Well, rapid increases in the price of Research Agreements actually have an anti-runaway mechanic built in: if I'm an era or two ahead of everyone else, each of my RAs will cost a lot more than other Civs' will (esp since AI civs that are behind will demand Gold before signing one with me).
That said, yes, the late-game RA prices are probably a bit too high (a 50% reduction seems too big). Either way, they're definitely an improvement from pre-v131 prices though, in my opinion.
Can we turn this multiplier down then? Or maybe my experience was a one-off... has anyone else noticed a mid-game CS with very little territory to its name?
Re: the culture discussion above, I agree that culture victories still seem too fast; in my current game, India's got 28 policies filled out by turn 220. Policy costs could stand to be scaled up a bit more, at least for the later (the 20th and on, say) ones.
Also, I'm very glad to report that I'm not finding myself awash in Gold by turn 220ish in my current game! The AIs are still managing to rake it in (917 net gpt for Arabia and 353 for India by turn 220, somehow... even with reduced maint, I don't know how that's happening...), although my finances would be in better shape had I kept a few more puppets for gold purposes I razed a few medium-sized cities to keep my Happiness up (most of my Science production is in my Capital, so I've been trying to keep that bonus high).
What strategy are you using to get two eras ahead of the AIs? I only play on king and emperor and in my two games in 131.12 the AI has completely blown me away in research (industrial before I hit ren for example). In release 131 I as able to get ahead in tech.
I agree that the RA prices are an improvement, but only as an indirect buff to the AIs for use of their extra cash. For me the strategy hasn't changed: Before it was always take RA and now it's never take RA (I'd much rather buy buildings and CSes than ~1/2 - 3/4ths of a tech over 30 turns).
Remember that due to a bug in the game, finishers count as a policy, so 28 could mean 4 finished trees. That still seems a bit fast though.