Computer Science's conventions solve these problems (although CS certainly has other problems):
* papers are generally posted to arxiv.org immediately on being finished, so everyone can access them
* conferences have fixed deadlines and relatively short, fixed/enforced review cycles
