What has happened to the sorting of answers on Stack Overflow? [duplicate]

Has the sorting of answers on Stack Overflow been changed? Because I've realised that the answers aren't sorted by

Votes -> Time


anymore, and they are sorted in a different way. It appears the time isn't taken in to consideration, now they are sorted by

Votes -> Random


If it is deliberate then what is the new order for sorting?

I have noticed that sometimes it sorts the old way, but then for others it appears completely random (apart from the votes).

marked as duplicate by gnat, rene, S.L. Barth, SPArchaeologist-様, WardJul 10 '16 at 14:27

• The second method looks very reasonable. – random Aug 26 '09 at 9:29
• Do you have a link to a question with randomly ordered answers (not random's ordered answers)? – Ladybug Killer Aug 26 '09 at 9:37
• Take your pick, it's happening to them all, as long as the votes are the same, click F5 a few times on the page and watch the answers dance about the screen. – Lloyd Powell Aug 26 '09 at 9:39
• E.G. I have an answer of a randomly selected question here. About 3 answers with 3 up votes in the following order: Aug 20 at 6:32 -> Aug 20 at 12:32 -> answered Aug 20 at 6:35....all with no edits. – Lloyd Powell Aug 26 '09 at 9:40
• @ThePower: Thanks for clarification. I can reproduce it when hitting F5. – Ladybug Killer Aug 26 '09 at 9:45
• @random I prefer random->random. – Mateen Ulhaq Apr 27 '11 at 4:58

I find this to be horrible decision (and I know I'm not the only one). We've spent a year now rewarding those who answer well and answer fast. I do not see why this is a problem.

If someone answers fast and they answer poorly, that gives them even more time to get voted down and let the better answers trickle to the top.

It also makes it entirely reasonable now for people to strategically downvote other answers even more so because they no longer have the "safety net" of being the first one keeping them above other answers with the same score. So expect to see a lot more lists of answers that are (0, -1, -1, -1). or even (-3, -3, -3, -3).

I do not see why we have to artificially reward someone for coming to the party late. If their answer is better, it should move up regardless of what time it was posted, and if it is a bad answer, it should move down regardless of what time it was posted.

• Looks like we have to use oldest as default view. – Ladybug Killer Aug 26 '09 at 13:54
• The old system worked great. I am very disappointed that Jeff was so spineless as to give into the complaints of a few new users who did not understand the system. – GEOCHET Aug 26 '09 at 13:54
• @Rich B: The Unpredictable God as spoken... again :-( – Ladybug Killer Aug 26 '09 at 14:00
• I wonder if Jeff has a rng he uses for making decisions as well as sorting the answers? – GEOCHET Aug 26 '09 at 14:35
• As someone who is consistently getting beat out by faster users, even I find this to be a horrible idea. The benefit of giving late answers a boost is going to be eclipsed by the amount of people who are going to abuse the system. – Brandon Aug 26 '09 at 14:35
• @Brandon: Lets not forget penalizing fast answers. This is not going to help us get good, fast answers. – GEOCHET Aug 26 '09 at 14:46
• I find this @RichB's a crucial point to consider: "Lets not forget penalizing fast answers. This is not going to help us get good, fast answers". – vmarquez Aug 26 '09 at 17:56
• Surely the problem here is tactical downvoting, not directly the ordering (though I personally prefer the old system too). Tactical voting needs to be eliminated and the only reasonable way I can see to do that is to prevent you voting on a question you also answer. – bananakata Aug 27 '09 at 7:56
• But I just had my question answer by a third person...all three have same 1 vote, but his answer moved to the top. WHY? – Honey May 29 '16 at 20:40

This is a terrible decision. It'll encourage duplicate answers and strategic downvoting. "Fastest gun in the west problem" was considered a "feature" of Stack Overflow. It encouraged quick answers. If your quick answer was wrong, you would just create more chance to get downvoted.

In most cases (not always, but most of the time) the better answer eventually floated to the top (esp. if was significantly better.)

The new system creates plenty of horrible problems (the slowest cheater in the east (SCITE) problem) just to solve FGITW, which was not a big issue.

For instance, today, I answered a question and a virtually identical answer appeared 3 minutes later and got upvoted. In this specific case, the guy himself was nice enough to upvote me but I honestly don't think it's a good thing to post a dupe answer after 3 minutes and leave it there.

Just bring the old system back.

• If it was long enough to post on it's own I would have posted 'Hear hear!'. – Lloyd Powell Feb 2 '12 at 10:28

After using this for awhile, I'm going to give the change a big thumbs up for one decisive reason.

I am now free to edit and improve my posts without fear of losing my spot among same-ranked posts. That alone far outweighs any other argument I've seen.

And that's a good thing.

• I agree with this one. If people are worried about tactical downvoting, I think another mechanic should be found to prevent this. The old system may not have encouraged fast bad answers, but it certainly encouraged fast mediocre answers. Why answer more than the bare basics, if it means you risk losing your spot? Why go the extra mile of adding information to your answer, when that puts you at the bottom of the list? +1 to this answer from me. Perhaps there's a third solution that would make everyone happy, but reverting to the old one seems like a step backwards. – jalf Aug 26 '09 at 17:52
• This system is still young. If a rep is down-voted occasionally by a "competing" user, they still only have one vote. If your post was really all that much better than theirs, it would be voted up by the majority anyway. In the bigger picture, reputation is an arbitrary number and the little anomalies thrown in here and there mean virtually nothing. As the user base grows, the influence single users have on the system will become less significant. Reputation is not a currency where we should worry about scraping and counting every penny. – Robert Cartaino Aug 26 '09 at 18:01
• @rcartaino, reputation is the fuel for this system. We can say "it's only rep" but if you took it all away, I wonder what would happen with participation levels. – hyperslug Aug 26 '09 at 18:21
• @hyperslug - I'm not denying the importance of reputation. I am 100% all for improving the efficiency of that fuel (to extend your metaphor) as a means of encouraging better content. I just wanted to point out that reputation is a cumulative process that is a pretty darn good approximation of [insert your favorite metaphor here]. The system works in the big picture. In my view, encouraging better content trumps the apparent misappropriations of single points by a small percentage of users in a temporary self-serving senario. So my thumbs up on the new feature. – Robert Cartaino Aug 26 '09 at 19:11
• I would upvote if I could ... – Ben Oct 4 '09 at 11:26

I've given this some more thought and basically I hate this change.

Just now I was reminded of a certain type of question that doesn't generate a lot of votes. I haven't thought twice about answering such questions before because either my answer is good and it'll get voted up or it's not and it won't. Now we have to contend with such answers now swimming about randomly in a bunch of Johnny-come-lately answers to the point where it's really not worth bothering because you now have to also hope that that when a random vote does come in, you're at the top at the right time.

Additionally, it's less worthwhile to put extra effort into answers to such questions as votes are even more of a dodgy proposition.

Also we risk another form of undesirable behaviour. Previously people would put in the placeholder, which I didn't think was a problem. This won't stop that. It'll make it worse. The reason is now you put in a placeholder answer and just wait for a real answer and then cannibalize it. If it happens within 5 minutes who's to say who had it first?

Are we going to rely on people to check timestamps where previously the order alone told them who was first?

How long before the user base at large actually cottons on to this change and just assumes whichever answer is randomly at the top was the first?

All for what?

Edit: I feel compelled to add my own thoughts (in more detail) to this. Well not just this but how it fits into a larger picture. Not that anyone will read it because the redditors will downvote it into oblivion but oh well.

• Just out of curiosity, why not link directly to your blog post? – Bill the Lizard Aug 26 '09 at 18:31
• 'cause the reddit comments on the last one were so interesting and insightful... ! – Shog9 Aug 26 '09 at 18:38
• @Bill: because the original "Why Stackoverflow Sucks" crapstorm was on reddit so it seemed the right place. Plus I thought I'd give the redditors a chance to read an informed opinion but the only way that would happen was if there were upvotes to counter the automatic downvotes that is reddit (ie the hopes people here and elsewhere would upvote if they were so inclined). – cletus Aug 26 '09 at 18:42
• @cletus, why not post your proposed solution here? "My personal preference is either to not have anonymous downvoting or not to have it when you also answer in the interests of full disclosure." That would provide some accountability. – hyperslug Aug 26 '09 at 18:46
• Can someone post a link directly to the blog. Reddit is blocked by Lightspeed systems. – Rob Allen Aug 26 '09 at 19:34
• @hyperslug: proposed solution? To what? This sorting thing is one thing and easy to fix: just don't do it randomly. Problem solved. My blog post is more about how this fits into wider issues. – cletus Aug 27 '09 at 2:22
• Again, the existence of placeholders because of free edits is the direct problem here, this new ordering system just enables it. – bananakata Aug 27 '09 at 7:57

It is by design. Jeff's comment on this answer says

we now sort answers of the same score, randomly

which was implemented to "give some boost to some really good answers that arrive late"

Edit to include my own POV.

I've seen the tactical downvoting right after the random ordering of similarly rated answers was implemented. It's crazy. For a question, around 5 to 7 answers were almost the same and almost all were correct but each got -1, and -2 votes. I think it's causing chaos and is very unhealthy for the community.

• Now I wonder why I got a downvote just now. Did I not give the correct answer? – Randell Aug 26 '09 at 13:57
• I suspect it's someone disagreeing with the change, not with your answer. – tvanfosson Aug 26 '09 at 14:12
• @tvan: Or someone trying to change the ordering of questions. You can't tell these days. – GEOCHET Aug 26 '09 at 14:38
• LOL. But accepted answers stay on top. – Randell Aug 26 '09 at 14:41
• @Randell: Hey, I am sure the same ignorance that made people think FGITW is a problem would lead them to believe they can or should change the accepted answer. – GEOCHET Aug 26 '09 at 14:43
• @Rich B, is this rhetorical? – Randell Aug 26 '09 at 14:49

The devil you know, or the devil you don't?

They've replaced what i considered a small, extremely limited opportunity for gaming with what feels like a big bold invitation for gamers to have at it.

But now you're playing craps instead of chess. So the fast and the good still have to compete with the gamers, but now the gamers can play around with chance instead of competing with quality.

So long, old devil. I wonder what the new one has in store for us...

I don't understand why people are making such a fuss of the new sorting order.

The most important metric on judging if an answer is good or not is the number of votes, and that's still what the questions are sorted by. Then they are sorted randomly.

If you don't like the new order, write yourself a greesemonkey script and god forbid stop whining.

Personally, the new order prevents those one-hit wonder answers of just posting a link when some of us actually try to write a complete answer and then gets ignored because someone posted a totally incomplete answer answer posted as such:

Which is not the kind of answers SO should have.

Downvote this answer all you wish, the only people which are disadvantaged by the change are the cowboys who tries to fire first with the smallest bullet they can find. At least now a GOOD, THOUGHTFUL answer posted 5 minutes later has the same chance as a one-link wonder crap answers that we are seeing way too often nowadays.

• "At least now a GOOD, THOUGHTFUL answer posted 5 minutes later" I see this all the time, and yet... I'm not terribly fast. Sure, i'll probably never see my name on the first page of users, but i do all right. I've many times posted a good answer late to a question where an existing sloppy or incorrect answer had already collected upvotes (which this change won't even touch), and still found my answer able to garner sufficient attention. – Shog9 Aug 26 '09 at 18:30
• Honestly, greasemonkey is never a good answer. If there's a problem or a needed feature that affects everyone, people shouldn't have to hack around it themselves. – womp Aug 26 '09 at 19:05
• +1 for the "This link will answer your question" point. I see this all the time when better, newer answers are ignored. – Alex Angas Aug 27 '09 at 5:58

On the one hand it's clear that people would put in placeholder answers and then fill them out in the next 5 minutes but, as has been previously discussed, I don't think that's a bad thing. Getting a quick answer is useful. If the placeholder isn't useful it gets downvoted and any benefit of first placement is lost (although you may find #2 or #3 tactically downvote it, even temporarily).

So the question is either what problem did this solve or what benefits does it bring to the table?

I can't help but think that maybe the "its not fair!" crowd have managed a vocal minority on this issue. By "its not fair!" I mean you get responses to issues like people having high rep that basically come down to "its not fair!" (that person X put 8 months more effort in that I did and thus has higher rep).

So one could argue "its not fair!" [tm] (that someone takes the time to quickly get an answer to a question).

It's also worth mentioning that the previous system discouraged "me too!" answers. Sometimes you see 3-4 people post the same answer within a short period. Fair enough. They're typically not aware of each other. But sometimes you'll see someone do it 20 minutes later. They can't not have seen the others so why bother to post the same thing?

This way their answer has as much chance as anyone's of being first (upvotes notwithstanding). So are we encouraging "me too!" noise with this?

• I, too, was going to post about my concern over tactical downvoting. It used to be enough to be fast, but now that votes are the only sort order, I suspect that there will be much more temporary tactical downvoting. – Kyle Cronin Aug 26 '09 at 13:48
• +1 simply for the point about the duplicate answers. If you have 5 people post the exact same thing, the first person should be getting the credit, not whoever just shows up first at the time the user clicks on the question and reads through the answers. – TheTXI Aug 26 '09 at 13:51
• @TheTXI: so we should make oldest the default ordering? I think that is what could be done mitigate this... – perbert Aug 26 '09 at 15:32
• +1 for the note on duplicate answers - they were enough of a problem already, and really didn't need this kind of encouragement. – Shog9 Aug 26 '09 at 15:56
• Placeholders get downvoted if they're clearly placeholders. Most "fastest guns" learn to put in a sufficient plaeholder and then flesh it out in the next few minutes. Those don't tend to get downvoted (if done right). – cletus Aug 27 '09 at 2:24
• but the point of the voting system is to give the "same measure" for answer of similar caliber - if 3 people answered the same thing, each of them deserve to have the same votes over time, regardless of who wrote the first answer. The random order gives you this. – Chii Sep 22 '09 at 13:50

I've read the points so far and can see some merit. However I've seen the change already help people who come in with a much better answer but later. Some answers take time to research and explain well (we aren't all Jon Skeet and never will be). This change really helps that.

If strategic downvoting strategies are employed then that is vote fraud, pure and simple. It should be detected and those votes reversed / users disciplined.

I didn't even notice the change.

• this is hilarious given the amount of angst I've endured over this – Jeff Atwood Sep 3 '09 at 21:21
• @[Jeff Atwood]: perhaps I'm just unobservant, or jaded, or uncaring, or... don't read questions/answers nearly as much or as carefully as the others ;-) – Steven A. Lowe Sep 4 '09 at 1:10
• @[Jeff Atwood]: that said, I think you may have tried to fix something that wasn't broken - time-ordering makes sense and provides a repeatable/consistent user experience; random ordering (for those that notice such things) not so much ;-) – Steven A. Lowe Sep 4 '09 at 1:11

I’m surprised this change got even noticed, let alone drew so much fire … because, honestly, you all sort by time anyway, right? right?

Oh dear …

Granted, the default sort order (by votes) makes sense for people who have just googled/bing’d for the problem and found the answer on SO. What they want is answers, quickly and they don’t want to read through pages of bad (but recent) ones to get to the gems.

But everyone who has complained about the new system so far has a huge reputation score. You’re the active folks, people! You don’t read, you contribute. You examine the existing answers, correct, amend, chastise, and complete. In one word, you interact. For you (or at least, for me) the answers aren’t isolated but related to each other and reading them in the wrong order (i.e. not in temporal order) would make as much sense as hearing random snatches of a conversation in the wrong order.

The bottom line of all this is: the new algorithm is completely irrelevant for me, since I don’t notice it. I think it might even be preferable for non-contributors. But for me (and for you) it shouldn’t matter. And just in case you’re wondering whether I’m asking you to change your answers ordering, let me remove all doubt: I am.

EDIT

Of course, if this measure is meant to prevent tactical downvoting, it’s a bad solution. Tactical downvoting could be prevented once and for all if they would just finally implement that tiny request that has been flying around for ages, that you can only downvote if you explain the downvote in a comment.

• Heh... I sort by time (oldest first) on Meta, but by votes on SO - when i'm looking for answers, i want to see the highest-ranked answer first. And yes, i probably read more than i write. – Shog9 Aug 26 '09 at 19:00
• it's not at all meant to prevent tactical downvoting. Something else is going in tonight for that. – Jeff Atwood Aug 27 '09 at 6:15

During this week I had a strong feeling that it was much easier to gain rep from the upvotes then before.

Don't know if it's linked or not, but I finally have the highest* score on SO.

* Among the users with default gravatar

• This is an unrelated issue. It's a side effect of "accepted answer rate." This thread is about "random ordering of answers with equal number of votes" – xmm0 Aug 26 '09 at 17:58
• @Mehrdad: upvotes seem to boost too. – Quassnoi Aug 26 '09 at 18:01