The Affero GPL does not solve the open source/cloud revenue dilemma

A number of people have recently raised the issue of the threat that cloud computing poses to the monetization of open source by specialist vendors, including Savio Rodrigues, Matt Asay, and Mike Hogan.

I believe that cloud computing provides an opportunity for open source specialists, but agree that cloud services based on open source code could potentially eat into the business opportunities for open source specialists since the cloud providers have no requirements to pay for a commercial relationship with the vendor concerned.

Some commenters have proposed that the Affero GPL is a potential solution to this problem since it applies the requirements of the GNU GPL to code distributed over a network, updating the GPL for a SaaS and cloud age. However, there are a number of reasons why this is not necessarily the case. Here are three of them.

1/ As Matt Asay points out, if MySQL were released under the AGPL Google would in all likelihood decide to find something else to base its development on, rather than be forced to contribute back its own code.

2/ The Affero GPL doesn’t prevent the deployment of open source software in networked environments, it just requires contribution of modifications. In the case of Microsoft supporting MySQL and Tomcat on Azure and Amazon using MySQL to create its Relational Database Service it is questionable whether there is any significant modification at all. Even if there is, it is likely to be very specific to the Azure and AWS cloud platforms. As such they are likely to be of no use to anyone else, or significant enough to trigger the avoidance of Affero GPL code, as above.

3/ The Affero GPL is unsuitable for vendors that want to encourage SaaS delivery by other parties. I spoke to an open source e-commerce vendor earlier this year looking at opportunities for cloud computing. A perfect candidate for the Affero GPL? Well no, because the company’s software is often embedded within applications from ISV partners that are delivered as a service. The Affero GPL would potentially force those ISV partners to publish their own code under the Affero GPL license.

Which is not to say that the Affero GPL does not have its uses, but it is not a solution to the problem that cloud providers can use open source code without paying for commercial relationships with the associated vendors.

Tags: , , , , , , , , , , , , , , , , , ,

14 comments ↓

#1 Twitter Trackbacks for 451 CAOS Theory » The Affero GPL does not solve the open source/cloud revenue dilemma [the451group.com] on Topsy.com on 11.23.09 at 8:22 am

[…] 451 CAOS Theory » The Affero GPL does not solve the open source/cloud revenue dilemma blogs.the451group.com/opensource/2009/11/23/the-affero-gpl-does-not-solve-the-open-sourcecloud-revenue-dilemma – view page – cached An open source blog by The 451 Group. […]

#2 Computer Internet and Technology Articles. » Blog Archive » 451 CAOS Theory » The Affero GPL does not solve the open source … on 11.23.09 at 8:31 am

[…] See more here: 451 CAOS Theory » The Affero GPL does not solve the open source … […]

#3 451 CAOS Theory » The Affero GPL does not solve the open source … | Open Hacking on 11.23.09 at 9:02 am

[…] the original post: 451 CAOS Theory » The Affero GPL does not solve the open source … This entry was posted on Monday, November 23rd, 2009 at 7:18 am and is filed under Linux, News, […]

#4 uberVU - social comments on 11.23.09 at 9:06 am

Social comments and analytics for this post…

This post was mentioned on Identica by caostheory: New post: The Affero GPL does not solve the open source/cloud revenue dilemma. http://bit.ly/6helNt

#5 Declan on 11.23.09 at 6:57 pm

Point 1 is fair enough – this is a use case where it does not make sense to choose the AGPL.

Regarding point 2, surely those modifications are
a) Generally uninteresting to other people so who cares if the changes to run on Azure are not released
b) “The Affero GPL doesn’t prevent the deployment of open source software in networked environments, it just requires contribution of modifications”. So what? AGPL is not about stopping people deploy in network environments. All it wants is the code back. This is no different to the intentions of the GPL regarding traditional applications.

Regarding point 3, could the code not be dual licenced for a fee to the ISV’s if they are partners? If it wasn’t for the AGPL they would not be tempted to do this?

#6 Matthew Aslett on 11.24.09 at 1:40 am

Hi Declan,

RE: So what? I agree, I wasn’t criticizing the AGPL but some people evidently think that licensing under the AGPL prevents Gooogle or AWS or whoever from running the code. I was just pointing out a misunderstanding.

RE: dual licensing. Perhaps, although only if the company owns all the copyright.

#7 The Art of Being Dorian » Tech News – Canonical Bringing Music to Ubuntu, LH Strikes Again & Affero GPL Can’t Fix the Cloud on 11.24.09 at 7:03 am

[…] is the emerging cloud computing landscape.  However Matthew Asslett (451 Group) points out the Affero GPL doesn’t negate the threat of no monetary contributions from cloud providers to comm….  Yes it doesn’t.  But the licenses were designed to get  source code contributions from […]

#8 Barry Klawans on 11.24.09 at 12:54 pm

I think the strength of the Affero GPL depends directly on the nature of the project it is applied to. If you are one of several open source implementations of a standard, the people building the cloud stack have choices. In the case of a RDBMS, if MySQL was under AGPL, I could use PostgreSQL, Derby, etc.

If I need to include a component in an area where there are no standards, then the implementations are not interchangeable. I will probably have to choose the one with the feature set I want to expose via my cloud based application, and live with its license.

Suppose my first choice is using the Affero GPL? I have three choices: 1) Live with it, and contribute my changes back to the project. 2) Choose a different project with a different license. 3) Write my own from scratch. I’ll argue that 2) isn’t really a choice – by definition I’m going with my second (or third or fourth) choice, and I’ll need to spend resources enhancing it to get the functionality I want that I would get automatically via 1). In this case, I can either re-apply my changes each time the project gets updated, or contribute my changes back so they are incorporated in the main code-base. Since contributing back is what I want to avoid, that leaves me with the choice of either re-appling my patches, will be a HUGE maintenance headache, or forking the code (if the license allows it.) Forking isn’t all that different from choice 3)

I think few if any companies can afford to either fork and advance an existing project or create their own from scratch. Even a company the size of Google will have to decide how much it wants to spend re-inventing a wheel to avoid licensing restrictions.

#9 Matthew Aslett on 11.25.09 at 4:27 am

Thanks for the feedback Barry. Good points from the cloud platform provider’s perspective.

#10 Lawrence Rosen on 11.24.09 at 7:38 pm

I read Matt Aslett’s original blog, and then the addendum he had to post on 11/23 because he left out the important AGPL alternative.

The AGPL plugs the GPL loophole that allows companies to provide services to third parties using *proprietary unpublished modifications to open source software*.

I agree with Matt that the AGPL (or even OSL 3.0 that I wrote and prefer to the AGPL) doesn’t solve all the problems he identified in his first post. His second headline calls the main problem “the Revenue Dilemma”, which is as fair as you can be in identifying any commercial company’s problems using copyright and contract law alone to protect a revenue stream. Age old problem you’re talking about, Matt. It may be news to Matt, but it isn’t news to those of us who have been helping open source and proprietary companies deliver profitable solutions for over a decade now. The revenue dilemma can only be solved with excellent products and services delivered with expertise.

I believe that the AGPL and OSL 3.0 both offer a solution to a type of free-rider problem, such as Google’s use of improved open source GPL and other code that is never shared with its competitors and customers. I have no problem in principle with that model, except that I believe that Google (and its ilk) ought to pay for that privilege to use free software for profit. The AGPL and OSL 3.0 plug that loophole. Even that is not necessarily enforceable strictly, because software can be run in lots of functional ways without invoking copyright infringement at all. And big companies like Google will continue to refuse AGPL and OSL 3.0 software altogether because their business choice is NOT to share.

Isn’t the notion of “privilege of subscribing for a fee” a fundamental basis for any open source company’s success? So if some customers (such as Google) refuse software under a strict license like the AGPL or OSL 3.0, then your company should offer them a commercial license for a fee. That goes a long way toward solving the revenue dilemma. Your customers will understand.

/Larry

DISCLOSUR: I wrote this comment originally as a member of the Jaspwersoft Advisory Board and am publishing it here with their permission.

Lawrence Rosen
Rosenlaw & Einschlag, a technology law firm (www.rosenlaw.com)
3001 King Ranch Road, Ukiah, CA 95482
Office: 707-485-1242 Cell: 707-478-8932
Apache Software Foundation, member and counsel (www.apache.org)
Open Web Foundation, board member (www.openwebfoundation.org)
Stanford University, Instructor in Law
Author, Open Source Licensing: Software Freedom and Intellectual Property Law (Prentice Hall 2004)

#11 Matthew Aslett on 11.25.09 at 4:24 am

Hi Larry,

Many thanks for the comment on the two posts. As you say, in many ways there is nothing new here. The revenue dilemma isn’t news to me, but I wrote about it to draw attention to what I see as a common misunder5standing about the Affero GPL: that is prevents the use of code in cloud environments.

I agree there is much more to this than I covered in either of these two posts and I have received some very interesting feedback from very knowledgeable people (both here and via email) which will will include in a more formal analysis of the situation in 2010.

Thanks
Matt

#12 La licence AGPL résout-elle tous les problèmes de l’open souce et du cloud computing ? – Philippe Scoffoni on 12.02.09 at 5:41 pm

[…] et voyons en quoi cette licence ne résoudrait pas tous les problèmes. Matthew Aslett, encore lui, a publié un billet dans lequel il évoquait au moins trois raisons […]

#13 ☞ Protecting Rights | Developer @ UniUrs on 01.11.10 at 4:32 am

[…] The Affero GPL does not solve the open source/cloud revenue dilemma And that's hardly a surprise becuase AGPL is purely about license-derived software freedom. We're on beyond licenses now, and we need to condition people to consider all the degrees of freedom and not just one. (tags: AGPL Licensing Software Cloud FOSS OpenSource) […]

#14 Tech News – Canonical Bringing Music to Ubuntu, LH Strikes Again & Affero GPL Can’t Fix the Cloud | The Art of Being Dorian on 08.23.10 at 1:39 pm

[…] is the emerging cloud computing landscape.  However Matthew Asslett (451 Group) points out the Affero GPL doesn’t negate the threat of no monetary contributions from cloud providers to comm….  Yes it doesn’t.  But the licenses were designed to get  source code contributions from […]