Processor of Natural Languages
@processorofnl.bsky.social
natural language processing
yeah yeah
November 14, 2025 at 6:00 PM
yeah yeah
Don't want to look too into it but... it doesn't really make sense to me why a researcher would be so desperate to publish a fraudulent paper, but it makes more sense to me why a college prep-focused high schooler and her father would want to.
November 14, 2025 at 5:58 PM
Don't want to look too into it but... it doesn't really make sense to me why a researcher would be so desperate to publish a fraudulent paper, but it makes more sense to me why a college prep-focused high schooler and her father would want to.
conferences are really awesome to attend, but the format feels outdated and unjustified at current scales. I feel the overwhelming positive reactions I see with COLM, a smaller and leaner conference with more focused interests, supports this idea.
November 14, 2025 at 4:20 PM
conferences are really awesome to attend, but the format feels outdated and unjustified at current scales. I feel the overwhelming positive reactions I see with COLM, a smaller and leaner conference with more focused interests, supports this idea.
incentives are incredibly misaligned but as a start we can stop distorting things even further by propagating beliefs like "top-tier confs==most rigorous peer review" or "you need X amount of publications to get a good position". good scientists probably know better, but more people should.
November 14, 2025 at 4:11 PM
incentives are incredibly misaligned but as a start we can stop distorting things even further by propagating beliefs like "top-tier confs==most rigorous peer review" or "you need X amount of publications to get a good position". good scientists probably know better, but more people should.
and it is also good(?) to see the frustrations surrounding peer review are dual-sided. 'reviewers are evil' or 'submissions are awful' discourse could be harmful in isolation... accepting flaws in the system as a whole is important, since we can only fix it as a whole (sadly can't use backprop)
November 14, 2025 at 4:01 PM
and it is also good(?) to see the frustrations surrounding peer review are dual-sided. 'reviewers are evil' or 'submissions are awful' discourse could be harmful in isolation... accepting flaws in the system as a whole is important, since we can only fix it as a whole (sadly can't use backprop)
on the other hand, being public has identified immoral submission/reviewing practices that I suspected of but couldn't confirm on a larger scale. they are now out for public discussion and (potential) remedying...
November 14, 2025 at 3:54 PM
on the other hand, being public has identified immoral submission/reviewing practices that I suspected of but couldn't confirm on a larger scale. they are now out for public discussion and (potential) remedying...
on public comments being allowed... guidelines to "keep things civil" is preventing complete collapse, but this cant apply to screenshots in other platforms. over time, redditification + pitchforking will emerge, which encourages onlookers to perform on self-justified outrage
November 14, 2025 at 3:52 PM
on public comments being allowed... guidelines to "keep things civil" is preventing complete collapse, but this cant apply to screenshots in other platforms. over time, redditification + pitchforking will emerge, which encourages onlookers to perform on self-justified outrage
it feels so unacceptable for people to screenshot the verbatim interactions and post it to their socials, where the poster's network will invariably come to the aid of the poster. what's worse is that they can also come into openreview and influence reviewers and area chairs.
November 14, 2025 at 3:48 PM
it feels so unacceptable for people to screenshot the verbatim interactions and post it to their socials, where the poster's network will invariably come to the aid of the poster. what's worse is that they can also come into openreview and influence reviewers and area chairs.
some points I thought about:
it is intellectually lazy and dishonest to characterize a side you don't like as AI generated. using ad hominem or ad machinam to avoid actual semantics is annoying.
it is intellectually lazy and dishonest to characterize a side you don't like as AI generated. using ad hominem or ad machinam to avoid actual semantics is annoying.
November 14, 2025 at 3:44 PM
some points I thought about:
it is intellectually lazy and dishonest to characterize a side you don't like as AI generated. using ad hominem or ad machinam to avoid actual semantics is annoying.
it is intellectually lazy and dishonest to characterize a side you don't like as AI generated. using ad hominem or ad machinam to avoid actual semantics is annoying.
btw, this in itself is a huge achievement and it is my mid-term goal: to be big enough that someone smart feels the need to clarify my work
October 31, 2025 at 3:03 PM
btw, this in itself is a huge achievement and it is my mid-term goal: to be big enough that someone smart feels the need to clarify my work
How does this paper from 2025 have 67 citations? Who cites this??
Only the most cutting edge frontier papers of course
Only the most cutting edge frontier papers of course
October 20, 2025 at 6:41 PM
How does this paper from 2025 have 67 citations? Who cites this??
Only the most cutting edge frontier papers of course
Only the most cutting edge frontier papers of course