For something between 2 and 4, the best I can come up with would be textbooks or seminars, both being extremely spotty in terms of quality and understandability.
In any case, a big problem you get is a cliff of information content going from 4 down to whatever the next step is. The incentive structure substantially motivates putting out new material, which must have some novel concepts. The focus on novelty and accomplishment leads to quite a mess. People put out half-baked work to be the first to write on a particular subject, which gets citations, which means the next round, also half-baked, is built on a half-baked foundation. When what's most needed in almost all cases is to parse the last generation of literature into something coherent, real, and replicable.
I suppose OP meant "review" or "survey" articles.If that was the case I totally agree.
For the other poster one nice source is of course The Annual Review journals. Arxiv of course too. The bibliographies in undergraduate/beginning graduate textbooks or syllabi are good sources too.
For something between 2 and 4, the best I can come up with would be textbooks or seminars, both being extremely spotty in terms of quality and understandability.
In any case, a big problem you get is a cliff of information content going from 4 down to whatever the next step is. The incentive structure substantially motivates putting out new material, which must have some novel concepts. The focus on novelty and accomplishment leads to quite a mess. People put out half-baked work to be the first to write on a particular subject, which gets citations, which means the next round, also half-baked, is built on a half-baked foundation. When what's most needed in almost all cases is to parse the last generation of literature into something coherent, real, and replicable.