Conferences now became quite close-minded and elitist: the same people, year after year, polished presentations, barely any input at all. There are a few large ones every year, but they are expensive and quite limited number of people are allowed in. Which raises the question: does anyone know a platform or a website where one could post unpolished results and discuss them with other in hopes of feedback, corrections or even collaboration?
It's good, but I'm looking for something more forum-like.
I think you need to reconsider your view and expectations on the scientific discourse machine. There is no better quick turn-around forum for discussing results/ideas than conferences and journals. It just works slowly:
1. publish an idea
1. no reaction --> probably garbage
1. paper citing you --> some reaction, discussion starts
1. rinse, repeat
It takes years to discuss ideas in this manner. And while you address your message to somebody, rarely you get a reaction from the same side, rather from elsewhere (if it works of course).
I came to believe that this is the best way in the end. Once you prolong your time-horizons to decades, a discourse evolving on time-scales of months and years starts to make sense.
It works reasonably well even despite the gatekeepers in the form of publishers. But the feedback loop length has become the limiting factor in the process. And some areas have lower entry barrier these days (like ML for instance), but journals and professional conferences still have high entry requirements.
Well, AI/ML and generally CompSci research works quite differently to other areas. In CS the primary discussion forum are conferences. It works like this:
1. you have an idea, you float it at a specific workshop (organised on fringes of large conferences) - space is typically 15 pages of text
2. once the idea is OK, you flesh out important fragments of it out, work out experiments and publish at (gradually) large conferences (typically up to 6 pages each)
3. finalise the paper for conference proceedings - the main track record of the discourse in the field
4. only after years of doing this, you summarise the approach in a journal paper (virtually unlimited space, it's not unusual to review 100pg journal manuscripts)
In other natural sciences, mostly physics et al., the process is the other way round: you work out journal papers and conferences are then only social and networking events where you speak about the stuff ex-post.
This CS "anomaly" indeed leads to a higher turn-around, but is still slow and buggy because major conferences are only annual, or even bi-annual events (IJCAI, ECAI, AAAI, AAMAS, etc.). There were interesting experiments such as VLDB, which (if I am not mistaken) first published papers in a rolling release of proceedings and only afterwards the conference was a selection of papers.
But afaik whoever experimented with commenting system around e.g., arxiv, it turned out to be a futile exercise. People simply do not participate as much as would be necessary for it to work.
