Hacker News Reader

Basically, reputation.

Arxiv isn't peer reviewed and thus you get a lot of crap on there (just have a look for proofs of goldbach's conjecture). Because the barrier to entry is low you dont get any kudos for publishing there, and the performance of academics in many institutions is essentially based on the kudos they get for publishing in big journals.

View on HN

Fortunately, many fields also have fully open-access journals and conferences which provide the prestige without making your papers hard to access. (In my field, e.g.: the ICDT conference, IJCAI, ICALP, LICS; the LMCS and JAIR journals.)

The reason why these haven't completely taken over is because prestige has a lot of inertia, and publisher-run titles retain their prestige (and still attract quality submissions and reviewing). But it's not true that you have to publish with closed-access publishers to get reviews and prestige.

Thanks for clarifying. I guess ML community is different though, even conference cannot keep up the pace of development, journal would be way too slow. But that might not be case for other domains.

Reply on news.ycombinator.com