Ask PG: Can we bring back the upvote/downvote numbering on certain posts?
It defeats the purpose of crowdsourcing when the numbers are hidden. I understand why its done on posts about politics or posts about argumentative topics. But sometimes people ask genuinely complex questions and the always-awesome HN crowd weighs in with great answers based on their experience. However, we can't see which answer has the most backers, or which answer most people consider right. It defeats the purpose of the wisdom of the crowd when the wisdom of the crowd is hidden.
Can you enable an option to allow a submitter in the "Ask HN" section to activate comment upvote/downvote counters?
6 comments
I really miss the numbering, I've got to say. It makes it significantly harder to skim through the comments section, and it frankly has made Hacker News less enjoyable and useful for me as a reader.
To counter, it has made HN more enjoyable for me. I perceive significantly less karma-whoring and ego games now that it is gone.
You can't see the numbers, but it's not really accurate to say "You can't see which answer has the most backers." The comments are still ordered by popularity. In general, the ones near the top have more backers and the ones near the bottom have fewer.
In general yes, but that doesn't tell you if comment number 2 is way ahead of comment number 3, or if they both suck compared to comment number 1. And see my comment below about replies. How are those ranked?
I cannot speak for pg, but I would guess that he is targeting the site at people who can make up their own mind given the arguments, with the voting system used to let the good stuff float up and lesser stuff sink.
Why do you care so much about the scores, either within the same post or related to other posts?
You are aware, I hope, that without the vote timeline, what you are asking for is completely meaningless -- replies posted when something is on the front page get significantly more votes/replies than those posted after the thing has dropped to the 2nd page, regardless of merit.
For the score to be meaningful as a number, scores have to be normalized with respect to no. of page views, time since post, and a lot of other things.
Why do you care so much about the scores, either within the same post or related to other posts?
You are aware, I hope, that without the vote timeline, what you are asking for is completely meaningless -- replies posted when something is on the front page get significantly more votes/replies than those posted after the thing has dropped to the 2nd page, regardless of merit.
For the score to be meaningful as a number, scores have to be normalized with respect to no. of page views, time since post, and a lot of other things.
I care because I want to know the right answer. Sometimes people post questions seeking the wisdom of the crowd, and its impossible to tell which comment gets the highest vote. Which negates the reason for asking the crowd in the first place.
Its kind of like asking your friends if you should buy A or B. If all of them say A, you get a good handle on the reasons to buy A.
This sort of question works fine if commenters just say A or B in the first comments. But what if someone replies to that and says "What about C, that's an even better alternative." And then everyone agrees C is better. But the current system doesn't allow us to see that accurately, so I'm asking if it can be turned on for people wanting to ask questions that way.
Its kind of like asking your friends if you should buy A or B. If all of them say A, you get a good handle on the reasons to buy A.
This sort of question works fine if commenters just say A or B in the first comments. But what if someone replies to that and says "What about C, that's an even better alternative." And then everyone agrees C is better. But the current system doesn't allow us to see that accurately, so I'm asking if it can be turned on for people wanting to ask questions that way.
> its impossible to tell which comment gets the highest vote. Which negates the reason for asking the crowd in the first place.
Wait a couple of days after the last comment you are interested in was posted; the time decay among different replies will be approximately equal, and the one on top is the highest, but
> I care because I want to know the right answer.
> Its kind of like asking your friends if you should buy A or B. If all of them say A, you get a good handle on the reasons to buy A.
The reasons are in the text, not in the vote, and you have to read them in order to appreciate if this is relevant to you or not. No way around it.
> This sort of question works fine if commenters just say A or B in the first comments. But what if someone replies to that and says "What about C, that's an even better alternative." And then everyone agrees C is better. But the current system doesn't allow us to see that accurately, so I'm asking if it can be turned on for people wanting to ask questions that way.
Come on.... What if that "what about C reply" was posted 5 days later, when no one who posted A or B was still looking? does the fact that it didn't get much upvotes make it any less valid?
Also, time of day matters -- at different times of the day, the answer statistics are dominated by people in Asia/Australia, people in Europe, and people in the US, etc, and they might have wildly different circumstances. You have to wait a couple of days for the dust to settle and everyone to chime in.
I'm sorry to say that, but your post have convinced me even more that pg did the right thing: There is much, much less information in those numbers than you ascribe. Even the rank is super biased. If you base your view of "what's right" on the number of votes (or even on their rank) rather than the actual content, perhaps HN is not the right place for you.
Wait a couple of days after the last comment you are interested in was posted; the time decay among different replies will be approximately equal, and the one on top is the highest, but
> I care because I want to know the right answer.
> Its kind of like asking your friends if you should buy A or B. If all of them say A, you get a good handle on the reasons to buy A.
The reasons are in the text, not in the vote, and you have to read them in order to appreciate if this is relevant to you or not. No way around it.
> This sort of question works fine if commenters just say A or B in the first comments. But what if someone replies to that and says "What about C, that's an even better alternative." And then everyone agrees C is better. But the current system doesn't allow us to see that accurately, so I'm asking if it can be turned on for people wanting to ask questions that way.
Come on.... What if that "what about C reply" was posted 5 days later, when no one who posted A or B was still looking? does the fact that it didn't get much upvotes make it any less valid?
Also, time of day matters -- at different times of the day, the answer statistics are dominated by people in Asia/Australia, people in Europe, and people in the US, etc, and they might have wildly different circumstances. You have to wait a couple of days for the dust to settle and everyone to chime in.
I'm sorry to say that, but your post have convinced me even more that pg did the right thing: There is much, much less information in those numbers than you ascribe. Even the rank is super biased. If you base your view of "what's right" on the number of votes (or even on their rank) rather than the actual content, perhaps HN is not the right place for you.
I think you're discounting the value of people's opinions too much. Basing a decision on what the denizens of Hacker news think is the best answer is a perfectly acceptable method of moving forward. Whole businesses and methods are based on this sort of wisdom (Crowdsourcing answers, the Facebook "Like" button, re-tweets on Twitter). Choosing the most popular answer might have it's drawbacks, but sometimes its the right answer.
You have a point in that time of day and all the other factors you listed might skew the results, but it might not skew them in a way that bothers me. If something is the best answer, I'm sure people in Asia or Europe or America can pick it out. They might have wildly different answers, or they could also be exactly the same.
I get the impression you think just because people "liked" one answer the best isn't a technical enough reason to go with that answer, and in fact you think the decision should be based on something more substantial. I disagree that this always needs to happen.
The answers near the top of the page got the most up-votes. The ones at the bottom got fewer votes or have a negative score.
And the replies to the answers? For example, if a reply gets the most votes. How do I know when that happens?
The reply will be listed first among all those replying to the same parent. There's also a time-decay factor that comes in -- in essence, when ordering, the votes on a newer post are worth more (and as time goes by, all votes equalize in value)
Right, but the relationship between replies is not known. Is reply number 1 far ahead of reply number 2? Or are they close. Plus, what about replies with different parents. How do you tell which reply has more votes?
Maybe that kind of complex question would be better suited to StackExchange instead of HN?
Comments are scored by length initially. This is short, hence it will be at the bottom.