[ffn]

I don't know about you, but in the field of selecting new frameworks I tend to essentially just (mindless) follow celebrities. Some people call me a sheep, but so far, I've found following wycats (rails, emberjs) and his cronies around has paid off tremendously in terms of framework stability for frontend javascript.

On the other hand, the worst way to go about selecting a framework is, in my opinion, benchmark testing (or looking at some blogger's benchmark tests). Your framework might run a million hello world programs faster than mine as of last month, but you have no idea if the people working on the framework will even be around in the next 6 months when your MVP is actually complete. And if you wind up having to solo-support your own dead framework while concurrently trying to run a small business, good luck finding time to build new features, hiring qualified programmers, and not coming around to hate javascript.

[andrewmcwatters]

There's a lot of validity to this, and it's quite wise advice, really.