GPLv3 adoption rates: Do they matter?

Since its release this June, GPLv3 certainly seems to have taken on the role of baby Damien within the open source licensing world. Sure, it's cute and cuddly at first glance, but behind its deep blue eyes lurk pure and unhinged evil ... at least if you believe the recent Evans Data survey (login required). With 380 developers who were working on open source software polled, only 6 percent of whom have adopted G

Then again, Palamida's posted conversion rates (for projects converting fully to GPLv3) seem to indicate that the Evans Data results may be a bit too pessimistic.

So who do we believe: Palamida or Evans Data? I think it's too early to bash either set of numbers. Obviously if Palamida continues to compile conversion numbers, we'll be able to more accurately compare notes between actual conversions and the Evans Data estimated adoption rate, which calls for 66 percent of the populace not adopting it within the next year and 43 percent staying fully away from GPLv3. But will that tell us whether GPLv3 is a flop?

Maybe the best way to judge the long-term success or failure of GPLv3 is by simply watching a number of key projects. As with major economics indicators, such as the Consumer Price Index, perhaps it takes only a few falling trees to eventually clear a forest. Here's a quick snapshot of the GPLv3 forest:

  • FreeBSD: No, "why would we?"
  • GlassFish and other Sun projects: Maybe for some parts
  • Apache: No need, just tweak for compliance
  • Linux Kernel: Uh, no no no no no no
  • SugarCRM: Oh yes, done deal
  • GNU OS, Emacs, gcc, Tar, Paint, etc.: Of course, already converted
  • Samba: Converted
  • MySQL: Changing license to navigate around GPLv3
  • SUSE: Yes but no ... who knows at the moment?

From this vantage point, it seems that with the exception of Samba, the majority of the infrastructure open source projects are staying partially or completely away from GPLv3, though some of the point solution products (such as SugarCRM) are happy to move the switch. In looking a little deeper into the forest of solutions catalogued by Palamida, it appears that the majority of GPLv3 solutions out there are geared toward Linux enthusiasts, developers, or the hackers/IT administrators. Here's a sampling:

  • Get Sudoku (number 9, number 9, number 9)
  • RadioDog (audio streaming software)
  • RxEvil (Ruby clone of the linux game, XEvil)
  • Mac-Ro-Knife (CLI for remote hosts)
  • LanShark (file sharing tool)
  • Hundiyas (a game like Battleship)

Mixed in with these tiny solutions, of course, there are some meatier solutions:

  • mwForum (forum software)
  • OpenBAM (business activity monitoring)
  • Hamster (ERP System)
  • uCommon (telephony libraries for embedded Liux)

Still, as with the high-level view, these products align with small communities or company-driven communities that would benefit from some of the protections offered by GPLv3. With that in mind, I think these numbers have little to do with the success or failure of GPLv3. Just as communities seeking wide distribution through ISVs have rallied around more permissive licenses such as Apache, I think we'll see companies that need a more commerce-friendly license such as GPL follow SugarCRM and Samba's lead. Either way, the numbers (Evans Data or Palamida's) don't matter in terms of winners and losers with open source licensing. We inhabit a multiple-license universe, and we should be glad for it because we can choose accordingly. A license, like an open source project itself, is nothing more than a specific means to a specific end -- even if that means looks a little creepy with blue eyes and black hair.


Copyright © 2007 IDG Communications, Inc.