Support period for unsuccessful games is wayyyy shorter than support for unsuccessful or buggy first release of open source softwares that can benefit from bug fixes and patches in the long run. Games are not in the long run.
Try to get Quake 3 support today. Then try to get support for ioquake3.
Alternatively, try to get support for Transport Tycoon. Now try to get support for OpenTTD.
It's not a matter of "game or not game". We have these examples of formerly commercial games for which commercial support no longer exists, but for which open source community support is plentiful.
I am of the opinion that Quake3 is an exception, not the rule.
So if the game is successful but doesn't work out of the box players will not bother and simply run the Windows version while the Linux gang will lament how it's not yet the year of Linux gaming. If the game isn't successful the Linux version will die out the same but faster.
Thinking an AAA game that missed its launch might benefit `from the community bug reports` is wishful thinking.
edit: And it's not due to the technological aspects or the software philosophy of the platform, it's about having the money and the time to support a product for a niche platform.