FOSS Gaming Needs Distributed Content Development

Posted: 2009-04-16

Looks like FOSS gaming needs more than just promotion and financial support. After my last post about it, i recieved a very insightful email from Tim Blokdijk since my comment form was broken, and i thought it would be worth sharing with all of you. Seems the biggest barrier to FOSS gaming development isn't code, but content.

I can't comment on your blog but I still feel a need to respond to your “Linux needs games, games need you” posting.
Up to about a year ago I was quite involved with the spring project, one of the more successful open source “games” out there.
To me the nr 1. challenge/problem is that “we” the open source people more or less cracked the problem of distributed code development. It's all text based and a lot of focus and effort has been put into that “way” of working. But it's different for content, as you cannot “diff” a picture or a blender file. Distributed content development is a pain in the ass. Spring has loads of content but it's very difficult to get a distributed team to work on integrating all this into one style that can compete with commercial games.
There is progress, the creative commons licenses makes it easier to do things as the GPL and other traditional licenses are to much focused on code. Broadband internet makes it easier to send around large binary source files. The 3D drivers for Linux are improving.. Gimp actually starts to get useful so artists are more open to the idea to look beyond Photoshop for a second. Wings3D is stable enough and useful enough to get people modelling. And Ton is finally committing himself to a gui rewrite for Blender. All these things used to be big blockers but it's still not practical to start a distributed content development project.
The only real successful open source content projects are “Elephants Dream”, “Big Bug Bunny” and “Yo Franky!”. But that involved getting 10+ people together for 6 months to work on a project. And even then they had problems managing the content files. They were using svn but had to use a small webapp to keep track who was editing what files.
A service like launchpad would need “a lot” of work to make it practical to develop content in the same way as it's possible to develop code. Your constantly running into/pushing the envelope. Dpkg repository's are not designed to cope with "source" files of many gigabytes. Executable files that need to be binary identical (or very close to) to allow people to play multiplayer games (across win/lin/mac platforms). And there are more issues like preventing cheating.. and stuff.
I personally think that you really need a company as canonical purely dedicated to game and content development to get to a level where you can speak of “reaching a tipping point”. The Blender institute is a contender but they lack a large amount of hard cash to accelerate the field.
So there you have it. Can the Blender Institute take on the challenge? Will Canonical step up to the plate?