3 ms·

I already replied, but I think this warrants a second reply: I would already be much happier if the voting policies were _known_, and I'll tell you why: For e

by sfk 17y ago

I already replied, but I think this warrants a second reply: I would already be much happier if the voting policies were _known_, and I'll tell you why: For example, I've frequently seen the idea that voting power should be based on some kind of average score of some recent comments. In the current bandwagon voting climate, this would mean that the guy who "wants to put bullets into the bad guys" (recent thread, +40 points) has more voting power than most other people here (a lot of _very_ informative comments linger at 1 or 2 points). If the voting policies were known, I could make an educated decision whether I still want to be part of this site.

nsrivast 17y ago
On the other hand, if the voting policies were known, people could more efficiently game the system. If the voting policy is unknown (and PG is committed to keeping it unknown), perhaps fewer people will worry about it's functional form, fewer people will care about what their vote multiplier happens to be, and more people would get on with the incredibly simple business of submitting interesting links and voting up what they find interesting. Perhaps. You should leave if the quality of the content deteriorates, not based on some background metric for an arbitrary scrip!
Perceval 17y ago
As I recall from past discussions, pg has indicated that his algorithm discounts outliers. So someone who has one 40+ comment and a string of comments rated 1 or 0 will not see the benefit of the outlier. A solution to piling on, if it persists, may simply be placing a hard limit on the maximum score a comment can garner. Slashdot and Kuro5hin both max out comments (Slashdot at 5, Kuro5hin used to max out at 5 but that was later changed to 3).
- pyre 17y ago
  > A solution to piling on, if it persists, may simply be placing a hard limit on the maximum score a comment can garner. Slashdot and Kuro5hin both max out comments (Slashdot at 5, Kuro5hin used to max out at 5 but that was later changed to 3). The problem being that sometimes there are posts that deserve well above the limit for their insight/knowledge. A logarithmic scale might be better. Up to 5 points there is a 1-vote-to-1-point correlation, after that it takes an increasing number of actual votes to increment the comment to the next point (i.e. 10 votes to get from 5 to 6 points, 15 more votes to get to 7 point, etc) That way a comment that is able to get a massive amount of votes gets 'rewarded' for it above a comment that is only able to get 5 votes.

by sfk 17y ago

nsrivast 17y ago

On the other hand, if the voting policies were known, people could more efficiently game the system. If the voting policy is unknown (and PG is committed to keeping it unknown), perhaps fewer people will worry about it's functional form, fewer people will care about what their vote multiplier happens to be, and more people would get on with the incredibly simple business of submitting interesting links and voting up what they find interesting. Perhaps. You should leave if the quality of the content deteriorates, not based on some background metric for an arbitrary scrip!

Perceval 17y ago

As I recall from past discussions, pg has indicated that his algorithm discounts outliers. So someone who has one 40+ comment and a string of comments rated 1 or 0 will not see the benefit of the outlier. A solution to piling on, if it persists, may simply be placing a hard limit on the maximum score a comment can garner. Slashdot and Kuro5hin both max out comments (Slashdot at 5, Kuro5hin used to max out at 5 but that was later changed to 3).

pyre 17y ago
> A solution to piling on, if it persists, may simply be placing a hard limit on the maximum score a comment can garner. Slashdot and Kuro5hin both max out comments (Slashdot at 5, Kuro5hin used to max out at 5 but that was later changed to 3). The problem being that sometimes there are posts that deserve well above the limit for their insight/knowledge. A logarithmic scale might be better. Up to 5 points there is a 1-vote-to-1-point correlation, after that it takes an increasing number of actual votes to increment the comment to the next point (i.e. 10 votes to get from 5 to 6 points, 15 more votes to get to 7 point, etc) That way a comment that is able to get a massive amount of votes gets 'rewarded' for it above a comment that is only able to get 5 votes.