Peer Review is still BROKEN! The NeurIPS 2021 Review Experiment (Video)

A look at the results of the 2021 NeurIPS peer review experiment.

https://www.reddit.com/r/MachineLearn

@ykilcher possibly arxiv needs its own reviewing system where anyone from the community kinda have a real vote on the final decision.

or as DAF, David Forsyth, said in ICCV 2021 “devalue the currency.”

Currency?
the peer-review stamp from top-tier venues.

How?
Back in 2013, a seasoned Pythonist venturing in computer vision in Colombia told me: “that conference model is a bit weird.” Why don’t they do as PyCon?

  • Satellite events. You can find at the granularity that you want from continent to country to city’s MeetUp events.

  • The most interesting stuff gets invited or proposed by others to a handful of global events.

Is it another instance of academics re-inventing the wheel as opposed to learning from other communities.?