Research agreements worth it?

Research agreement are just another reason to keep the weak civs alive.
In my domination games, I always keep the weak civ with 2-3 cities to milk their cash flow and RA(even if this means giving them money). if an AI kill another, I liberate a dead AI city. Resurect him to milk him
I wait till I fill most early tech so I don't crap ones. Afterwards, I usually have 4-5 RA running all game.

RA and GS are GREAT!

This whole 'randomness screws me over thing' seems a little silly to me. If you have a chance between getting rifling or sailing from your reasearch agreement you should either be backfilling your tech tree before the RA completes (older techs tend to only take 2 turns, which really isn't a big deal, even when you stack two or three of them turn after turn), or you shouldn't be signing RAs and then complaining when you get something rubbish

I don't really see RAs as a way of slingshotting tech any more than Tech trading in Civ 4 was about slingshotting. It's just a different way of backfilling your tech tree without losing access to current technologies. In Civ 4 you traded to backfill while using your research to get modern techs; in Civ 5 you sign an agreement, backfill for 30 turns and then, when your agreement pays off, those thirty turns haven't been wasted because you should get something decently modern no matter what you get. Congratulations, you've just filled in all the missing parts of your tech tree whilst still advancing.

It's not a Great Scientist, it can't be controlled to slingshot techs for you specifically for the reason that you'd become even more unbalanced in tech - use them to buy you time to fill in those bits of the tech tree you really should have but don't want to divert your research to and you'll start seeing more success, and a lot less randomness

I dont know if this has been brought up but keep in mind that. If you refuse a pact with an AI they will then seek out another AI to make a pact with. This can turn out very badly on higher difficulties. While I focused a lot on CS in my lower difficulty games, I find myself more inclined to grab a bunch of pacts with all the AIs to vastly catch up then out tech tech them while I focus on building an army so I don't look weak enough to invade.

It's that the randomness is bad for balance, not that it's weak. It's expected return is worth it, but the split between the best case and worst case can be big. Like the lottery can have an expected return larger than the $1 to play, but your actual result is either lose the dollar or win millions.

Many play with ruins off because their power is so varying and high. A research agreement is like a ruin you can hit every 30 turns.

My game last night I got a free slingshot from one, I had just bulbed education and 2 turns later my first research agreement gave acoustics. I think the other 3 options were fishing, iron working, or civil service.