Skip to content

Open Source and the Future of Network Applications

Two weeks ago today, the folks from Joyent open sourced under the GPLv2 two of their core software assets: Connector and Slingshot. A nice announcement, certainly. What was really interesting, however, was why they did it.

No longer will those of us committed to open source need to use products from companies that do not open their source. This is Bastille day for software on the web. The cracks in Windows dominance on the desktop are visible. Let’s not repeat the mistake of building monopolistic overlords on the internet. We are not interested in making some proprietary stack of applications the Windows of the Web. We hope you agree…

Let’s not wait for G***le or Y***o or M*N or someone else to give us the web we want. We can do that now…

Connector and Slingshot. Better than free. Open.

First and most obviously, this is a call to arms. Join us, pleads Joyent, before we trade one dictatorship for another. Underlying the recruiting attempt, however, are a set of implicit assumptions worth extracting.

  1. Microsoft’s desktop dominance is threatened
  2. The primary source of the threat is free but non-open source SaaS offerings from Google, MSN, Yahoo
  3. The predicted outcome will see users forced to trade one dominant provider for another
  4. Open source is the last, best defense against that future

Speculative and reactionary though these comments may be, they are reasonable enough in my opinion to be warrant further debate. But not here, and not now.

Suffice it to say, for time being, that the Joyent folks are not the only ones concerned by the prospect of future technology landscape dominated by the likes of Amazon, Google, eBay, Yahoo, et al. As evidenced by developments like Joyent’s decision and the GNOME Online Desktop efforts, it’s increasingly apparent that open source and Web 2.0 are on a collision course.

While these two dominant technical trends or directions have much to learn from each other, the convergence is likely to have its painful moments if OSCON is any indication. Indeed the talk of the conference was the somewhat shocking public swipe at Tim O’Reilly by one of the GPLv3’s chief architects, Eben Moglen. As documented elsewhere, Moglen absolutely dropped the hammer on Mr. Web 2.0, arguing that “that the FSF has ‘done the heavy lifting’ and ‘carried your water’ for the last decade, and that the era of Web 2.0 distraction (buzz about who is making money, who will get acquired, etc) will need to be replaced by a serious conversation about freedom.”

Whether you agree with Moglen’s tactic here – I do not – it cannot be debated that the questions raised are quite legitimate. If software is increasingly if not exclusively transitioning to network delivered services, what should software freedoms – whether they’re defined by Apache or the FSF – mean in that context? Especially considering questions of data ownership and availability.

As nearly as I can determine, no one has many answers here yet – though smart people are beginning to ask the right questions.

A couple of my own random thoughts and observations:

  • Although much education remains, traditional software firms and practitioners have learned a great deal in the past decade or so about the implications of free and open source software. I suspect that SaaS/Web 2.0/pick-your-term-of-choice has a similar learning curve in front of it, and it is likely to be similarly painful (just ask Google).
  • Like many others in the open source world, Moglen appears to be believe that P2P is a legitimate (and less threatening) architectural alternative to a redshift future, observing during his talk that the PCs in the room offered – in aggregate – supercomputing power. While many people I know and respect are convinced of this, I’m skeptical. With the exception of examples like Skype, which does leverage users’ PC’s (though few of its users know that, I suspect) for processing, scale out seems to have cemented its role as the most economical model for large scale application delivery. I suspect, therefore, that P2P advocates are underestimatng the cost and complexity of assembling P2P based alternatives to the most popular web based applications we use today. Nor do I think some of the drivers of the mainframe to client server transition – scarcity, for one – will be as common as they were decades ago.
  • Moglen and his former organization the FSF have long considered themselves advocates of users’ rights rather than developers’ rights. The reciprocal nature of the GPL, for example, is explicitly designed to prevent developers from removing freedoms – as defined by the FSF – from the software subsequent to its distribution. Unlike, say, BSD licensed code which gives developer near absolute freedom for consumption and usage. Despite this, or perhaps because of it, the GPL has proven phenomenally popular amongst the developer populations generally (e.g. Sourceforge) and active projects specifically (Linux, MySQL). I remain unconvinced, however, that the SaaS world will be similarly concerned with the freedoms protected by the GPL (assuming, of course, that the developers that employ the GPL are indeed concerned).

    It was easy to prefer free and open source licensed MySQL, as an example, over expensive and proprietary alternatives such as IBM and Oracle. It will be less easy to prefer, I believe, a free and open Google over a free and slightly less open Google; the subtlety is likely to be appreciated only by a minority.

  • The ongoing tension between open source licenses and the SaaS applications it cannot control seems to be abating somewhat. It’s not resolved, by any means – there are still some those who are livid that Google can heavily leverage open source software and not share the resulting derivative works back to the communities they come from. But with the adoption of the GPLv3 by SugarCRM, the OSI approval of the Common Public Attribution License, and the in flight Affero GPLv3, it seems as if compromise – once assumed to be impossible – has become less so. Opinions may differ on whether that’s good or bad, but I’m generally a believer that compromise is a friend to progress.

What, if anything, you conclude from the above is largely dependent on your perspective. With so much as yet undecided, opportunities for debate and disagreement are numerous. What cannot be disputed, however, is the fact that the two most powerful trends in software – open source and software-as-a-service – are converging quickly, and the ripples from the resulting collision will impact all of us.

Categories: Network, Open Source, Open Standards.

Comment Feed

4 Responses

Some HTML is OK

or, reply to this post via trackback.

Continuing the Discussion

  1. […] Stephen O’grady points out, you may not agree with the tactics Eben used (I also don’t), but the conversation is an […]

  2. Web 2.0 vs Open Source?…

    Stephen O’Grady, one of my must-reads, has very insightful commentary coming out of OSCON ’07 (O’Reilly’s open source conference.) With respect to Web 2.0 vs. open source:……

  3. […] esteemed Sr. O’Grady writes about open source and the network (related Asay note), noting among many other things that Joyent is an option for those who want […]

  4. […] But none of that is what I’m interested in writing about – apologies for the digression. What I personally found fascinating about the 0.5 release was the way it continues the blurring of the lines between network and client application that I’ve been expecting for some time. […]