3 ms·
In science and math, too often only the final result gets credited or published. Maybe we should start crediting and publishing our intermediate results and at
by niemandhier 18d ago
In science and math, too often only the final result gets credited or published.
Maybe we should start crediting and publishing our intermediate results and attempts.
Humans would gain credit for providing a part of the solution to tough problems, LLMs would profit from the resulting data flywheel.
- thaumasiotes 18d ago> In science and math, too often only the final result gets credited or published. > Maybe we should start crediting and publishing our intermediate results and attempts. All mathematical results are "intermediate results". What would it mean for something to be a final result?
- veunes 18d agoI think "final" here just means the unit that currently gets treated as publishable and credit-worthy: a theorem with a proof, a solved conjecture, a completed paper, etc
- jonathanstrange 18d agoIf it's not worth publishing, it's not a result at all. That being said, the problem really is that since LLMs excel at sifting through piles or papers, publishing more would aggravate the problem instead of alleviating it.
- mitxela 18d agoThen it's circular: "only the units that are publishable are published"
- veunes 18d ago[dead]