Here are the comments he made in reference to removing average karma [1][2][3]:
> dang 781 days ago
> It's based on comment scores only. My sense is that it's an irrelevant distraction and we should get rid of it.
> For example, we caught a bunch of users who were gaming it by deleting any comments that brought their average down. We fixed that by treating deleted comments as comments of score 0 for average-computing purposes. It's an example of how, once you publish a metric, people start to care about it and do things based on it, regardless of how meaningful it is.
> Comment average used to be used by a few algorithms (like comment ranking) but we turned that off as an experiment a while ago and nothing seemed to get worse. If anything, I think it may have helped a little.
---
> dang 572 days ago
> What brudgers said is correct, but I'll add that we're probably going to get rid of comment averages. We've looked at the data extensively and it hasn't proven to have much value. We've phased out the code that used to rely on it, and what's left (which is only for display) is probably not worth the cost.
---
> dang 567 days ago
> ... We got rid of average because (a) after looking extensively at the data we didn't see any value in it, (b) we had evidence of people gaming the metric, (c) its implementation was complicated, and most importantly (d) we think upvotes are wrong thing to optimize for. Optimize for saying substantive things.
I'll add to this that statisticians taught me averages are actually meaningless when applied to individuals. The number is tied to a data set but that data set can be all over the place in terms of what it contains. You can't know where you're at relative to others in a data set just by comparing to an average. Unless it's the median where you know you're above or below the middle person without further data on what that means. Pretty useless.
From what I was taught, averages are mainly (or solely?) there to measure trends in sets of data. Basically, where it's collectively going over time. Applying it to individuals is meaningless. So, that feature should never exist unless you're intentionally exploiting psychology & common misinformation for personal gain. Example: showing players how they compare to an "average" so at least half of them think they're better than average and keep playing to get number higher. Another example: differentiating a software company with terrible defect rate by comparing it to the industry average, which is worse, in an industry full of shoddy software. The software looking good is still crap but "better than average." ;)
> dang 781 days ago
> It's based on comment scores only. My sense is that it's an irrelevant distraction and we should get rid of it.
> For example, we caught a bunch of users who were gaming it by deleting any comments that brought their average down. We fixed that by treating deleted comments as comments of score 0 for average-computing purposes. It's an example of how, once you publish a metric, people start to care about it and do things based on it, regardless of how meaningful it is.
> Comment average used to be used by a few algorithms (like comment ranking) but we turned that off as an experiment a while ago and nothing seemed to get worse. If anything, I think it may have helped a little.
---
> dang 572 days ago
> What brudgers said is correct, but I'll add that we're probably going to get rid of comment averages. We've looked at the data extensively and it hasn't proven to have much value. We've phased out the code that used to rely on it, and what's left (which is only for display) is probably not worth the cost.
---
> dang 567 days ago
> ... We got rid of average because (a) after looking extensively at the data we didn't see any value in it, (b) we had evidence of people gaming the metric, (c) its implementation was complicated, and most importantly (d) we think upvotes are wrong thing to optimize for. Optimize for saying substantive things.
---
[1]: https://news.ycombinator.com/item?id=9297678
[2]: https://news.ycombinator.com/item?id=8264220
[3]: https://news.ycombinator.com/item?id=9322256