I'm not sure about this. Obviously there's a cost in the actual port of the game, but I suspect there's a lot of work besides that to support any new platform.
I made some amateur games in python and they're trivial to run in Windows and Mac (being Linux my dev platform), but there are lots of small details that can ruin the user experience just because I didn't do proper testing in all the platforms (like some Windows OpenGL implementations not supporting shared contexts, or openal glitches in Mac).
My point is that is not just having a port, you have to support it, and that costs time and money that may not be worth the effort given the number of potential users of that new platform.
Obviously there is an additional cost; I think the point the parent was making is that the cost to add your third platform is lower than the cost to add your second platform, especially since one of the hardest bits (changing graphics stack) is mostly shared work between Mac and Linux.
The cost is still non-zero, of course, and for the most part Linux gamers have to live with a lower degree of polish than Windows or even OSX gamers because of that.
I made some amateur games in python and they're trivial to run in Windows and Mac (being Linux my dev platform), but there are lots of small details that can ruin the user experience just because I didn't do proper testing in all the platforms (like some Windows OpenGL implementations not supporting shared contexts, or openal glitches in Mac).
My point is that is not just having a port, you have to support it, and that costs time and money that may not be worth the effort given the number of potential users of that new platform.
EDIT: sent the comment too early