Friday, May 30, 2014

Commercializing Open Source Licenses

Nearly a year ago in this blog, I had a post up arguing that Richard Stallman's position on the necessity of using strong copyleft licenses to protect the open source movement was misguided. I'm following that post, now, by explaining that, in fact, not only are strong copyleft licenses inappropriate for certain business cases, but, in others, they are a powerful tool in the monetization of commercial software - where Stallman seems to want to live in a world where copyleft licenses exist only to promote the open source movement as a whole. Let's review:

A "strong" copyleft license is a software license that requires all distributed derivative works of that software to be licensed under the same terms as the original license, which typically includes distribution of source code. E.g. GPL.

A "weak" copyleft license may allow works that are bundled with the original software to be distributed under a different license, as long as the original copyleft software remains unaltered and under the same license. E.G. BSD, MIT, Apache, LGPL (to a lesser extent).

Stallman has argued that unless we all use GPL for all of our libraries, the open source movement will be eaten by the commercial software industry. He is wrong for three sets of reasons:

1. In many circumstances, the GPL is fundamentally incompatible with business needs, and these business needs are simply not going away.
2. The free and open source movement has been shown to co-exist harmoniously with the proprietary, commercial software industry.
3. Strong copyleft licenses have been a powerful tool for the commercialization of proprietary commercial software, under a scheme of dual licensing, totally turning Stallman's vision for the GPL on its head.

(1) Business Needs
If you are distributing software1 that has trade secrets embedded in source code, clearly, a strong copyleft license would be inappropriate, as it would require public disclosure of trade secrets. This could be a trade secret ranging from anything from a financial hedging algorithm or controls of precision machinery, to graphics processing or even your secret fantasy football handicapping scheme. Laying your code bare would give all these secrets away, which, from a trade secret law perspective, would invalidate their standing as trade secrets. This is a big no-no.

Additionally, the terms of the Apple App Store make it very difficult to include GPL licensed software in apps. For instance, the App Store may distribute your software outside of the united states, and the GPL requires that you cannot restrict licenses for GPL licensed software, which includes geographic restrictions. So, if you are distributing software that has export restrictions, either due to technology, agreement, or privacy laws, you find yourself in a very tricky situation. It is navigable with clever engineering and proper lawyering, but it is an enormous headache.

(2) Harmonious Coexistence with Commercial Software
As stated in previous posts, the Ruby on Rails community basically lives off of the MIT, BSD and Ruby licenses, all of which are weak copyleft licenses. This is simply a fact that cannot be disputed. It may be the case that when Stallman first founded the GNU foundation, strong copyleft licenses were a necessity for the success of linux - given the nature of the atmosphere back then, with only a few large companies controlling the balance of commercially viable developers. However, as time has passed, the number of developers has grown tremendously, and they are not all controlled by a handful of old-world corporations. As a result, there are now many totally viable motivations for contributing to open source projects beyond mere legal compulsion to do so - the Apache foundation is an excellent example of this motivation. Quite simply, developers enjoy having access to tools with large user bases, the prestige and reputation of being an open source contributor, which may further a commercial career, and the sense of community that comes with being part of an open source project.

(3) Dual Licensing
Stallman has accepted that while Dual Licensing is legal, he is not a fan. In essence, Dual Licensing, (in certain circumstances called Single Vendor Commercial Open Source Business Model) is where a company may make their proprietary software available under the GPL and also under a commercial, proprietary license. MySQL is an excellent example of the success of this license. It can be thought of as a type of "freemium" model - as long as you are not distributing your software, you are free to use, study and modify GPL licensed software pretty much to your hearts content. This allows for academic use, and purely internal commercial use, e.g., a hedge fund can download MySQL and use it internally, modifying it as much as they want, without worrying about the copyleft provisions. However, if they want to license their hedge-fund approved version of MySQL, but don't want to release their entire codebase to the public, they need to pay Oracle for a commercial license. This is precisely what happened to MySQL, which has very successfully used dual licensing to create a substantial business.

In the end, I think that it is clear that strong copyleft software has a permanent place in commercial applications - but I also believe that at this point, the alternative motivations for contributing to open source software - beyond mere legal compulsion to do so - are more than sufficient to allow for a vibrant weak copyleft open source community to thrive.

However, if you are considering integrating GPL code into your proprietary, commercial software before it ships, I highly suggest you find a very competent lawyer.

Thursday, May 8, 2014

Revisiting AOL - Profits down 90% in Just Two Years

Last August I wrote this post:
Of the $541M in revenue, $361.2 comes from advertising, or almost precisely 2/3 of all revenue. Further, an entire $166M comes from subscriptions, which is codeword for dial up subscribers. That's right, a full 30.6% of AOL's quarterly revenue comes from people with dial up modems. So, AOL generated 97.6% of its Q2 income from advertising and dialup. That means that all of AOL's other products, besides advertising and dialup, account for less than 3% of its income. That is not a good sign.
Just to review, AOL defines "subscribers" in its 10K as:

As of December 31, 2013, we had approximately 2.5 million domestic AOL subscribers. Our subscribers are important users of Brand Group and Membership Group properties and engaging our subscribers, as well as former AOL subscribers who continue to utilize our free service plan, is an important component of our strategy. Our paid subscription plans provide bundles of products and services ranging from online storage, privacy and security solutions to technical support, back-up and unlimited dial-up internet access options, computer protection and partner discounts.

Today, Ars Technica has the headline: In one short year, AOL’s quarterly profits plunged 66 percent.

According to their 10K:

This decline has been the result of several factors, including the increased availability of high-speed internet broadband connections and attractive bundled offerings by such broadband providers, the fact that a significant amount of online content, products and services has been optimized for use with broadband internet connections and the effects of our strategic shift to focus on generating advertising revenues, which resulted in us essentially eliminating our marketing efforts for our subscription access services and the free availability of the vast majority of our content, products and services. Although we provide many additional products and services as part of our subscription plans, there can be no assurance that our subscribers will value the bundle of services we offer and they may cancel their subscriptions at any time. If any of these factors result in our Subscriber base declining faster than we currently anticipate, our subscription and advertising revenues could be adversely affected.
Here's the year-over-year:

Not looking so hot for AOL.

Wednesday, May 7, 2014

So it Begins: Startups Getting Hurt By Net Neutrality

But don't take my word for it.

Via MIT Technology Review, here is Brad Burnham of Union Square Ventures:

The cable industry says such charges are sensible, especially when a few large content providers like Netflix can take up a large fraction of traffic. But if deep-pocketed players can pay for a faster, more reliable service, then small startups face a crushing disadvantage, says Brad Burnham, managing partner at Union Square Ventures, a VC firm based in New York City. “This is absolutely part of our calculus now,” he says. 
Burnham says his firm will now “stay away from” startups working on video and media businesses. It will also avoid investing in payment systems or in mobile wallets, which require ultrafast transaction times to make sense. “This is a bad scene for innovation in those areas,” Burnham says of the FCC proposal.
Just a reminder to everyone, that recently, Mozilla proposed the following changes to the proposed FCC Net Neutrality Regs:

Mozilla's plan is a somewhat crafty attempt to avoid the worst of the political mess that reclassification would cause, by arguing that there are two separate markets: the markets for broadband providers to end users (i.e., our own broadband bills) and then a separate market for the relationship between internet companies (what Mozilla is calling "edge providers") and the broadband providers. Mozilla is saying that since these are separate markets, the FCC could reclassify just the connection between internet companies and broadband providers as telco services, and leave the last mile setup unchanged as an information service. Thus, it's arguing that the transit market more accurately reflects a telco service, and thus would be much easier to reclassify. In a sense, this would also be a way to attack the interconnection problem, which is where the net neutrality debate has effectively shifted. 
GigaOm comments on that same plan here, saying:

But can this work? It’s a neat way to call Wheeler’s bluff on the reclassification issue, which is so politically charged, that he truly can’t touch it. Instead of attacking the cable and telcos from the front on reclassification, he could sneak around from the side. However, Wheeler’s made statements in the past that indicate he’s okay with a double-sided market for broadband, which means he may not want to impose this new relationship on ISPs. 
Such an action would also undoubtedly lead to lawsuits if it were implemented, which throws net neutrality into doubt for even longer. However, it’s about time someone changed the terms of this debate to reflect how the internet has changed since 2002 when the FCC decided it wasn’t a utility. Since then, as people have abandoned ISP-specific email, portals and more to surf for content and choose services delivered from the wider internet, it’s clear that ISPs are a conduit for content and services, not a provider of them.
Mozilla seeks to get the FCC to recognize this in a way that might be politically viable. Hopefully the agency takes Mozilla up on the idea.

And, as a parting note, let's not forget that Tom Wheeler, the current head of the FCC, is a former Cable Industry Lobbyist. Or, to quote Consumerist on the matter: FCC Chairman: I’d Rather Give In To Verizon’s Definition Of Net Neutrality Than Fight. How, exactly, did Wheeler get this job again? Oh right. May have something to do with that 700k he raised for the Obama campaign. Smashing.