3 ms·

> The real answer is, the metrics are flawed. Not sure what you're saying here. The metrics of karma, whether you like it or not, is what Hacker News is based

> The real answer is, the metrics are flawed. Not sure what you're saying here. The metrics of karma, whether you like it or not, is what Hacker News is based off. It would be hard to say that we are discriminating against posts with less karma by not letting them be on the front page. However, the other metric system of the account's age is unrelated for the most part and therefore should not be a factor in most things. > One of the more interesting ironies is the people claiming the quality is declining due to new users are actually new users I'm not sure if this is entirely true. PG is an exception, but he doesn't speak for all of the early-adopter users.

jcr 16y ago
What I meant by "the metrics are flawed" is, how you personally evaluate the reputation of another user is a personal choice based on very limited and potentially deceptive input. Additionally, how the system itself evaluates reputation of a particular user, article or post is also imperfect. Only part of the algorithms for evaluating reputation are revealed in the HN source code [1] but other parts are not public. PG has stated HN runs with some of his own personal, unpublished modifications for the sake of preventing people from gaming an imperfect system. The up/down vote and 'flagging' feedback from users is important input, and does help the system more fairly evaluate things. Since the vote/flag input is from human beings basing their decisions on incomplete information and personal preferences, you can see how the system will remain imperfect. A good example is below I cracked a joke about divulging PG's secret to success being talking to himself in a soundproof closet... --the downvote I got tells me someone didn't find it funny. Such is life. On the bright side, if you can think of a way to avoid the imperfections in a reputation evaluation system, you could make millions, and you would certainly have PG's attention. It's a painfully difficult problem to crack, potentially undecidable, so even a better imperfect solution would be extremely valuable. [1] http://www.arclanguage.org/ http://www.arclanguage.org/

jcr 16y ago

What I meant by "the metrics are flawed" is, how you personally evaluate the reputation of another user is a personal choice based on very limited and potentially deceptive input. Additionally, how the system itself evaluates reputation of a particular user, article or post is also imperfect. Only part of the algorithms for evaluating reputation are revealed in the HN source code [1] but other parts are not public. PG has stated HN runs with some of his own personal, unpublished modifications for the sake of preventing people from gaming an imperfect system. The up/down vote and 'flagging' feedback from users is important input, and does help the system more fairly evaluate things. Since the vote/flag input is from human beings basing their decisions on incomplete information and personal preferences, you can see how the system will remain imperfect. A good example is below I cracked a joke about divulging PG's secret to success being talking to himself in a soundproof closet... --the downvote I got tells me someone didn't find it funny. Such is life. On the bright side, if you can think of a way to avoid the imperfections in a reputation evaluation system, you could make millions, and you would certainly have PG's attention. It's a painfully difficult problem to crack, potentially undecidable, so even a better imperfect solution would be extremely valuable. [1] http://www.arclanguage.org/ http://www.arclanguage.org/