254

First, apologies for not creating this feedback thread earlier - in hindsight, a change to such an established feature of our sites really does warrant some explanation.

We're running an A/B test on Stack Overflow only that places half of registered users with >= 15 reputation in a variant group that will see voting controls like so (shamelessly stolen from ᔕᖺᘎᕊ's answer here):


sticky vote controls example


A few points about this current experiment:

  • If you're in the variant group, the controls will only move if you can vote on the post; your own posts will not have the sticky controls
  • This will run for another three weekdays, so it should finish up next Wednesday, 2015-06-10.
  • We've set acceptance criteria at 5% more voting than the baseline.
  • After the experiment concludes, I'll update this question with the results.

Edit 2015-06-08 01:50

We're into the next variant with these changes, based on feedback:

  • stickiness is applied if the post's body is longer than 2/3 of the browser window's viewport
  • stickiness is applied regardless of a post's votability, e.g. the controls will be sticky on your own posts (if they're long enough)

Edit 2015-06-09 19:55

Experiment has completed; preparing a summary.

Edit 2015-06-12 18:30

TL;DR this feature will not be implemented, as it had too little of an impact on voting.

We ran two* variants of these sticky voting controls:

  1. variant 1 results: movement was only enabled on posts you could vote on; no minimum height requirement to the post body; ~3.8% vote loss from baseline
  2. variant 2 results: movement was enabled on all posts if you had at least 15 rep; post body height had to be longer than 2/3 of the viewport; ~2.1% vote gain from baseline

Some definitions for the above result links:

  • baseline: the current behavior, i.e. voting controls do not move from their initial placement
  • trial: a question page was rendered to a user who could at least upvote something on the page
  • success: the user upvoted or downvoted one of the question or answers

We placed a 5% minimum improvement threshold for acceptance; this is completely arbitrary, but it's a common floor in our experiments, as new features == new upkeep. Also, judging from the extreme polarization of the responses to this feature, we would need a user preference to disable it, something we're hesitant to provide for any feature.

Going forward, we might try another experiment with duplicating the controls at the bottom of long posts, but there's no ETA for this.

If you did like the functionality, it can be added via ᔕᖺᘎᕊ's SE Additional Optional Features script.

* an initial variant was stopped after only an hour and this change was applied

  • 128
    This is the only time I've wished I had 125 points on meta. If you deploy this, please add an option to turn it off. – Praetorian Jun 6 '15 at 7:19
  • 6
    Just curious, was this change sitting in a drawer for long time just waiting for someone to ask for it, or was is an ultra spontaneous decision? – Shadow Jun 6 '15 at 10:26
  • 9
    I take it that it's the A/B test results ("is there 5% more voting?") which you'll use as your "feedback". So what kind of feedback are you looking for in answers to this meta-question? – ChrisW Jun 6 '15 at 10:39
  • 9
    "We've set acceptance criteria at 5% more voting than the baseline." is there any basis for this percentage? Shouldn't the acceptance criteria be "any improvement and no significant complaints"? – usr Jun 6 '15 at 15:51
  • 93
    I will be short on it: I HATED IT! This is exactly what I hate. It feels like it is begging to vote. Like those awful adds on some websites that follow you, screaming "CLICK ME! CLICK ME! NOTICE ME!". – Ismael Miguel Jun 6 '15 at 16:55
  • 34
    I like it - it reduces scroll time on the really long posts (which are often the most vote-worthy). – Hosch250 Jun 6 '15 at 17:40
  • 13
    @JarrodDixon THEN DON'T DEPLY IT. Please! Before deploying new features, you could showcase them and figure out if the users want it. But before that, you could solve the existing bugs first. – Ismael Miguel Jun 6 '15 at 18:55
  • 11
    I like it personally. I sometimes write books (well, maybe short stories) as answers. I'll get a ton more up votes this way! :D J/K! I do like the idea though. I'm a firm believer in voting ... it's what makes the SE work. If there is some simple way to remind people to vote, we should do it. This is simple and innocuous. Get 'er Done! – Pᴀᴜʟsᴛᴇʀ2 Jun 6 '15 at 19:19
  • 15
    What is that horrible animation supposed to do for us? Stop it, please! – Martin F Jun 6 '15 at 19:46
  • 25
    @IsmaelMiguel As always when some new feature is added to a product, the "fix existing bugs first" complaint is fallacious. Just because they're adding this it does not mean they are not working on other bugs, nor that the advantage of adding this would not be greater than those of fixing other bugs. – nico Jun 6 '15 at 20:47
  • 14
    @IsmaelMiguel again, your reasoning is fallacious. Are the people working at this the same that are supposed to be working at the IE bug? Would they work at that bug if they were not working at this? Is that bug affecting a sufficient amount of people to be worth investing time solving? And so on... Unless you work at SO it's difficult to judge these things, as we most likely don't have the larger picture. It's absolutely fine if you don't like the scrolly votey thingy. I am just saying that "solve bugs first" is irrelevant here, as the discussion is on "do you like the scrolly thingy". – nico Jun 6 '15 at 21:04
  • 8
    I just want to say, now that I'm over the initial shock and stare of this, I think it is an amazing feature and hope that it doesn't go away. I love it so much and it's almost like it offers a sense of security knowing those buttons are always right there when I need them. Don't change a thing :) – Diminutive Colossus Jun 6 '15 at 22:40
  • 11
    Agree with @Praetorian. Please at least make it optional/configurable. – aroth Jun 7 '15 at 2:59
  • 7
    Somebody in this thread really hates A/B tests. – BoltClock's a Unicorn Jun 7 '15 at 8:31
  • 38
    Can it be made to blink too? Maybe with a marquee border around it. – j08691 Jun 7 '15 at 19:43

39 Answers 39

1

When there are many comments, and you scroll past the post to see the comments, the vote icon should not follow you past the post.

Besides that the only thing I don't like about it is that it does add complexity (more for the user to think about), and if you add a switch to turn it off, that would add complexity as well. Obviously we have to balance the overall complexity of the system with the overall featureset.

1

It is much easier to deal with now (phase 2). Phase 1 I wanted to scream, now I do not even notice it.

1

Just to report a quick "bug" I found on this subject (not quiet sure if you guys already found this yet or not).

Bug A/B

So what you see is the votes of question (which has 0 votes) on top of votes of the 1st and only answer (which has 3 votes).

This "bug" was found using FF38.0.5 with Win7

Despite this, so far, I like the new sticky votes, although I think in a UX point of view (IMO) the old way is better.

1

The vote buttons get lost in case there's a long list of comments, i.e. it does not scroll more downward than the length of the question or the answer.

I often read comments if available, because some of them have good points, and then vote. In case of a long list of comments, even with the new feature enabled, I have to scroll back to reach the controls.

If the vote buttons follows me, I'd expect them to follow more.

  • Deliberate change. Perhaps to avoid accidental voting for a "comment". – Bill Woodger Jun 9 '15 at 16:32
0

Having control buttons in a fixed position is something quite common on today's web experience. You see the top navigation bar fixed at the top of a lot of news paper site and e-commerce sites. More annoying for some users (who don't use an AdBlock tool), is the fact that advertisement banners, usually on the right side of the page and not inline with content in some cases is fixed to the top while scrolling.

What is new here is that the buttons also fixes at the top of the page while scrolling downward in the same answer. These are inline controls and apart from sticky ads I think the experience from users will be positive, once you get used to it. It fix the problem scrolling back up to vote, and would probably improve the number of votes.

The con is that we are no longer looking at content in fixed position. This is especially true when asynchronous loading of pages mess up the current viewport and move what you're reading away. But the implementation here, doesn't do that cruel design choice. And we have been moving away from fixed pages since the introduction of scripted pages in the middle of the 1990s. This is another invention in that direction.

So from a UX perspective I believe it's a good thing where pros outrank cons.

-1

That looks creepy! No doubt on that. I just started to see this and thought that either I need a sleep right now or SO implemented a new feature without telling us first. Then I found this post. Well as feedback, I am sure this is a nice feature. I first saw this in my Windows Phone when the list of apps are scrolled the alphabets stick to the top and keeps changing whenever the new alphabet comes out. That looked nice there, but frankly I felt creepy. That's possibly because I didn't have enough sleep or because I am not used to this on SO. In logic that does look like a productive feature (you do not need to scroll long answers to vote.), but my eyes are not comfortable with it..yet.

So I would suggest as everyone has suggested to make this an option. You can check how many of the users have turned it off. So the benefit would be that the user will have an option to choose whatever they feel comfortable with and you will get the data of how many wants that feature. (Of course you will need a way to tell people about where is the setting to turn it off).

Personally I will turn it off for few weeks and then turn it on only when I am fully sober. So that my eyes doesn't feel the pain when they are most vulnerable.

-2

Please do not implement this or if you do then please allow an option to turn this off. I would vote but I do not have enough reputation on this site.

-4

This is a very interesting idea. I'm not in the lucky testing group, but I've been thinking for a while that a change was needed to call out the voting more.

I think it would be even more interesting, effective, and fair, if you hid the vote counts behind an ellipsis or something similar and reveal on click. The answers would still be ordered by score, but it might reduce the likelihood that some users will upvote just because they liked an answer and it already has a high score, not knowing if it's truly correct. Ie, reducing "score bias".

Design-wise it looks uncomfortable to me in the animated example because it's floating and drawing the eye constantly on scroll. I'm imagining it might be more comfortable if it sat on, or next to a line and then disappeared into the 'share | edit | flag' line, which shouldn't be floating either.

I think if you introduce a sticky moving element, you need to carve up and visually define the structure more. The minimalism is too minimalist :) For example, with a border between the content and the vote counter here on meta, there would be a stable visual anchor at the start of each line. All speculative ideas.

-5

Here is a possibly more effective way to give feedback:

If you are in the testing group, and don't like the feature, I suspect the best way to voice your disapproval will be to abstain from voting while the test is running.

The metric (5% increased voting) could be weighted heavily in the website's final decision. By abstaining from voting, you can send a message that the effect is not appreciated via a side channel.

Similarly, if you are in the testing group and like it, vote with abandon.

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .