermann S
CrossRefGoogle Scholar
11.
1. Li Y,
2. Tan CH,
3. Teo HH
CrossRefGoogle Scholar
12.
1. Deci EL,
2. Koestner R,
3. Ryan RM
13.
1. Raymond E
(1999) The cathedral and the bazaar. Knowl Tech Policy 12(3):2349
CrossRefGoogle Scholar
14.
1. Henkel J
CrossRefGoogle Scholar
15.
1. Wasko M,
2. Faraj S
Google Scholar
16.
1. Bianchi AJ,
2. Kang SM,
3. Stewart D
Google Scholar
17.
1. Lakhani KR,
2. von Hippel E
(2003) How open source software works: Free user-to-user assistance. Res
Policy 32(6):923943
CrossRefGoogle Scholar
18.
1. Hertel G
CrossRefGoogle Scholar
19.
1. Hemetsberger A,
2. Reinhardt C
20.
1. Hackman JR,
2. Oldham GR
(1976) Motivation through the design of work: Test of a theory. Organ Behav
Hum Perform 16(2):250279
21.
1. Morgeson FP,
2. Humphrey SE
22.
1. Schwartz SH
CrossRefGoogle Scholar
23.
1. Oreg S,
2. Nov O
CrossRefGoogle Scholar
24.
1. Engelhardt S,
2. Freytag A
(2013) Institutions, culture, and open source. J Econ Behav Organ 95:90110
CrossRefGoogle Scholar
25.
Google Scholar
26.
1. Birnbaum A
(1968) in Some latent trait models and their use in inferring an examinees
ability. Statistical Theories of Mental Test Scores, eds Lord FM, Novick MR
(Addison-Wesley, Reading, MA), pp 395479
Google Scholar
27.
1. De Leeuw J,
2. Mair P
(2009) Gifi methods for optimal scaling in R: The package homals. J Stat
Softw 31(4):121
CrossRefGoogle Scholar
28.
1. Yen W
(1981) Using simulation results to choose a latent trait model. Appl Psychol
Meas 5(2):245262
Abstract
29.
1. Cook JR,
2. Stefanski LA
30.
1. Dabbish L,
2. Stuart C,
3. Tsay J,
4. Herbsleb J
(2012) Social coding in GitHub: Tansparency and collaboration in an open
software repository. Proceedings of the ACM 2012 Conference on Computer
Supported Cooperative Work (Burnham Inc Pub, Chicago), pp. 12771288
Google Scholar
31.
1. Tajfel H,
2. Turner JC
Google Scholar
32.
1. Bohn A,
2. Feinerer I,
3. Hornik K,
4. Mair P
Google Scholar
33.
1. Wu CG,
2. Gerlach JH,
3. Young CE
(2007) An empirical analysis of open source software developer motivations
and continuance intentions. Inf Manage 44(3):253262
CrossRefGoogle Scholar
34.
1. Kish L
(1965) Survey sampling (John Wiley & Sons, Inc. New York)
Google Scholar
35.
1. Armstrong JS,
2. Overton T
36.
1. Rizopoulos D
(2006) ltm: An R package for latent variable modeling and item response
theory analyses. J Stat Softw 17(5):125
CrossRefGoogle Scholar
37.
1. Stefanski LA,
2. Cook JR
(1995) Simulation-extrapolation: The measurement error jackknife. J Am Stat
Assoc 90(432):12471256
38.
1. Lederer W,
2. Kchenhoff H
Google Scholar
n-Depth
A leopard can't change its spots, but Microsoft's embrace of open source looks to prove
otherwise. Critics in the community have their doubts, but business realities are changing the
climate in Redmond.
By Steven J. Vaughan-Nichols
08/28/2015
Satya Nadella couldn't have found a better way to tell the IT world there's a new sheriff in town
than when he said these three words: "Microsoft loves Linux." If Nadella had uttered that
blasphemous phrase a decade ago, his predecessor and former boss CEO Steve Ballmer may
have fired him on the spot -- or at the very least thrown a large object at him. Ballmer made very
clear his view of Linux and the open source movement.
"Linux is a cancer that attaches itself in an intellectual property sense to everything it touches,"
Ballmer said in June 2001 in one of his legendary quotes. Setting the stage for Ballmer's
proclamation was Jim Allchin, senior vice president of the Microsoft Platform Group in February
2001. "Open source is an intellectual-property destroyer. I can't imagine something that could be
worse than this for the software business." Also setting the table for Ballmer was Senior VP
Craig Mundie. "As history has shown, while this type of model [open source] may have a place,
it isn't successful in building a mass market and making powerful, easy-to-use software broadly
accessible to consumers," Mundie said.
Microsoft's militant stance against all things open source rang unabated for much of the last
decade, despite rapid growth in deployments of the technology. Ultimately Microsoft started to
soften in 2006 with various acts of preservation in the ensuing years including the publication of
its 2008 interoperability principles, in which it promised to share APIs and documentation of its
protocols. At the time it was perceived as a bold move, yet few took it seriously and even fewer
trusted Microsoft.
Nadella was into his ninth month as Microsoft's newest CEO when he showed his affection for
Linux. At a press and analyst event in San Francisco among other things touting the
multiplatform support for the Microsoft Azure public cloud service, Nadella announced 20
percent of the instances were Linux-based. Nadella promised Microsoft would provide first-class
support for Linux. And then he said it: "Microsoft Loves Linux."
Wait! What!? I know the message hasn't got to everyone yet, but 2015's Microsoft is not the
fortress of proprietary software it was in 2001 or even 2011. This is not Steve Ballmer or Bill
Gates' Microsoft. It's Nadella's and he sees Microsoft's road ahead as walking hand-in-hand with
open source software development methods.
Azure clearly has become a multiplatform cloud. Indeed, 25 percent of Azure virtual
machines (VM) are now running Linux. In addition, Azure now supports five Linux
servers as VMs: CoreOS, CentOS, Oracle Linux, SUSE and Ubuntu.
Windows gurus Mark Russinovich and Jeffrey Snover had starring roles at the recent
Chef Software Inc. ChefConf 2015.
Now, this doesn't mean Microsoft is going to open source Windows or Office. Nor does it mean
that the folks in Redmond will be releasing MS-Linux... well, not anytime soon, anyway, though
Russinovich didn't even rule that out when asked recently. "It's definitely possible," Russinovich
responded to a question at ChefConf back in April. "It's a new Microsoft."
But, this is not the Microsoft that has throughout its history emphasized its embrace-and-extend
philosophy and spread fear, uncertainty and doubt (FUD) about open source. It's not that
Microsoft didn't see the value in open source software. Gates and company actually did. For
example, the first use of TCP/IP in Windows NT 3.5x and Microsoft's initial releases of HotMail
were both based on FreeBSD. Publicly, however, Microsoft set itself up as open source's mortal
enemy.
Responding in kind, open source developers have hated Microsoft. Linus Torvalds, for example,
quipped in 2003, "I've never seen it as a Linus versus Bill' thing. I just can't see myself in the
position of the nemesis, since I just don't care enough. To be a nemesis, you have to actively try
to destroy something, don't you? Really, I'm not out to destroy Microsoft. That will just be a
completely unintentional side effect."
Years later, Torvalds added, "I may make jokes about Microsoft at times, but at the same time, I
think the Microsoft hatred is a disease." (click here.) Microsoft hatred, however, is a common
open source disease. For years, there was no love lost between Microsoft and open source users,
developers and companies.
Microsoft in Transformation
So, what happened? Why is Microsoft now working hand-in-hand with open source projects?
Why is it partnering with open source companies such as Canonical Ltd. to enable Windows
Server to run as a guest OS on Ubuntu and OpenStack?
Allison Randal, president of the Open Source Initiative and a distinguished technologist at
Hewlett-Packard Co., said it best at the OSCon 2015 conference in Portland, Ore., the annual
gathering of the open source technical community. Randal said that with 78 percent of
companies now using open source solutions, and 64 percent participating in open source
projects, "it's table stakes to get in the software development game. If you're not doing it, you're
just going to get left behind." (Video of her OSCon 2015 presentation can be seen here.)
"If you're not doing it, you're just going to get left behind."
Allison Randal, President of the Open Source Initiative and a Distinguished Technologist at
Hewlett-Packard Co.
Microsoft knows this and is aware it can no longer take a "not invented here" attitude any more.
Microsoft's new management realizes the old proprietary software business model that served it
well for so many decades has been milked as much as it can be. Hence, Microsoft has been
moving to a service-oriented business model and, at the same time, the company is moving to
open source software to power those services.
Hard to swallow? Listen to Nadella, who said in a Wired interview, "If you don't jump on the
new, you don't survive." Exactly so. Eat or be eaten.
As early as 2007, Microsoft partnered with Novell (now SUSE), to bring SUSE Linux Server
Edition (SLES) VMs to Windows and Windows Server VMs to SLES. Today, this looks like an
incremental step. Then, it was a radical shift.
But, outside of SUSE, almost no one in the open source community really believed Microsoft
wanted to do anything except to show that Linux infringed on Microsoft intellectual property
(IP). At the time, it was probably both to provide the first VM bridge between Linux and
Windows and to establish a business claim to Linux IP. The latter would, in time, become the
template for Microsoft's billion dollar Android IP claims.
Even when Microsoft did try to work and play well with open source, it didn't go well. For
example, one early successful Microsoft project was IronPython. I say "was" because while the
language, an open source implementation of the Python programming language that's tightly
integrated with the .NET Framework, is alive and well, Microsoft spun it off on its own in 2010.
Jim Hugunin, IronPython's lead developer, left Microsoft for Google Inc. after this. Hugunin said
Microsoft's decision was the catalyst for his move. While Hugunin did his best to put a good face
on the move, he also said the "first release of IronPython was clearly broken in many ways. It
was released under an open source' license that was specific to Microsoft and not trusted by the
community."
The project finally came out successfully. "The culmination for me of the work on IronPython
was the creation of the Dynamic Language Runtime (DLR) which brought many of the ideas we
developed for IronPython deep into the .NET platform," wrote Hugunin. But, he ended by saying
he would no longer work on the project despite the fact that "it would also be very satisfying to
work on IronPython outside of the corporate constraints I've been living in for the past six years."
Burn!
At OSCon 2015, I conducted an informal survey of several dozen attendees. My question: "Do
you believe Microsoft is committed to creating open source software?" The answer,
overwhelmingly, was "no." With the occasional "Hell, no!" thrown in for spice.
They fear Microsoft is invading the open source community to further its real mission of
embrace and extend. Many remain convinced that Microsoft is simply trying to lull open source
companies to sleep, before trying to extinguish them. In short, their level of trust is low.
Eric S. Raymond, author of the seminal open source essay, "The Cathedral and the Bazaar,"
summed it up nicely when he said, "[It] might be sincere this time. Who knows? I don't, and long
experience makes me skeptical."
He has reason. In October 1998, Raymond uncovered the "Halloween Documents." These
internal Microsoft documents revealed that Microsoft already saw that open source software
"poses a direct, short-term revenue and platform threat to Microsoft, particularly in server space.
Additionally, the intrinsic parallelism and free idea exchange in [open source software] has
benefits that are not replicable with our current licensing model and, therefore, present a long-
term developer mindshare threat." The memo continued, "The ability of the [open source
software] process to collect and harness the collective IQ of thousands of individuals across the
Internet is simply amazing. More importantly, [open source software] evangelization scales with
the size of the Internet much faster than our own evangelization efforts appear to scale."
The memos also described how Microsoft might be able to fight off open source software by de-
commoditizing protocols and applications: "[Open source software] projects have been able to
gain a foothold in many server applications because of the wide utility of highly commoditized,
simple protocols. By extending these protocols and developing new protocols, we can deny
[open source software] projects entry into the market." In short: embrace, extend and extinguish.
Today, what's most striking about the Halloween Documents isn't their descriptions of how
Microsoft wanted to snuff out open source. It's how they accurately predicted that Microsoft
already knew, well more than a decade ago, that with its 1990s business model, Microsoft would
be unable to successfully compete with open source software. Hence, Nadella's conversion to
Linux.
Some open source users don't buy Microsoft's change of heart for one minute. Roy Schestowitz,
a software engineer and longtime Microsoft critic, still believes Microsoft is open source's mortal
enemy. Schestowitz recently wrote, "The Microsoft loves Linux' nonsense cannot be put to rest,
as that tired old lie keeps resurfacing in the media... Microsoft hates GNU/Linux. When it
participates it's in order to make Linux Windows-dependent (see Hyper-V, for instance) or
devour the platform in various other ways so as to make Microsoft's non-Windows cash cows
take over, in due course."
Another common feeling was expressed by Glenn Holmer, a Linux administrator. Holmer said,
"None of [what the company's doing] means anything as long as [it's] still patent-trolling
Android. When [Microsoft stops] doing that, [it] can start [its] journey toward credibility in the
open source world."
But, and this is different from the days when Linux and open source programmers and leaders
uniformly saw Microsoft as the evil empire, today, some open source developers, users, and
leaders are willing to give Microsoft a chance.
For example, Jos Poortvliet, community manager for ownCloud, an open source, private cloud
company, says, "Like many, I remain skeptical. But I'm open to be convinced."
The Linux executive leadership seems to be more willing to accept Microsoft's olive branch. Jim
Whitehurst, president and CEO of Red Hat Inc., says: "Last year when Nadella said Microsoft
loves Linux,' it marked a significant shift in tone and rhetoric from his predecessor who at one
point referred to Linux as a cancer.'"
"Microsoft seems very sincere in [its] willingness to work with not only
competitors, but also the open source community at large."
Whitehurst sees Microsoft's shift as a huge milestone for open source. "That open source has
gone mainstream and to be a major player in IT -- you have to embrace it. More and more
developers and businesses are turning to open source solutions in ways they've never done
before. Since his statement, Microsoft seems very sincere in [its] willingness to work with not
only competitors, but also the open source community at large, the open sourcing of .NET being
one key example of the company's growing acclimation toward open code."
ndeed, Jim Zemlin, executive director at the Linux Foundation believes the Microsoft open
source push is for real. "Microsoft gets open source," Zemlin says. "[The company is] both an
active contributor and consumer of open source software with large commitments to the Core
Infrastructure Initiative and Apache Software Foundation, among many others."
Moreover, he continues, it's not just talk. "I have personally worked with people at Microsoft on
IoT initiatives such as the AllSeen Alliance and Node.js and the latest major open source effort,
the Open Container Initiative. These individuals are smart, humble and enthusiastic in their
approach toward open source. Microsoft has clearly changed when it comes to open source and
is a better company as a result."
Sam Ramji, CEO at Cloud Foundry, an open source Platform-as-a-Service (PaaS) cloud
foundation, has a unique view of Microsoft's efforts. Before this position, he spent almost four
years as a Microsoft senior director of Platform Strategy and de facto leader of Microsoft's first
worldwide open source and Linux strategy organization.
"Open source is catching on and it's on the verge of being mainstreamed
at Microsoft. The real proof will come when open source is used in product groups."
Sam Ramji, CEO, Cloud Foundry, and Former Open Source Strategist at Microsoft
Ramji believes that under Nadella, "Open source is catching on and it's on the verge of being
mainstreamed at Microsoft. The real proof will come when open source is used in product
groups."
In particular, Ramji is watching what Microsoft is doing with Docker containers. "I'm told that
Microsoft really will be standardizing on glibc containers straight from Linux. The Microsoft
approach will be to call native existing Windows APIs. If Microsoft does deliver on this, and I
think [it] will, the barbarians will be inside the castle."
Still, Jacques likes what Nadella is doing. "He's the right captain, he's picked the right journey,
he put his boat on the journey, but it will take a lot of work to get there. Personally, I think
[Microsoft] could be more aggressive. It's great to see [the company] participating in the open
community, but I haven't seen [it] taking a leadership role. [it needs] to understand that We is
greater than me.'"
So what should Microsoft do? Jacques suggests it "look to Brocade and Cisco, which seem to be
doing well at this."
Specifically, "Microsoft should hire people with significant open source chops who could take a
true leadership role both inside Microsoft and in open source projects," suggests Jacques.
"[Microsoft] could do this by hiring someone with the right technical and collaborative DNA.
For example, hiring some senior people from Red Hat would be a great move."
What's Next?
So what is Microsoft really doing? I've been covering Linux almost since day one. Before Linux
was a gleam in Torvalds' eye I had been reporting and reviewing on all kinds of OSes,
development tools, and end-user programs, including Windows, and before that, MS-DOS. In
short, while I'm best known for my Linux and open source coverage, I know Microsoft quite
well, too.
I see Microsoft in a sea change. Given its druthers, Microsoft would love to be in a world where
licensed software came in shrink-wrapped boxes. Those days are gone.
Ballmer realized part of this. That's what led to Microsoft moving to a Software-as-a-Service
(SaaS) model. Nadella knew this wasn't enough. He also stepped into the CEO suite
understanding it wasn't enough to give lip service to open source, Microsoft must become an
open source power in its own right.
The writing was on the wall, in the Halloween Documents, years ago. Nadella is now reading
them to his company.
This will not be an easy move for Microsoft, but it is a necessary one. The Linux Foundation's
Zemlin is fond of saying that "Linux and open source are on the right side of history." He's right.
Apple Inc., with its wildly popular top-to-bottom proprietary hardware and software stack, can
survive with its old model. Microsoft, as its Nokia failure has shown, can't follow that path. For
Microsoft to stay the 21st century technology giant it was in the last century, it must transform
itself into an open source company.
Steven J. Vaughan-Nichols is an Ashville, N.C.-based freelance journalist who has covered the
open source software industry extensively from its beginnings and is currently a regular
contributor to Computerworld and ZDNet.
Given intense competition for the worlds best engineering talent, can your company really
afford to lock up its code behind proprietary licenses? Sure, if youre in the business of selling
software, giving it all away may not make sense. But the vast majority of companies dont sell
software, and should be contributing a heck of a lot more as open source.
How much more? (Almost) everything, to quote GitHub co-founder Tom Preston-Werner. The
post is a few years old, but judging from how the industry continues to treat software like
something to be hidden, not nearly enough people have read it.
Reasons To Open Up
If it sounds like business model suicide to open source (almost) everything, its not. At least,
not for the 99.999% of companies that sell services, not software. For this overwhelming
majority, Preston-Werner offers several reasons to open source code:
1. Open sourcing code is great advertising for you and your company
translat[ing] into goodwill for [your company] and more superfans than ever
before;
2. If your code is popular enough to attract outside contributions, you will have
created a force multiplier that helps you get more work done faster and
cheaper. More users means more use cases being explored which means
more robust code;
3. Smart people like to hang out with other smart people. Smart developers like
to hang out with smart code. When you open source useful code, you attract
talent;
4. If youre hiring, the best technical interview possible is the one you dont
have to do because the candidate is already kicking on one of your open
source projects; and
5. Once youve hired all those great people through their contributions,
dedication to open source code is an amazingly effective way to retain that
talent. Lets face it, great developers can take their pick of jobs right now.
These same developers know the value of coding in the open and will want to
build up a portfolio of projects they can show off to their friends and potential
future employers.
Keep in mind that open source really only helps you with a particular kind of audience. Chances
are, open source wont appeal to your end customers, whatever your product happens to be. But
thats OK, because open source is really just a way to engage the people who will help you build
your product.
Self-Selection Bias
Still not convinced? Take a look at Facebook, the company Ive called the worlds largest open
source company. In an excellent Fast Company article, Facebook open source chief James Pearce
touts the benefits it derives from open source, most notably recruiting:
It turns out that large percentages of our engineers will have known about our open-source
projects before they will have joined and they will say that it contributed positively to their
decision to join the company. Its a great window into the world of the sorts of problems that
we solve, and of course were hoping there are world-class engineers around the world who
would relish those kinds of opportunities and when they see the problems were solving will feel
the urge to take a look.
Ive written before about the importance of open source to recruiting the industrys best
engineers. Much of todays best software is written in the open, be it Hadoop or Spark or
MongoDB or Android. Its how companies engage with developers long before they engage with
end-customers.
Its also how companies learn from their peers, as Box did from Facebook.
Its why Netflix bothers to write technical blog posts that only a geek could love. (There are a
couple of things you should think about though before trying to score 100 billion records in
Pig.)
Its why your company needs to stop hoarding software that offers no competitive advantage and
start sharing. It will help you attract and retain talent which, in turn, will help your company
better serve end customers.
Thats how open source helps. And its why you should give (almost) all of it away. Now.
Daniele Procida,
In business, investment expects a return. So, what's the return on funding open source software?
It's not always clear to people why companies like Divio - businesses that need to make a profit -
would want to give money away to open-source communities and projects. Giving money away,
after all, is not the first thing one would put on a business plan.
In our case, we sponsor numerous projects with ongoing regular contributions, including the
Django Fellowship programme, Read the Docs and Django Girls. We also sponsor well over a
dozen open-source conferences each year, from large, established events such as PyCon US and
DjangoCon (Europe and US) to small, nascent ones such as Django Weekend and Python
Namibia. And not to mention Django Under the Hood, Write the Docs, PyCon UK - the list goes
on.
Why do we do it?
Some of this support does make obvious business sense. Companies like ours need to keep
recruiting, so sponsoring and attending a conference like a DjangoCon or a PyCon is an excellent
way to meet potential recruits. Being a sponsor also - quite rightly, because these events can't
even take place without sponsorship - earns a lot of goodwill, and that's worth something too,
because it's good for us to be known, liked and trusted.
This isn't the whole story though. It's true, some of these are great events at which to find new
recruits with valuable skills and great ways to earn goodwill. But: some of the events are not
aimed at highly-skilled developers at all, and in case, just how many new recruits and how much
goodwill does a company actually need?
In fact, a large part of the return on the investment is less immediate and less obvious.
A community is an organic thing. If it's to thrive, then even the parts of it that don't represent
immediate commercial value must be nurtured and invested in. There may be no immediate or
direct commercial value in sponsoring something like DjangoGirls or Python Namibia, but they
are valuable investments - long-term investments that are good for the community and its future.
Still, given a choice between making an investment that will pay off quickly, and one whose
benefits are spread out over time and amongst everyone involved with the software community,
why would it make sense to choose the latter? Surely it would make good business sense to let
other companies make those investments, so we get to benefit from their generosity? And why
spend money supporting an ecosystem or community - such vague, intangible things - when it's
possible to support just the particular projects and packages in it that really matter to us?
In other words, you might think that a sensible business should support open source software
communities in ways that will bring immediate benefits, and certainly not in ways that will
benefit its rivals as much as itself.
Firstly, if every company adopted that apparently smart and beneficial strategy, then it would
cease to be beneficial; we'd all lose out, and we would all have harmed the ecosystem that we
depend upon.
The fact that many companies - successful, profitable, sensible ones, like Divio - do make those
long-term investments in the community is an indication that this is a wise way for them to spend
their money.
Secondly, it is actually important for the benefit to be spread around. We rely on particular tools
and packages, but not just on them; we also rely on the health of the ecosystem and the
community in general, that makes it possible for those tools and packages to exist in the first
place, and for new ones to emerge.
Finally, there's another reason why, and it's easy to understand. Divio, just like most companies,
is staffed by people. Maybe businesses can't really love things, but the people in them can, and
the people who work with open source software care about and identify with the communities
behind their software.
It makes perfect sense for an individual to want to care for their community, and to find value
and reward in contributions to it, sometimes simply its own sake and not even for any personal
benefit in return. People enjoy knowing that the organisations they're involved in support and
belong to their communities in these ways.
When these values are shared by individuals at all levels of a company, when it's part of
company's culture and ethos, then 'good business sense' doesn't have to compete with altruistic
community support. Instead, the business is part of the community, and what's good for the
community is good for the business too - especially when the business is able to contribute to
that good.
And all this is why it makes perfect sense for companies like Divio to support not just events
such as the DjangoCons and PyCons that benefit us directly, but also the events like Django Girls
and Python Namibia that don't. It's why we are concerned not just about the health of projects
like Django and Read the Docs, that we use all the time, but also about the health of projects we
might never even use.
lint Finley Business
But last week, the company took this idea to the next level. It gave away all rights to Kubernetes,
a cloud computing system originally designed by Google engineers, asking a non-profit to
manage its development. It didnt just share some software code with the world. It agreed to let
an independent party oversee the development of the code.
Dubbed the Cloud Native Computing Foundation, the organization is just the latest in a series of
high profile new foundations now stewarding opens source projects created by large tech
companies. In the past year, weve also seen the launch of the Cloud Foundry Foundation to
govern a project originally released by VMware, the establishment of the Node.js Foundation,
thanks cloud services company Joyent, and the founding of the Open Container Initiative, thanks
to several different companies, most notably Docker and CoreOS. All four of these new
organizations are under the umbrella of the Linux Foundation, the organization originally
founded to manage the Linux Kernel, the core of all Linux operating systems.
Open source foundations are nothing new. Linux Foundation has been around since 2007, and
other major projects like the Eclipse code editing tool and the Apache web server have been
governed this way for even longer. Many of the most important open source projects in recent
years, such as the Hadoop big data crunching platform and the database system Cassandra, are
managed by the Apache Foundation. But its unusual to see so many new foundations created so
quickly.
While the Linux kernel, Apache server, and Hadoop platform were all started initially as non-
commercial projects, some of these new projects, like Cloud Foundry, have always been
corporate products. And though some major open source projects remain corporate property,
such as the MongoDB database, its becoming harder and harder to name significant projects that
arent now part of a foundation. Thats a good thing.
Giving It Away
Why are so many companies giving away their intellectual property? Its not happening for
altruistic reasons. In his keynote at the OReilly Open Source Conference in Portland, Oregon
last week, Cloud Foundry Foundation CEO Sam Ramji argued that the shift is being driven by
economics.
Companies like Google want others to use their open source software since it can help drive the
use of online services, like Googles cloud computing tools. They want others to contribute code
to this software too. But increasingly, others dont want to use or contribute to projects unless
theyre independently managed.
Venture capitalists have been pouring money into companies dedicated to commercializing open
source projects, whether thats by providing support and services for open source projects or by
building proprietary products atop open source code. Docker, for example, has raised $162
million million according to Crunchbase. But new business models for open source create new
challenges for the companies that maintain them.
Turns out with all of that money, you start to generate distrust, Ramji said. So some of the old
nature of the open source model is starting to break down as these projects become bigger and
more valuable, I think thats where foundations come in.
For example, IBM and HP offer cloud services based on Cloud Foundry. It would be in their best
interests to contribute back to the main Cloud Foundry project, but they might be disinclined to
do so by the fact that they compete directly with Pivotal, a company spun out of VMware and
EMC in part to commercialize Cloud Foundry.
When Pivotal owned all of the work that HP and IBM developers were putting into Cloud
Foundry, there was always the possibility that the company could change the open source
licensing, reject a competitors contributions, or make changes to the project that didnt fit with
the plans of outside contributors. One company was calling the shots, and it had to put its own
interests first. So by putting its code into a neutral, independent organization, competitors can
make mutually beneficial decisions without having to worry that all of their efforts will disappear
overnight.
The Downside
But there are downsides to the foundation model, argues Derek Collison, one of the original
creators of Cloud Foundry and the founder of Apcera, a startup that makes tools for managing
cloud infrastructures. Foundations move slower than projects governed by a small team with
clear goals, and innovation can be hampered by infighting, he says, echoing the reasons many
open source project leaders have given for not placing their projects into foundations over the
years.
Design by consensus has never worked, he says. And it never will unless you have someone
who says I know the pain points, I know how to drive it forward.'
But Collison isnt totally against the idea of foundations. In fact, Apcera is part of the Open
Container Initiative that is helping standardize an increasingly popular technology called
containers. He argues that relatively little design or innovation has to happen within the
foundation. With the standard established, the member companies are free to innovate as they see
fit, adding their own unique features atop the standard.
Standards, Please
Standardization is a major driving force many of these new projects, Linux Foundation director
Jim Zemlin told us last year. Providing a huge standards document to a light bulb manufacturer
wont help it make better, cheaper bulbs, Zemlin told us. But if you hand them the open source
code, then they can just start doing it.
And even if foundations move more slowly than companies, they might still win out in the end.
Ramji cited a study conducted by Henrik Ingo in 2010 to compare governance models of open
source projects. Ingos conclusion: There appears to be a glass ceiling limiting the growth of of
single vendor projects. In other words, if you want your project to grow, giving away the code to
an independent organization is the best way to do that.
Skip Social. Skip to: Latest News. Share this story on Facebook Share this story on
Twitter Share this story on Pinterest Share this story via Email
Linux
open-source
Share
Share 6 992
Tweet
Pin
Its the latest in series of projects recently open sourced by large tech companies all focused on a
branch of AI called deep learning. Google, Facebook, and Microsoft have mainly used these
systems for tasks like image and speech recognition. But given Amazons core business, its not
surprising that the online retailers version is devoted to selling merchandise.
We are releasing DSSTNE as open source software so that the promise of deep learning can
extend beyond speech and language understanding and object recognition to other areas such as
search and recommendations, the Q&A section of Amazons DSSTNE GitHub page reads. We
hope that researchers around the world can collaborate to improve it. But more importantly, we
hope that it spurs innovation in many more areas.
Along with the idealistic rhetoric, open sourcing AI software is a way for tech industry rivals to
show off and one-up each other. When Google released its TensorFlow framework last year, it
didnt offer support for running the software across multiple servers at the same time. That meant
users couldnt speed up their AI computations by stringing together clusters of computers the
same way Google could running a more advanced version of the system internally.
That created an opening for other software companies like Microsoft and Yahoo to release their
own open source deep learning frameworks that support distributed computing clusters.
Google has since caught up, releasing a version of TensorFlow that supports clusters earlier this
year. Amazon claims its system takes distribution one step further by enabling users to to spread
a deep learning problem not just across multiple servers, but across multiple processors within
each server.
Amazon also says DSSTNE is designed to work with sparser data sets than TensorFlow and
other deep learning frameworks. Google uses TensorFlow internally for tasks such as image
recognition, where it can rely on the Internets vast store of, say, cat photos to train its AI to
recognize images of cats. Amazons scenarios are quite different. The company does sells
millions of different products. But the number of examples of how the purchase of one product
relates to the purchase of another are relatively few by comparison to cats on the Internet. To
make compelling recommendationsthat is, to recommend products that customers are more
likely to click on and buyAmazon has a strong incentive to create a system that can make good
predictions based on less data. By open sourcing DSSTNE, Amazon is increasing the likelihood
that some smart person somewhere outside the company will help the company think of ways to
make the system better
Why the open source business model is a failure
Most open source companies can't thrive by selling
maintenance and support subscriptions. But the cloud may be
the key to revenue generation.
By Paul Rubens
| Follow
Video
Why you need Open Source
Open source software companies must move to the cloud and add proprietary code to their
products to succeed. The current business model is recipe for failure.
That's the conclusion of Peter Levine, a partner at Andreessen Horowitz, the Silicon Valley
venture capital firm that backed Facebook, Skype, Twitter and Box as startups. Levine is also
former CEO of XenSource, a company that commercialized products based on the open source
Xen hypervisor.
Levine says the conventional open source business model is flawed: Open source companies that
charge for maintenance, support, warranties and indemnities for an application or operating
system that is available for free simply can't generate enough revenue.
"That means open source companies have a problem investing in innovation, making them
dependent on the open source community to come up with innovations," he says.
Why is that a problem? After all, the community-based open source development model has
proved itself to be more than capable of coming up with innovative and very useful pieces of
software.
Revenue limits
The answer is that without adequate funding, open source businesses can't differentiate their
products significantly from the open source code their products are based on, Levine maintains.
Because of that there's less incentive for potential customers to pay for their products rather than
continue using the underlying code for nothing. At the very least it limits the amount that open
source businesses can hope to charge putting a cap on their potential revenues. It's a vicious
circle.
"If we look at Red Hat's market, 50 percent of potential customers may use Fedora (the free
Linux distribution,) and 50 percent use Red Hat Enterprise Linux (the version which is supported
and maintained by Red Hat on a subscription basis.) So a large part of the potential market is
carved off why should people pay the 'Red Hat tax'?" Levine asks.
You could argue that this is actually good for businesses, because the availability of open source
software at no cost provides competition to open source companies' offerings based on the same
code, ensuring that these offerings are available at a very reasonable price.
But if open source businesses can't monetize their products effectively enough to invest in
innovation, then potential corporate clients can't benefit from the fruits of that innovation, and
that's not so good for customers.
The problem is compounded when you consider that open source companies' products are not
just competing with the freely available software on which their products are built. It's often the
case that they also have to compete with similar products sold by proprietary software
companies. And that particular playing field is often an uneven one, because the low revenues
that open source companies can generate from subscriptions mean that they can't match the huge
sales and marketing budgets of competitors with proprietary product offerings.
It's an important point because although sales and marketing activities are costly, theyre also
effective. If they weren't, companies wouldn't waste money on them.
So it follows that open source companies miss out on sales even when they have a superior
offering, because having the best product isn't enough. It's also necessary to convince customers
to buy it, through clever marketing and persuasive sales efforts.
The result, says Levine, is that open source companies simply can't compete with proprietary
vendors on equal terms. "If you look at Red Hat, MySQL, KVM in every case where theres a
proprietary vendor competing, they have more business traction and much more revenue than
their open source counterparts."
As an illustration of the scale of the problem, Red Hat is generally held up as the poster child of
open source companies. It offers an operating system and a server virtualization system, yet its
total revenues are about a third of specialist virtualization vendor VMware, and about 1/40th of
Microsofts.
Hybrid future
This is why Levine has concluded that the way for open source companies to make money out of
open source software is to abandon the standard open source business model of selling support
and maintenance subscriptions, and instead to use open source software as a platform on which
to build software as a service (SaaS) offerings.
"I can run a SaaS product by using Fedora as a base, but then building proprietary stuff on top
and selling the service. So the monetization goes to the SaaS product, not to an open source
product," says Levine. "I think well start to see an increasing number of SaaS offerings that are
a hybrid of open source and proprietary software."
He adds that many SaaS companies including Salesforce, Digital Ocean and Github (two
companies Andreessen Horowitz has invested in) already use a mix of open source and
proprietary software to build their services.
And Levine says that Facebook is the biggest open source software company of them all. "I was
shocked when I realized this, and Google probably is the second biggest," he says.
Facebook has developed and uses open source software for the infrastructure on which its social
network is built, and adds its own proprietary software on top to produce a service it can
monetize. Google also generates a large volume of open source infrastructure code, although its
search and advertising software is proprietary, he adds.
While the existence of free-to-download software undoubtedly makes it harder for open source
businesses to monetize the same software by adding support, maintenance and so on, it's also the
case that these low-cost alternatives must make life more difficult than otherwise for proprietary
vendors trying to sell their products into the same market.
That's because these low-cost alternatives necessarily make the market for proprietary software
smaller even if proprietary companies have higher revenues that they can use to innovate,
differentiate their products, and market them.
This could help explain why some proprietary software companies are moving their products to
the cloud, or at least creating SaaS alternatives. A mature product like Microsoft's Office suite
can largely be functionally replicated by an open source alternative like LibreOffice, but
Microsoft's cloud-based Office 365 product takes the base Office functionality and adds extra
services such as file storage, Active Directory integration and mobile apps on top.
That's much harder for anyone to replicate, open source or not. And it suggests that in the future
it will be all software companies, not just open source shops that move to the cloud to offer their
software as a service.
To comment on this article and other CIO content, visit us on Facebook, LinkedIn or
Twitter.