An alternative to Dual Licensing

In a previous post, we have made our point that dual licensing was unfair and community debilitating. 

In a search for a better solution, I met this weekend with Ludovic Dubost CEO, Founder and original contributor of XWiki. XWiki.com  is making its income out of enterprise support, software guarantees and integration projects. This last part being a substantial source of revenues for XWiki.com

In January 2006 Ludovic faced an important decision: at that time, XWiki was a GNU GPL project with a reasonably growing community but because of it Ludovic was loosing integrations/outsourcing contracts. Some of his potentially largest customers were not comfortable with the GNU GPL.

Many companies in XWiki.com's situation would have opted for a Dual-Licensing (DL)  strategy but Ludovic thought that it would be somewhat turning back to the community and he decided for a GNU LGPL  instead.

Today, the company is growing with both large enterprise customers and a solid group of external contributors to whom XWiki doesn't even ask to sign any dual-license CLA/release.

Yet, the fsf thinks that the GNU LGPL is a  necessary evil of sort:

Proprietary software developers have the advantage of money; free software developers need to make advantages for each other. Using the ordinary GPL for a library gives free software developers an advantage over proprietary developers: a library that they can use, while proprietary developers cannot use it.

Except that nowadays, proprietary software companies are using the GPL through DL schemes to maintain a monopoly over the commercial exploitation of open sources released under the GPL!

As a result,  DL companies don't have any incentive to really take care of the community, not even to contribute. But under the  GNU LGPL, XWiki.com is naturally pushed to tightly cooperate with the community to which the real value has been transfered: XWiki must foster a happy community.

What about the contributions? On the developer side everything is rosy: no more dubious dual CLA. On the open source side it's even better. Ludovic tells me that more than 50% of his integration projects lead to significant open source contributions.

Why would you ask since most of those developments are made for (non-distributed) internal operations? Why would XWiki.com insists on their customers so that contributions could be released back to the community when it's not even mandated by the licenses? Why would the customers accept?

Three main reasons:

1. It keeps the community happy, growing the flow of new quality contributions
2. It makes the proprietary maintenance less burdensome: it helps keeping the proprietary customer projects in sync with the main release of XWiki.
3. Customers are not locked-in anymore. Since any company can use the LGPL they know they can use another integrator in the future.

Would XWiki have chosen the DL path:

1. XWiki would have had much less incentive to please their community since a DL/GPL would have given them a quasi commercial monopoly
2. Less customers would have chosen XWiki not only because  of  perceived FUD factors but also because it would have meant a lock-in with a specific integrator: XWiki.com
3. The developer community would have been prevented to develop beyond a slow-growing group of mostly opportunistic patchers.
   

But, what if one of XWiki.com customers decides to release an XWiki-including product? Under the LGPL they won't be forced to contribute back to the commons.

Maybe. Think about it:  if they decide to release an XWiki-based product, this new company will be faced with the same challenges than XWiki: keeping the community happy and decreasing their maintenance risk.

Bottom line:

• It's an alternative to perverting the GNU GPL into a commercial monopoly-making strategy

• It's an alternative to demanding DL/CLA and therefore to controlling developers IP

• It's good for the code: with the LGPL fostering both developer community and enterprise adoption the common code should grow much faster than if it had stayed in the limbos of the dual/GPL.

• It's good for the developers: the community will gain many more friendly companies looking out for them

                       How could the GNU LGPL be bad for open source?

 

This article  is cited or quoted by:

• Scott Ruecker on Lxer
• Roy Schestowitz on Boycott Novell
• D. Girard on Application-Servers.com
• FreeSoft News
• Root.cz

Comments

[Matt Asay]

You've forgotten two key things, Pierre:

1. The company you cite is making very little money and so is hardly something to which a MySQL might aspire.

2. Most companies will receive very few outside contributions. It's not the CLA that prevents people from contributing to MySQL, JBoss, etc. It's the fact that it's very hard to find the time and aptitude to contribute significant amounts of code to anyone not paying your bills. This is why most of the Linux kernel is developed by paid developers - people getting paid to write the Linux kernel.

I wish it were different, but it's not.

[mtg]

@matt (Thanks for taking the time to comment)

[matt] 1. The company you cite is making very little money and so is hardly something to which a MySQL might aspire.

[mtg] Well JBoss was satisfied with the LGPL model and according to M.Asay :) http://www.cnet.com/8301-13505_1-9853461-16.html
JBoss valuation was commensurate to that of MySQL.

[mtg] Furthermore according to SUN http://java.sun.com/developer/technicalArticles/mysql_acq/
and also to c|net Gordon Haff http://www.news.com/8301-10784_3-9853877-7.html
MySQL has had a total of 100M downloads for a current yearly revenue of 70M USD. That gives us a ratio of 0.7 USD/download to date. From what I've heard XWiki is 10 times (!!) this ratio with currently 10K downloads / month.

[matt] 2. Most companies will receive very few outside contributions. It's not the CLA that prevents people from contributing to MySQL, JBoss, etc. It's the fact that it's very hard to find the time and aptitude to contribute significant amounts of code to anyone not paying your bills. This is why most of the Linux kernel is developed by paid developers - people getting paid to write the Linux kernel.

[mtg] For once XWiki too hired some "professional contributors": "Don't we all already know that Open Source if Free once it's paid for?" Told me Ludovic at somepoint. :-)

[mtg] But the difference here is that a LGPL model generates much MORE contributions from *organizations* who otherwise would haven't. For instance , www.curriki.org accepted to contribute to XWiki and likely wouldn't have on a dual GPL/proprietary model.

[mtg] No to mention the crux of the issue with DL: forking. It's painful for developers (finding new key people) and IMPOSSIBLE for a company (since having the DL procures a direct unfair advantage)

And if the mere possibility of forking is so difficult to envision then it doesn't play its role of balancing the behavior of the DL company vis-à-vis the community...

[dsf]

>> In a previous post, we have made our point that
>> dual licensing was unfair and community
>> debilitating.

Having followed your articles for a while, I can't say as I see anyone agreeing with you! The comments to the above referenced article were all negative.

So your argument is enterprise support, software guarantees and integration projects.

How many of you have bought support and/or guarantees? And just how many projects have tons of "integration" opportunities?

And once again we're talking of the same companies who have had some success, while tens of thousands crash and burn.

But hey, lets all be social and communal!

[mtg]

@dsf

My argument is that the DL model is buggy and that using the GPL to promote a de facto commercial monopoly is really at the complete opposite of what open source is/should be about.

So I'm trying to identify companies making a living out of business models that don't include DL.

Some people agree, some don't but in both cases I welcome constructive and insightful participation.

[dsf]

What's a business to do when programmers have collectively decided they won't pay for software? Programmers hurting programmers! Imagine any other profession’s workers saying “I won’t use your product unless it’s free.” Imagine:

- Artists: “I won’t pay for your painting, as all art should be free.”
- Doctor: “I won’t pay for your surgery, as all medical treatments should be free.”
- Grocery Store Owner: “I won’t pay for your food, as all food should be free.”
- Car Dealer: “I won’t pay for your car, as all cars should be free.”

Yes, it is ridiculous but it is where we are in the software world.

DL came about as more and more companies started to feel the pressure from the open-source community. "I won't use your software unless it's open-source." and "I won't pay for it if it is." is the rule for the day. So, in an attempt to remain in business, businesses came up with the idea of dual licensing their software. Unfortunately, we still have the "I won't use your software unless it's open-source." and "I won't pay for it if it is." mentality, so no one is winning, and in the end all software will end up being developed in every one’s spare time (by programmers with nothing better to do).

It all appears to be a lost cause! There are little to no opportunities to start your own "successful" business any more, and we are all doomed to work for the corporate IT department. At least until they figure out how to get what they need for free.

[mtg]

@dsf

You're right: JBoss was acquired $350M (basically for free) by a company selling free products (reflected by its market cap of $4.5B).

Here is a free book by Lawrence Lessig. I know you're going to love it:

It's called Free Culture: http://www.free-culture.cc/

See, always this conflict between the Farmer and The nomad ( http://blog.milkingthegnu.org/2008/05/nomadism-open-s.html )

[dsf]

Yes, and what I'm saying is we are always referring to the same companies (RedHat, JBoss, etc.) as there are so few. Furthermore, $350m is a perceived value that is not backed by real assets.

[dsf]

You guys blow me away. So the anti-dual licensing model you are really talking of here, is creating an open-source server solution that is packaged as a desktop trial version, and available as a full fledged server version for much larger sums of money (1 K€ for packaging, and 3 K€, 6 K€, 12 K€ for database sizing). So, it’s open-source, but to actually use it as intended you have to pay at least 4 K€.

Interesting, but how does this all pertain to the open-source GNU GPL license? It appears to me that XWiki makes money by providing a useable configuration (that isn’t possible or easy to do from the available source), instead of a desktop trial version that’s freely available. Again, nothing to do with the GNU GPL license, and since the software is server based and isn’t being redistributed by the average customer, using GNU GPL is irrelevant. In this case using GNU GPL only becomes relevant when a business uses XWiki as part of its non open-source product offering when released to the public.

Furthermore, I find it hard to believe their profits come mostly from integration with other products, when they are charging 1 K€ out of the starting block, and 3, 6, or 12 K€ per year for a database sized subscription.

Which, by the way, has me asking “Why are we talking of support, guarantees and integration here when you can’t use their product without a minimum monetary output of 4 K€?”

[mtg]

@dsf

BTW XWiki focuses on enterprise solutions.

If you're alluding to the hosted solutions I don't see the problem, many companies are offering a hosting package, the cost of which is related to their target market.

As to the sources and the packages I don't know what you're talking about; All the packaged downloads are here: http://www.xwiki.org/xwiki/bin/view/Main/Download
and all the sources accessible here:
http://dev.xwiki.org/xwiki/bin/view/Community/SourceRepository

[Thomas Hansen]

OMG, you just won't give up...?
I think I have yet to read one of your blogs (and I've read quite a few) which didn't have at least 80% of commenters commenting negatively towards you and your logic...
Here's my third;
Regarding the "services business model" I can assure you that it is a) unfair, b) not working and c) creating crappy products, at least in the end. a because the "extra services companies" ends up not helping their users unless they can bill them meaning that you cannot have support, help or documentation unless you "pay a little bit extra". Which leads inevitable to b. c comes into the picture since they don't earn money on the stuff they SHOULD be doing so they often have incentives to creating crappy products since that will in the short terms increase revenue due to being able to sell extra support, consultancy and documentation.

If a company making its living out of selling "documentation", "support" and "consultancy" on top of their own products (which are free) let me ask you this question; How long do you think it'll take before they understand that by "obscuring" their APIs, adding extra load to the "learning curve" and basically releasing a *crappy* product they'll actually EARN MORE MONEY?

"Extra Services Business" model does NOT work, I know because I am sitting next to 40 people which are doing this. They used to be 80 people 6 months ago, and they're basically facing bankruptcy due to not being able to compete with another product which haven't had the financial incentives to create crappy products as they themselves have.

Unless you earn your money where you SHOULD earn your money(which is on your PRODUCT), your product will ROT! It might take some time before this happens, but eventually that is the inevitable outcome of your fancy "free beer business model"...

Unless you have a product which people are creating derived works of and these people are distributing their derived works or you are getting paid by IBM to work on the Linux Kernel or something you can just forget about earning money on Open Source. Look at the FUD coming out from e.g. MySQL about "licenses needed" and so on to get an example of how wrong things might turn out. Nobody was creating "derived products" from their products, still everyone asking about help about understanding their license would get a standard reply back saying; "it's probably safer to purchase a license, but contact a lawyer if in doubt".

That answer will NEVER be given by Trolltech or Gaiaware (us).
The reasons for this is pretty straight forward; we don't have financial incentives to say such things! (creating FUD about GPL)

Now if you are more happy paying for the "services" we can probably cut you a deal for our product where you get a commercial/proprietary enabling license for FREE and you PAY us for the rights to be able to post in our forums. Would you purchase our "services" then in order for you to be able to use our product? Or is your agenda just to get "free beer"...?
If it's the first one you can send me an email, you know where to find me and I'll make sure you get the PRODUCT for FREE but that what you're paying for is the SUPPORT. If it's the latter then you're free to go and find yourself a "free beer" product to use. We're NOT there, neither is Trolltech or any of the other Dual License companies and they will probably never be there either...

[mtg]

Well Thomas, thanks for reading assiduously a blog you don't agree with. It proves you have a great mind ready to discuss new ideas.

As to the "service doesn't work" Red Hat proved it worked long enough, at least long enough for them to get large customers after which they started indeed to "close" a little bit more their environment.

Anent XWiki, remember a large part of their revenues comes not only from services but also from integration: basically they are building various enterprise application for large companies, the building blocks of which are handed back to the open source community.

I agree that this model might not work with every company, maybe not in every country/market. But it works for them.

Just like service worked for Cygnus, Red Hat and many others.

BTW just curious, why don't you post under one name only?