This mostly seems to be a bitter rant about the time Facebook tricked TechCrunch by rolling out a new feature that only TechCrunch staff could see.<p>See the results of the prank here: <a href="http://techcrunch.com/2009/09/10/facebook-now-lets-you-fax-your-photos-i-have-no-idea-why-anyone-would-want-to-do-this/" rel="nofollow">http://techcrunch.com/2009/09/10/facebook-now-lets-you-fax-y...</a>
Uh, yeah. Is it really that unimaginable that they want to A/B test (or beta test) the features before they roll them out to everyone?<p>And keeping the media away the new features is just common sense. They're going to publish a huge story about a new change... Once that isn't complete and isn't ready for everyone. That just can't end well.
Being able to turn on and off features based on login or randomness seems like a pretty common feature when you get big enough for it to be meaningful to do tests on portions of your userbase, or have multiple tests running at once.<p>Disqus have open-sourced a similar system for Django-based sites, Gargoyle: <a href="http://code.disqus.com/code/projects/gargoyle.html" rel="nofollow">http://code.disqus.com/code/projects/gargoyle.html</a><p>You basically stick if(abc-is-active)s throughout your code, and then it has a fancy frontend for defining when abc should be active, based on user, ip, randomness, etc. They showed it off in their PyCon 2011 presentation, at about the 10:00 mark: <a href="http://python.mirocommunity.org/video/4256/pycon-2011-disqus-serving-400-" rel="nofollow">http://python.mirocommunity.org/video/4256/pycon-2011-disqus...</a>
Big whoop. All the websites that deal with similar scale have a six-month (or longer) pipeline, and almost all of them test code "live" in one way or another.
Which means Facebook is not pressed by competitors as of now. They hold onto a bunch of cool features ready to swith on, just in case competition strike. Any competitor will face doubly uphill battle -- both against the network effect and against a fresh reserve of new features, enabled as soon as the Facebook notices them.<p>My take-away from the story -- little competition results in slow progress.
I believe that Google uses a similar development methodology with Chrome to separate the channels out. Everyone works on the Chromium trunk, and features are enabled/disabled depending on which channel they are appropriate for. Single trunk development makes a lot of sense when you make a product which doesn't rely on distinct version releases.
I can't help but ask "so what?" I know the media likes to fascinate themselves with facebook, but adding websites does not really make the bog standard concept of not-yet-released products news.<p>After wondering why they're reporting on this, I assumed that they might be drawing some conclusions or commenting on what this actually means. They have 6 months of code in the pipe, are they still testing? can't handle the volume of change? think their users can't handle the volume of changes? or just are happy enough in the market that they're comfortable sitting on a few things that are ready to go live? Anything?<p>Nope. The story came across with the voice of a schoolgirl giggling "he he he, facebook mentioned us." With trite reporting like this, who can be surprised they've implemented the level of control to blind techcrunch at will?