User Experience Stack Exchange is a question and answer site for user experience researchers and experts. It's 100% free, no registration required.

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

Suppose there is site with an upvote/downvote (like SE) feature, or a "liking" feature (like Facebook). A user posts something to this site and other users can cast their vote on it. Now suppose the poster is allowed to edit this post at any time, should this affect the votes that were cast on the original text?

I can see in some scenario's leaving the votes/likes as is could be problematic, the user could post:

Rand Paul for president

And get thousands of likes, potentially with the upvoters name associated with it, but then edit the post an hour later to:

Trump for president

And now to anyone who views the post all of the upvoters appear as Trump supporters.

However, not leaving the votes as is, for example resetting the votes upon edits, could also be problematic. Suppose the original poster just realized there was a spelling error and fixed it, that could cost them thousands of "likes".

Should votes/likes be left alone after an edit is made?

share|improve this question
I would suggest to inform the people who voted or liked with a notification and leave it to them to unlike or downvote again. – Yakke Feb 2 at 18:18
This question can't be answered as it depends too much on the context. It is too broad and/or asking for opinions. Can you add some context or ask for resources like best practices? – jazZRo Feb 2 at 18:34
@Yakke, I would hazard against notifying users. It may protect you from edge cases where the OP is trying to be misleading or trolling, but it degrades the UX in that you're now notifying potentially hundreds/thousands of users that something they looked at hours, days, months ago was corrected for spelling? – Daniel Brown Feb 2 at 18:45
This is actually an issue that Facebook hasn't addressed yet. For example, you will see a page promoting a contest sponsored by a famous brand, like Audi, to win a free car. A few seconds of close inspection will reveal that the page is probably fake and in no way affiliated with the actual brand. After they reach a certain amount of likes, the page will be sold off to a company/band/group/person/etc. who will then change the name and content. All likes, shares, followers and such are kept. This is essentially buying/selling likes. – Ivo Coumans Feb 3 at 12:58
@ChrisH, there are certainly ways to improve how and when you notify users of changes. What has been suggested sounds a little like an "auto-mod" or criteria that would trigger some red flags and ask humans to verify that the content still holds its original intent. To some degree, Stack Overflow uses its "moderation queues" in this way. As others have pointed out, the biggest factor is what kind of community you have. I would argue that a lightweight MVP is what I've described in my answer; from there you can grow the solution appropriately. – Daniel Brown Feb 3 at 13:20
up vote 42 down vote accepted

Most websites that allow editing also do the following:

  • When a post is edited, add (edited x <timespan> ago) to the end of the post.
  • An option to view past version(s) of the post is made available
    • A "history" link, or making (edited x <timespan> ago) clickable
  • The ability to "flag" posts as inappropriate/dishonest

An approach like this, while requiring more work, allows you to avoid a lot of potential UX pitfalls. As you've pointed out, users may have very legitimate reasons for editing their post where votes should not be removed. Also, most users will likely get annoyed if they're constantly notified of small spelling corrections.

share|improve this answer
I think this is the way I'll have to go, perhaps including number of votes per different iteration in the history view so people can see that most people voted for the first version before it was edited, etc. – DasBeasto Feb 3 at 17:36
Also, StackOverflow has implemented review queues, for when an item is edited it can be reviewed and accepted/rejected or improved. – Ismael Miguel Feb 4 at 10:02
@IsmaelMiguel: Only when the edit is made by a low-rep user. There's really nothing stopping someone with more than, what is it, 2k? sneaking in a "fraudulent" edit. – Lightness Races in Orbit Feb 4 at 11:02
@LightnessRacesinOrbit I am aware of that... – Ismael Miguel Feb 4 at 11:11
@LightnessRacesinOrbit Well, there are! You can stop those fraudulent behaviors by having moderators! – Thales Pereira Feb 4 at 16:31

The absolute vote count should always be left alone. Reddit's masking of the true up/down vote count was a controversial matter due to the lack of transparency around how much activity constituted 'hot'.

I proposed (elsewhere) that the upvote counter should have a detail-panel (on hover) that depicts the percent of current content that was voted on.

Thus, consider when 100 votes occurred, but 20 were after an edit. The edit was inconsequential and changed 4% of the characters in the post. 80% of the people agree with 96% of the post. 20% of the people agree with 100% of the post.

I haven't decided on a visualization for it yet.

A diagram like this shows these percentages, though requires a certain visual literacy.

enter image description here

Agreeing with @Deduplicator, I have considered that this visualization is inherently opaque to semantic value. However it does 'dimensionalize' sentiment (up/down) over time and edit-distance. That was an intention of mine, to implicate when semantic value may exist, by correlating 'sentiment surges' with inflection-points, i.e. ƒ", the second-order derivative.

Compare this to Apple's choice for the same, in the App Store. Score is (was?) calculated based on reviews left on the current version of the app. This renders the sentiment caused by the changes.

A similar strategy would be to weight old-version votes differently than current-version votes. Unfortunately, the weighting value is editorialization, and not an honest representation of the total opinion – which can only be described as polyphasic once votes arrived after the original post was edited.

share|improve this answer
For a more technical user base, I think this would be a cool way to attack the problem. It may be a little on the complicated side for most to understand at a glance. – Daniel Brown Feb 2 at 19:40
@DanielBrown I think the idea is implicitly complicated, and is best-suited for a user base that can initially handle this kind of complexity. Once they have 'alpha tested' it, what will emerge is language and simplifications that work for less-technical, so-called-naive, users. Then the idea can be trialed among them. – New Alexandria Feb 2 at 19:44
Well, the trouble with this is that edit-distance, or whatever metric you use for discounting pre-edit votes for the post-edit content, is inherently not a good metric. You need a human to read and understand both texts, and even then the inherent looseness of natural language might make that evaluation pretty much useless. Still, a nice chart. – Deduplicator Feb 3 at 14:54
+1 - This is a very cool idea but may be more work than it is worth, I imagine only a certain niche would care about the upvote history to this degree of complexity. Still, it is an excellent way to display that data if this route were taken. – DasBeasto Feb 3 at 17:39
Thanks @Deduplicator, I updated the answer ^ – New Alexandria Feb 4 at 4:13

In most cases I know, changing the upvotes or likes to an item seems not to be the solution to your problem. It rather causes confusion and probably frustration. In general, you always have to regard the intention of the upvote counter, if there is a public edit history and/or if votes are directly linked to a specific version of the upvoted item.

E.g. if the intention of the upvote counter is to inform people about how many users agree with the statement, there is no public edit history, the upvoter names can be seen and you look for an automatic mechanism to handle this, then you now have the following options:

  • Remove the likes. Disadvantages: the author won't edit minor changes. If he does, he will be frustrated about the lost likes. You only want to remove the likes if the content changes to the contrary, like in your president example. So if this is an exceptional case, think of a manual moderation of such items.
  • Keep the likes and inform the voters. Disadvantages: active users will get a lot notifications. It depends on the kind of content the application is about.
  • Improve your concept. For example show that this item has been edited, provide a public edit history, link likes to a specific version of the item (see play store ratings). In my opinion, this is the way to go rather than spamming notifications or removing likes.

As pointed out in the comments this question is a bit too broad. To give a good solution, you have to regard the context. To stay with your example of StackExchange sites: Here the intention of the upvote count is to show, how many people find the answer useful or agree with it. Votes are private so it isn't as critical as they weren't (e.g. in Facebook). Report the post if you see such an edit (in the public edit history) and let the moderators handle it, e.g. by contacting the author or reverting the changes (again, it depends on the context). If you see a strange like of your friend on Facebook, check the edit history and contact your friend if it is critical. This should normally be enough here; no automatic mechanisms required for these special cases.

share|improve this answer

This is a bit of a made-up question. Who asks this sort of stuff?

Who should be president?

However on this site at least, I think if you had an answer which got a lot of up-votes, and subsequently the answer was changed to something completely different then someone could flag it for moderator attention.

The moderator could then revert the answer, and then lock the question to prevent it happening again.

It's an interesting question. Edits are supposed to be for clarification, not for gaming the system.

share|improve this answer
Someone with (I hope) a sense of humour edited my answer and reversed the meaning of it. :P However it was flagged and a moderator rejected the edit. – Nick Gammon Feb 4 at 4:23

Guess I would allow adding extra comment vs full edit after some considerable time passed.

[original post]

Update: [comment] Update 2: [comment 2]

Did you research how Quora and Stackoverflow here do it?

share|improve this answer
Very fair suggestion. However, I assume they already did research SE or other sites, and anyway, the question is about a hypothetical other site and how it could best reduce such issues. – underscore_d Feb 4 at 13:45

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.