3 ms·
I wonder how hard it would be to make a blind test of comment quality by year, like a hot or not of comments... Other than setting it up, my questions would be
by corin_ 12y ago
I wonder how hard it would be to make a blind test of comment quality by year, like a hot or not of comments...
Other than setting it up, my questions would be a.) How to filter out comments that give away when they were posted ("wow, didn't expect them to announce a smaller ipad!") b.) How to deal with comments that are part of a chain (and therefore their relevance may not be visible on their own) c.) How many comments would one person need to rate to get a statistically sound reading of what period in HN's history they think is better
- rdl 12y agoI think the easiest test (at least for my "1% metric") is reading HN comments on articles after they're populated (a day?), and then deciding "was there anything in there which was amazingly good, such that I'd actively seek it out in the future?" The disappearance of grellas is a major factor against this recently.
- sillysaurus3 12y agoGrellas seems just as active as he usually is: https://news.ycombinator.com/threads?id=grellas https://news.ycombinator.com/threads?id=grellas 50+ avg comment karma can't happen weekly. Substantive comments require waiting for substantive stories, which are relatively rare.
- corin_ 12y agoAsking yourself that question allows you to decide whether the current quality is "<subjective scale from awful to amazing>, but it doesn't help compare now vs. previous though, there are far too many cognitive biases to objectively remember and compare