-9

Surveys preferably conceal their results to remove response bias. Blind survey votes naturally induce trust, we should embrace that reality.

Here is a partial list of the biases blind votes eliminate, making them intrinsically more valuable than naked votes:

  • Status quo bias: Preference for the current state of affairs or "Bandwagoning"

  • Outcome bias: An error made in evaluating the quality of a decision when the outcome of that decision is already known. E.g., "If I downvote this post it will be closed; therefore, I will not downvote."

  • Omission bias: The tendency to favor an act of omission over one of commission. E.g., "These votes suggest the post is controversial therefore: I will not vote."

  • In-group favoritism: Favoring members of one's in-group over out-group members. E.g., "Moderators like me upvote, I should upvote!"

  • Anchoring: An individual depends too heavily on an initial piece of information offered -the "anchor" - when making decisions. E.g., "This post has a -10 score. I will look more critically at this post to decide."

Systematically removing bias promotes trust in new users, which will increase our user base.

For these reasons I feel voters who choose to vote in the blind should earn reputation slightly faster as an incentive to self-abstain from prejudice. As a user preference, vote scores are not displayed on any post until you have voted on it.

Applicability across different exchanges likely will vary. I feel hidden voting is universally beneficial yet naked votes should remain the default option.

Moderators on this site have confirmed that they use the vote score to decide if they themselves will cast a vote (fallacy could be due to any of the above biases). As such vote scores are influencing site moderation to the detriment of objective post content evaluations.

Some moderators argue a vote score helps make a decision whether or not to close a question. This is another argument in favor of this feature because moderators will have an incentive to cast a vote - they want to see the score. Yet in either case, the option is ...an option. Mods can elect to vote in the clear. The moderator workload would also be greatly reduced. As a feature it addresses many SE questions, such as these linked below.

“Spam Voting” - Should we stop it?

Would voting for posts based on their score be considered fraud?

A better serial voting trap


Useful answers should go beyond "yes" or "no" but should have constructive content including ideas, pro's, and con's of:

  • Use cases

  • Definitions

  • Implementation

  • Reward

  • Related issues & considerations

  • 3
    hmm. Might actually worth trying out as a userscript – Journeyman Geek Oct 8 at 23:32
  • 2
    Less transparency leads to greater paranoia (speaking personally), I'd prefer to go another way. – 011358 smell Oct 8 at 23:33
  • The same paranoia exists in posters assuming voter bias. The votes are displayed - poster assumes voters will "bandwagon" them. (see my links, it's a real thing) Which group of paranoids should we cater to? – Vogon Poet Oct 8 at 23:35
  • 4
    There are two use cases here. Power users that look at any interesting question and then vote, and persons trying to find answers. Now for a user to find answers you generally order the results top to bottom. The user then votes on the answer that suits him, which is of course one of the top ones. But the top ones obviously have the highest score. So I can see this working for the first use case but certainly not for the latter. Maybe blind voting for a day and then resume as normal? – Maarten Bodewes Oct 9 at 0:00
  • I wouldn't go as far as "certainly not" for the latter. The ordinal position of the answer is very different from the vote score. The voter has much more subtle information to bias them. Without knowing the "spread" between two answers, they are more likely to be honest. A 20-point spread will be discouraging and may loose the vote altogether. Even possibly, since this is a user preference, the answer order is arbitrary for the answer seeker - who does very often disagree with popular opinion. I personally would leave this "always on." – Vogon Poet Oct 9 at 0:27
  • related – gnat Oct 9 at 6:24
  • 1
    Anecdotal evidence: after I wrote my answer, the other one received three upvotes, mine 0. Obviously that is because people are biased by the already high voting count for the first answer ;-) – GhostCat Oct 9 at 7:31
  • Honestly, ask yourself if you will easily up-vote a question with a -30 score - I don't go with the flow and I have my own mind. I vote as I please so, I can both downvote a +500 post or a upvote a -50 one without hesitation when I find they deserve it. – shiny-metal Oct 9 at 7:45
  • @shiny-metal The main problem I see with "herd" voting is: a newbie asks a not-ideal question. For this or that reason, he acquires -5 or worse in 10 minutes. Not because the question is really that bad, but people are jumping on the train, too. And the real problem is that even when the question is fixed 10 minutes later, the question still ends up with a (undeserved) negative score. – GhostCat Oct 9 at 8:01
  • 2
    @GhostCat There could also be the exact opposite effect. When I see a question at -5 in the question feed, it is very likely I will never visit that question. If it is at zero and it is a bad question, then I will visit it and downvote. So hiding the score will make it more likely for me to vote and it will cost me a lot more time because I would read tons of questions others have already marked as low-quality. – BDL Oct 9 at 8:23
  • 2
    @BDL Agree. That really depends on your motivation in that moment. I probably click on all negative scored questions (not to add the n-th downvote, but to put in a close or delete vote). On the other hand, I really look at most questions anyway ;-{ – GhostCat Oct 9 at 8:28
  • 1
    Not seeing vote count can be extremely negative for poor questions. If you are user with moderating privileges, votes are signal that question needs attention. If it is down voted, maybe it is close worthy. Also if I see poor question that cannot be salvaged, I will down vote it, but I will not pile up, if it already has few down votes, I will rarely need to add my own. If I would not see it is already down voted, I would down vote and such question can easily get more down votes than it needs (deserves). – Her Majesty Queen of ARC Oct 9 at 9:43
  • 1
    @VogonPoet You completely missed the point of my comment. I don't down vote or up vote based on other peoples votes. I only restrain myself from down voting question I feel deserves down vote, but already has few down votes. There is no need to down vote question into oblivion. – Her Majesty Queen of ARC Oct 9 at 14:04
  • 1
    @VogonPoet Are you even reading what I wrote. I am ALWAYS voting on the content (quality) of the post. The only thing I don't do is pile up down votes on post that already has too many, giving the chance to OP to improve. Since down votes count toward question ban, having too many for beginners can push them over the edge before they get the chance to learn how to ask. You insist on bandwagon voting... if I don't down vote on already down voted post, how is that bandwagon voting? – Her Majesty Queen of ARC Oct 9 at 15:08
  • 1
    I give up....... Life is too short... – Her Majesty Queen of ARC Oct 9 at 15:27
12

A very interesting idea, but it only addresses one specific use case, namely:

  1. user X "finding" question Y.
  2. And only to vote, X clicks on Y, and carefully digests all content whether "worth upvoting".

But that isn't the standard use case.

Most people come here because some search engine send them this way. They have a problem that maybe question Y addresses. And what happens then?

  1. You open the question
  2. You determine whether the question matches your problem
  3. You try to identify those answers you deem helpful to you
  4. And, maybe, then you vote.

But how do you identify what is most likely helpful, in case there are 5 or 10 or 20 answers? By first looking at the vote counts.

In other words: these questions are not polls!

They represent a knowledge base, and the entries in that knowledge base are intended to be ranked according to quality. Most users need to see the vote counts first, otherwise they are likely going to waste their time with low quality content you find at the bottom end of highly frequented questions.

Long story short: I do think there is sense in this idea, but in order to really "roll" such a feature, we would need to gather a lot of statistics upfront that tell us when and why people vote where.

  • 2
    It is my experience that the vast majority of users, who have earned the so-called ‘privilege’ of upvoting (a mere 15 points), do so sensibly. When an answer is highly upvoted, it is not the user behind the post who is being awarded it is the content of their answer/question that is endorsed by his/her/their community. – Mari-Lou A Oct 9 at 8:00
  • @GhostCat - I address the user case you mention by not making this mandatory. Blind votes are rewarded, not required. Your search user case is not impacted at all and this should be a non-issue. To the contrary, that new search becomes a new user interested in hastening their reputation climb, they are encouraged to self-restrain themselves from bias. They opt-in for blind voting and increase the objectivity (value) of their own votes. – Vogon Poet Oct 9 at 14:59
  • @VogonPoet My point: adding such a change would be a serious investment. It also adds complexity to the UI. As said: in order to justify such kind of change, you would have to provide a really big convincing set of data points. And even then it wouldn't happen. The community asked for many "really" significant changes many times, and few things ever happen. – GhostCat Oct 9 at 15:02
  • @VogonPoet No. Simply expectation management. One should carefully plan where to invest time and energy here. – GhostCat Oct 9 at 15:05
  • @VogonPoet 30 or 60 or 90 days of repeated frustration, because of banging heads against unmovable walls ... just saying: if you plan to long-term contribute here, it helps to avoid patterns that well, only create frustration. – GhostCat Oct 9 at 15:12
  • @GhostCat - "we would need to gather a lot of statistics upfront that tell us when and why people vote where" - This information already exists in the science of surveys. I don't think we need to or should re-invent the wheel. We are very simply surveying a group of experts for opinions on the fitment of a post. Voting is an exercise in social psychology. I have documented why this feature fits our strategic objectives, using that science as a reference. We can implement known best practices with what we have. – Vogon Poet Oct 9 at 16:00
11

If you prefer not to see post scores until you have voted, I've created a userscript for that.

HideVoteCounts

  • hides vote counts as soon as page is loaded
  • prevents fetching vote counts before voting
  • re-hides vote counts if you unvoted
  • always show vote counts on your posts
  • intentionally does not work in mod flag queue or review queues
  • if you really need to view a post's score, now you can double-click on the ?

Possible feature requests:

  • sort by newest by default?
  • sort answers randomly?
  • check if user has the privilege to up/down vote
  • If there is a place to make this script "beta tested" the experiment would give interesting insight! I will hold myself to a rule never to "unvote" based on the revealed score. Wonderful! – Vogon Poet Oct 9 at 2:55
  • @Vogon I just brought the possibility to get this experimented to a CM's attention, but it definitely needs more thought and work put into it – Samuel Liew Oct 9 at 7:59
  • I give up, how do I run this? I googled "Chrome Extensions" and "run file" from the developer console. I'm out of touch - javascript was so 1997 for me! – Vogon Poet Oct 9 at 19:53
  • I guess folks will ask you to create an opposite userscript soon – gnat Oct 9 at 20:54
  • @Vogon I have some userscript instructions here under the downloads header stackapps.com/q/7994 – Samuel Liew Oct 9 at 20:55
0

The idea is not foolproof (nothing is), however a productive user will tire of the steps needed to get the hidden voter benefit without being hidden (such as by searching while logged out, writing notes, then logging in). The implementation needs to ensure the time wasted circumventing it will not be worth the ROI.

Ultimately this solution in progress renders jumping in and out of blind fairly useless. As the user accumulates time in the blind the benefit increases.

Implementation:

Reward: The best reward is the status of being known as a self-disciplined objective user. Another consideration at this point to simply gain 1 reputation for each 3 blind vote credits.

  • Additional reward may be that blind votes only account for 3/4 of a naked vote toward daily total. Thus a blind voter may cast a theoretical max of 50 votes per day vice 40. Because they accumulate at the 3/4:1 rate, the daily count is still technically 40.

  • Alternately simply allow 10 blind votes beyond the 40.

Use cases:

  • Power users: These users benefit from hastening their decisions. None of the above biasing elements are available to distract them from the only useful voting criteria - content.

  • Answer seekers: These users benefit by the fact that the option is an option, not a standard practice. This doesn't impact non-voters in any way.

    that look at any interesting question and then vote, and persons trying to find answers

Definitions:

  • Blind voter: is any user who has their voting preference set to blind. User avatar and profile clearly delineate these users from the standard display. Time in the blind is also displayed graphically, implying more trustworthiness.

  • Naked post: Any post or portion of a post with the vote score displayed.

  • Established post: A post which has achieved a lower threshold in score variance (by linear regression) and has been viewed by a significant audience. New post scores tend to vary wildly.

  • Bracketed post: is any post displayed in a list by order of score where the scores above it and below it are naked, such that the blind score can easily be deduced. Bracketed posts are blind-ineligible. The blue question below is bracketed by scores of 48 and 39. It is blind-ineligible and the score would be displayed.

Bracketed post
(fig. 1)

  • Blind-eligible post: is any post which is eligible for blind voting by that user. This is a user-specific characteristic, no post is ever intrinsically blind-eligible. A running history of naked viewers with the last view time and the naked vote count follows each post to drive this decision; any user on this list will see this post blind-ineligible. Reinstatement removes users from the list. Whenever a naked score is displayed the associated post is blind-ineligible until the post meets the reinstatement criteria. Blind-eligible posts are clearly identified by the missing vote score, and also would be a slightly different shade, and sport a "blind" icon.

  • Vote Transparency: The ratio of naked to blind votes issued on a post, where 100% naked votes render a score transparent, and 100% blind votes render the score opaque. Vote transparency is not viewable, it is a fuel gauge graphic. E.g., an answer with 8 blind votes and 4 naked votes cast (up or down) will have a 66.7% full (opaque) fuel gauge. The fuel gauge is empty for posts with 100% naked votes. The concept graphically suggests that blind votes are a shield from bias which inspire confidence. (This could be better)

  • Blind vote: A blind vote is any up or down vote that is cast on a blind-eligible post by a blind user. It is tallied with naked votes on each post but also individually to calculate vote transparency. Search users see a single vote score but a fuel gauge graphically indicates the vote transparency. naked:blind vote ratio. (Rough graphic provided below)

Vote Transparency
(fig. 2)

A blind vote for the purposes of this feature has the following characteristics:

    • The voter has opted in to blind voting before viewing the post or the post score by any means (including in a search).

    • The vote has been cast on an eligible post.


  • Blind vote credit: The blind vote benefit locks in after the vote has remained unchanged while the post receives either 25 additional views, or the greater of 3 votes or a 25% increase in votes (not score).

  • Reinstatement: A post which is not blind eligible may become blind eligible by removing the user from the blind-ineligible list. This is called reinstatement. Only established posts can be reinstated. When the post vote count (not score) increases by n%, the post is reinstated. When the time since naked viewing (per user) reaches a set point, When the vote count changes by m%, the post is reinstated.

  • Grandfathered post: is any post which can be assumed to be viewed prior to blind implementation, because the system did not exist. Grandfathered posts are blind-ineligible until they meet reinstatement criteria.

  • Related issues and Considerations:

Un-voting - this is addressed in the definition of a blind vote credit

Post order by score: This remains unchanged, however the blind voter only sees the ordinal rank of the post (first, second, third, etc.) without knowing the scores or point spread between them, or even if any of the posts are actually "good"

Answer score bracketing: See fig. 1. If only one answer in a group is "blind" the post above and below will partially reveal a post score. Bracketed posts are ineligible for blind voting, in the definition. The score is displayed.

Vote transparency - This concept conveys how objective a post score is. A 100% opaque gives users confidence that every vote cast was protected from common biases. As a graphic gauge it doesn't reveal anything more than a ratio. The transparency number remains hidden to obstruct calculating votes. A score of 2 with a ~50% gauge can still be either 1/1 or 23/25 (naked/blind). The display votes privilege is protected.

New variables: This system adds variables to users, questions, answers, votes, profiles, and searches. It is far-reaching to do right however I believe improved objectivity will reap rewards in both site traffic and users.

  • 1
    I don't think adding a personal gain be granting reputation for voting will help your cause. This would only tingle the greed receptors in some users who then care less about the content and more about getting these 50 votes to get "teh repz". – Tom Oct 9 at 22:38
  • If they don't care about the content, then what exactly are they going to vote on? With no score to make their decisions they really have to look at the question, or randomly vote? Interesting points worth considering - what other incentives are possible? I do believe being known as a "blind voter" is itself a status symbol. Rep & votes were candy so I didn't sound stingy. – Vogon Poet Oct 9 at 22:40
  • Please advise if this post should be a wiki. – Vogon Poet Oct 9 at 22:54
  • 1
    Yes, that's what I mean, they don't vote based on the content anymore, they vote just to earn the reputation points. Reputation provides a certain attractiveness due to additional privileges or "fame" amongst other users, so granting it in an easy task that doesn't cause any backlash for yourself can be troublesome for the system itself. This greed factor even applies on rather meaningless stuff like badges, where people do bad stuff, just to get them easier, for example 'robo reviewers' (people who sprint through review queues just to accumulate "completed reviews" for badges). – Tom Oct 9 at 22:54
  • 1
    I agree the bonus blind votes were a better choice, reputation for votes should be out... later. Got chores – Vogon Poet Oct 9 at 22:59

You must log in to answer this question.

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