I'm asking this as a follow-up to http://lesswrong.com/lw/d5y/why_academic_papers_are_a_terrible_discussion/, which was written a few years ago, and which I find very interesting.
Many of the arguments advanced in http://lesswrong.com/lw/d5y/why_academic_papers_are_a_terrible_discussion/ (especially inaccessibility) could just as well apply to conferences, too.
I'd also wonder - would you consider conferences to also be a terrible discussion forum? What do you think would be some good alternatives?
The audience for conferences is limited, and people seem to remember only a tiny tiny fraction of everything they've encountered in a conference. The ideas in conferences don't seem to do much for building up platforms of public discussions around the new subjects that are often announced in conferences (rather than, say, on online platforms).
I suppose one could advance the argument that ideas often get brought up/discussed at conferences that wouldn't be conveniently discussed in any other medium (for now..). But is this mostly because people are too comfortable with what they're been brought up with?
My experience with conferences (in machine learning):
1) It seems almost universally true that the first few conferences one attends (especially the first one) are incredibly high marginal value, for the reasons that Mars_Colony notes.
2) After that, the marginal value seems to decrease somewhat. I attend 1-2 conferences per year and still get value out of them, but perhaps not enough to justify going except that it's a great venue in which to spread one's ideas to a larger audience. On the other hand, if I only went once per year (or per 2 years) then I would probably get more value per conference due to networking (in the good way) with colleagues who I had not seen in a long time.
3) Workshops at conferences can also be very valuable; they often act as a venue for a budding subfield to try to figure out its directions/priorities, or for a budding not-yet-subfield to coalesce into existence.
4) It is not my experience that conferences are a paid vacation. The last 4 ICML locations were: Edinburgh, Atlanta, Beijing, Paris. This seems consistent with what you would choose if you wanted to rotate through Europe, North America, and Asia while picking centralized locations near major airports. The last 4 NIPS locations were: Lake Tahoe, Lake Tahoe, Montreal, Montreal. This is admittedly at least partially optimized for skiing, but I also think it's pretty hard to call these places boondoggles. Yes, I certainly enjoy being able to go to Beijing for free, though I don't really understand how this is a strong criticism (note that many companies also pay for their employees to travel places as well).
5) I agree with some others in this thread that it's annoying for people to randomly opine on how terrible academia is without being informed. (For some reason, it seems in vogue on LessWrong to be anti-academic; I don't really understand what causes this?) I agree that academia has tons of problems (much like every other human endeavor) but I don't think that uninformed speculation will do much to solve them.
I think it's founder effect.