Basically the thought process is thus:
1) Should we have matchmaking?
2) Well, it'd have to be pretty robust, are there any good off-the-shelf ones we can use?
a) Steam? Then it'd be Steamworks in every build, which we know closes other distribution channels.
b) GameSpy? ... ugh. We looked, we really did, but no, not a good thing.
c) Impulse? Didn't have the features we needed; looked like they were planned but doesn't seem to have ever happened.
3) Hmm, so what about building our own?
a) We did that with Tidalis, put quite a bit of effort into it, and all it got us was a few favorable comments in the rather stellar reviews... and the game did very poorly commercially. Despite selling a fairly large load of copies (in a really steep discount in one of the special bundles on Steam during the holidays) the matchmaking service is next-to-unused.
4) So what's the gain if we did do it?
a) Occasionally someone fusses at us for not having it. If we did it then that would stop and be replaced by people occasionally fussing at us for the specific way we did it.
b) It might help players find games easier.
5) Ok, how many players, how much would it help them?
a) Judging by feedback and the mantis votes and whatnot... not very much at all.
6) Ok, sounds like a cool thing to do sometime, particularly since we worked out a lot of the basics in the system we built for Tidalis... but definitely not a priority.
And no new information (facts, angles of thinking about it, etc) has been presented in this thread. In fact, some of the claims and assertions being put forth are manifestly false (AIW not doing well or doing worse than before in revenue, etc). Anyway, no changes to the inputs, therefore no changes to the premises of that thought process, and therefore no changes to the conclusion.
If the players really want this more than any other new feature and that is objectively demonstrated (rather than "proof by emphatic assertion", which is good for comedy value but not decision-support)... we'll do it. We've got the basics of the framework and the hardware to back it, but it's just not a priority by any of the various measurements we use (personal interest, player interest, etc) and frankly we don't think it would really help all that much.