4 ms·
> Notice how there is a comment on that answer saying that it's wrong but that comment has only 38 upvotes whereas the answer has 56 upvotes. A youtube comment
by Etherlord87 3y ago
> Notice how there is a comment on that answer saying that it's wrong but that comment has only 38 upvotes whereas the answer has 56 upvotes.
A youtube comment won't get more upvotes than the video it refers to. It's just a matter of structure, of visibility.
It's nice that there is a comment. The author apparently also doesn't know he can (propose to) edit the existing answer. If he doesn't have enough reputation, it will have to be approved by another member (it will be in a moderating queue).
I can agree that it is frustrating sometimes to see an invalid answer being upvoted... Especially if you write the correct answer but the invalid one remains with the higher score just because it was there first, or because the voters are unable to tell it's incorrect and unable to tell your critique is correct. But that's not really related to SE.
- pwdisswordfishc 3y agoThis is a matter of a skewed voting system too. The comment explaining the answer is wrong has 38 upvotes, but the answer has only 12 downvotes. This means there are at least 26 people who agreed with the criticism, but were either unable to cast a downvote or refused to cast one (in the name of Positivity™, probably).
- Etherlord87 3y agoYou spend your reputation to downvote an answer (a question can be donwvoted freely). The positivity® is an issue too, I agree. Or maybe anti-negativity. On blender.stackexchange.com, where I'm active, we have a problem of few people caring to upvote questions. The best way yo motivate people to upvote a question is to... Downvote it. As long as a question isn't absolutely terrible, if it has negative score, people will upvote it to 0. Which generates reputation for the author of the question, because he gets -2 rep for a downvote and +10 for an upvote.
- ahmedfromtunis 3y agoBut SE is not YouTube. On the latter, it doesn't matter what comment has the most likes. On SE however, one can say that what answer gets the most upvotes is the most important aspect of the site. For me, the date of the answer had become a key information to check before I even reading the answer. I use Django a lot and it significantly evolved over the years. And for too many times, I'd spend time implementing a solution, only to realize it no longer worked since version x.y that was published z years ago. The thing is, stack exchange was built to serve an industry characterized by its very rapid evolution overtime. Yet, it fails, IN MY OPINION, to adress such a key element in a graceful way. The documentation industry solved this years ago. Going back to Django again, its doc site allows users to select the version they're interested in order to show them relevant content. I think stack exchange needs to implement something similar for its answers. How can that work? Well, users can be allowed to add tags to mark versions where they know the solution doesn't apply anymore. It may not be the earliest version where this issue arises, and that's understandable. Other users can later refine the threshold data. SE can then display the info where searchers can easily spot it [Answer not applicable since x.y] and avoid them some confused head scratching. We can push this even further with multiple correct answers for a single question, depending on the version. Users can even add the version of different software they're interested in and see the correct answer that applies higher, regardless of number of upvotes. (Written on my phone. Apologies for typos.)
- Etherlord87 3y agoOn Steam there's a separate "recent" rating. I think SE/SO would benefit greatly from it as well. Allow people to set in their preferences to disregard old scores and use only the new score.
- Avamander 3y agoVote decay probably has been discussed a lot already on Meta StackExchange.