Although of course I did not answer the original poster's question, don't you agree it is a bit humorous that by trying to answer most people probably will try to vote up to check if they have the same problem?
There's perhaps even something to learn from that.
Yes, true. I could've hinted that this kind of post could also be interpreted as an attempt to game the voting system in a creative and thus perhaps interesting way. But I didn't.
It also happens to me when up- or downvoting comments.
I think there are safeguards in place to prevent us from voting too rapidly. E.g. when I go through a thread and "adjust" it to my liking (up/down vote as I see fit) and then reload, then usually half of my votes didn't make it.
I'm trying to get into the habit of voting immediately, but keep falling back to "buffered I/O" when I'm not paying attention...
Someone will probably correct me, but does it have something to do with your average karma per post? I remember reading something about being under 2 average effects your voting.
Not all votes are counted, and not all votes that are counted are displayed (for instance, when you downvote someone even if they are at -4 the votecount seems to be stuck at -4 but the karma of the person is still decremented).
Though is the "secret sauce" relating to HN specific voting fraud detection, etc, open source too? As I don't "do Arc" I've never bothered to look but pg has mentioned a few "techniques" they have for detecting fraudulent voting and accounts, and I can't imagine that's part of the core code..?
Yes it is, the code is written in a language layered on top of scheme called 'Arc', you can get the whole package here: http://arclanguage.org/ , the package includes a slight variation (the 'secret sauce' is missing) on the code running HN.
Funnily enough, my karma recently crossed the 200 point threshold, but I didn't recall the down arrows appearing for comments. I just logged in to tell you all that there must be more to it, since I've got 200+ karma and no down arrows. But lo and behold, there they are. I'm not sure if those are actually new or if I simply never noticed them before. If the threshold is not 200, it must be close to that.
If the arrow is back after reloading the page, the reference ID of the upvote URL timed out. This occasionally happens to me with long articles - I'll open the HN comments page in one tab, read the article in another, then come back to upvote it after the page has been open for a while, especially if I ended up doing something else in between.
22 comments
[ 14.3 ms ] story [ 798 ms ] threadAlthough of course I did not answer the original poster's question, don't you agree it is a bit humorous that by trying to answer most people probably will try to vote up to check if they have the same problem?
There's perhaps even something to learn from that.
It redirects to the login form, and then when you return, the up-vote arrow is gone but the vote isn't counted.
It also happens to me when up- or downvoting comments.
I think there are safeguards in place to prevent us from voting too rapidly. E.g. when I go through a thread and "adjust" it to my liking (up/down vote as I see fit) and then reload, then usually half of my votes didn't make it.
I'm trying to get into the habit of voting immediately, but keep falling back to "buffered I/O" when I'm not paying attention...
Not all votes are counted, and not all votes that are counted are displayed (for instance, when you downvote someone even if they are at -4 the votecount seems to be stuck at -4 but the karma of the person is still decremented).
This question and a bunch of others I've written up in this article: http://news.ycombinator.com/item?id=1755533 .
No, then it wouldn't be secret anymore.
Security by obscurity is not perfect but in this case having that knowledge out there would only benefit the spammers and the 'gamers'.
Is news.ycombinator.com open source ?
Yes it is, the code is written in a language layered on top of scheme called 'Arc', you can get the whole package here: http://arclanguage.org/ , the package includes a slight variation (the 'secret sauce' is missing) on the code running HN.