4 ms·

The claim seems factually accurate and supported. We wouldn't consider an expert saying "you don't understand transformer ML models" with quotes supporting that

The claim seems factually accurate and supported. We wouldn't consider an expert saying "you don't understand transformer ML models" with quotes supporting that claim, as being putting down the user, no? I often read technical conversations on HN in which one user tells another they don't have an accurate understanding of some topic. Just because it's a historical or cultural matter doesn't mean it's ad hominem to state plainly and dispassionately that someone lacks understanding.

dang 5y ago
It's just the same when the topic is ML models. HN readers shouldn't be making pejorative comments about each other in those conversations either, and when we see it we moderate it the same way. There's no need to comment pejoratively on the other person. The post is stronger without it, and the expected value of such swipes, in terms of the impact they have on subsequent discussion, is sharply negative.
- jancsika 5y ago
  > HN readers shouldn't be making pejorative comments about each other in those conversations either, and when we see it we moderate it the same way. Moderation appears highly selective. E.g., "This argument is clownish and we should be embarrassed it's on the front page."[1] That is the top comment for that post, but I don't notice your reminder there about pejorative comments. If I had to rankly speculate, I'd say moderation happens consistently on topics where the average HN user tends to have a low level of knowledge (unionization, civil rights history), and less consistently for topics where there is a greater probability of a critical mass of participants with domain expertise (digital security). I also rankly speculate that you don't state this outright to prevent the tinder fire that would result from making users aware of their own lack of knowledge on the topics more likely to be moderated.
  - pvg 5y ago
    Moderators can't see everything and HN doesn't operate under some 'no negative comments, ever, at all' ruleset. It's one thing to point out, in some detail, that the argument in a particular article is very poor, it's a completely different thing to directly attack another user. Just like anyone is free to write clownish, provocative invective on their blog, people can point out its nature when it pops up on HN - especially when the main reason it does is its clownish provocativeness in the first place. You could always flag and/or email the moderators if you think something needs a further look, though.
  - dang 5y ago
    Moderation always appears highly selective no matter what - this is a psychological inevitability (https://hn.algolia.com/?dateRange=all&page=0&prefix=true&sort=byDate&type=comment&query=%22always%20feels%22%20mods%20by:dang https://hn.algolia.com/?dateRange=all&page=0&prefix=true&sor...). The truth is far more boring than your speculation, which I suppose is a consistent problem with speculation. We moderate guideline breakage when we see it. Moderation is inconsistent because we don't see everything (as pvg already pointed out). https://hn.algolia.com/?dateRange=all&page=0&prefix=true&sort=byDate&type=comment&query=likeliest%20by:dang https://hn.algolia.com/?dateRange=all&page=0&prefix=true&sor... There's also some selection involved, but it's more along the lines of "do I have the energy to make a post about this one", "am I prepared to deal with howls of protest in this case", and "is it borderline enough to just let go".
TheCoelacanth 5y ago
Saying "you don't understand transformer ML models" is also not a productive form of discussion. Discuss what is wrong with what they are saying. You don't need to make inferences about what personal failings might be leading them to make an incorrect statement.

dang 5y ago

It's just the same when the topic is ML models. HN readers shouldn't be making pejorative comments about each other in those conversations either, and when we see it we moderate it the same way. There's no need to comment pejoratively on the other person. The post is stronger without it, and the expected value of such swipes, in terms of the impact they have on subsequent discussion, is sharply negative.

jancsika 5y ago
> HN readers shouldn't be making pejorative comments about each other in those conversations either, and when we see it we moderate it the same way. Moderation appears highly selective. E.g., "This argument is clownish and we should be embarrassed it's on the front page."[1] That is the top comment for that post, but I don't notice your reminder there about pejorative comments. If I had to rankly speculate, I'd say moderation happens consistently on topics where the average HN user tends to have a low level of knowledge (unionization, civil rights history), and less consistently for topics where there is a greater probability of a critical mass of participants with domain expertise (digital security). I also rankly speculate that you don't state this outright to prevent the tinder fire that would result from making users aware of their own lack of knowledge on the topics more likely to be moderated.
- pvg 5y ago
  Moderators can't see everything and HN doesn't operate under some 'no negative comments, ever, at all' ruleset. It's one thing to point out, in some detail, that the argument in a particular article is very poor, it's a completely different thing to directly attack another user. Just like anyone is free to write clownish, provocative invective on their blog, people can point out its nature when it pops up on HN - especially when the main reason it does is its clownish provocativeness in the first place. You could always flag and/or email the moderators if you think something needs a further look, though.
- dang 5y ago
  Moderation always appears highly selective no matter what - this is a psychological inevitability (https://hn.algolia.com/?dateRange=all&page=0&prefix=true&sort=byDate&type=comment&query=%22always%20feels%22%20mods%20by:dang https://hn.algolia.com/?dateRange=all&page=0&prefix=true&sor...). The truth is far more boring than your speculation, which I suppose is a consistent problem with speculation. We moderate guideline breakage when we see it. Moderation is inconsistent because we don't see everything (as pvg already pointed out). https://hn.algolia.com/?dateRange=all&page=0&prefix=true&sort=byDate&type=comment&query=likeliest%20by:dang https://hn.algolia.com/?dateRange=all&page=0&prefix=true&sor... There's also some selection involved, but it's more along the lines of "do I have the energy to make a post about this one", "am I prepared to deal with howls of protest in this case", and "is it borderline enough to just let go".

TheCoelacanth 5y ago

Saying "you don't understand transformer ML models" is also not a productive form of discussion. Discuss what is wrong with what they are saying. You don't need to make inferences about what personal failings might be leading them to make an incorrect statement.