Jon Ericson asked, Does the tag synonym suggestion system work?
The answer, unfortunately, is no.
Tag Synonyms were introduced in August 2010:
It’s been clear for a very long time now that we needed some kind of tag synonym system, one that the community itself could maintain and control without having to bend the ear of a moderator.
The issues are:
- The community suggestions are taking forever to get approved
- A majority of the approved synonyms are moderator-handled
With a lot of help from Shog, I did some analysis of Tag Synonyms and how they get approved.
I took the 500 most recent tag synonym suggestions on SO and looked at what percent has been approved by moderators vs. the community (the people intended to maintain the synonym system per the blog post above).
Out of 500 proposed synonyms, 228 were pending, and 272 were approved (54.4% approval rate). Of the pending tags, the average time since the synonym suggestion was created is 159 days (median 157 days). This is only the 500 newest synonyms created within the past year (the oldest request was from Nov. 30th, 2012).
Of the 272 approved, 194 were created by moderators (someone with a diamond next to their name on the 'creator' field). That is 71.3% of the approved synonyms having been handled entirely by someone other than the community (a quick look at the tag-synonyms and synonym-request tags will show that this is a generally more reliable method of getting a tag synonym finished).
Shog also did his own analysis on the last year. 89 out of 372 synonyms were approved by mods. So I got 71.3%, he got 76.1% handled by mods rather than the community.
He also provided two data dumps of time before approval. He gave the number of days before a synonym got approved by the community, from the beginning of time, and just for the past 2 years. While the time to get a synonym approved was only 15 days from the beginning of time, over the past 2 years it has taken 46 days. The pending synonyms currently have an average age of over 150 days as stated above.
I do not want to make voting super easy, also I want to restrict voting to tag experts.
He has definitely succeeded in the first part, but unless the mods are omniscient (not saying they aren't), chances are that there could be some improvement in the second part.
First we need to make it easier for the experts to start voting on synonyms.
This is the current workflow for finding a tag synonym and voting on it.
Step 1: Browse to a Tag
Step 2: Click "Synonyms (3)"
Note: Despite it saying
Synonyms (3), that doesn't mean there are 3 pending synonyms. It means the tag has 3 synonyms, and doesn't even indicate whether one is pending or not.
Step 3: Vote on the Synonym
So our ambitious tag-synonymizers need to click on a link with no knowledge of whether or not a synonym request is actually pending, and then they have to actually have enough score in that tag to actually suggest the synonym. This means that there is a barrier to finding the suggested synonym even before the restriction on voting. Waffles has definitely succeeded in making it difficult to vote on these.
Handling of Synonyms
There are two different 'types' of synonym creation. For the sake of this example, let's talk about google-geochart being made a synonym of google-visualization.
If you look at google-visualization you will notice that google-geochart is a synonym now. But there are still questions tagged google-geochart, what gives?
Normal tag synonyms created by your average user will not retag existing questions. Searching for [google-geochart] as a tag will redirect you to [google-visualization], but the old tag will still remain as a marker until those posts are edited.
So the old tags sticks around as a marker, but there is no actual way to use the marker for anything of value. That means that making a synonym has the effect of removing a piece of useful information which may be one reason it isn't taken so lightly and is hard to implement for normal folks.
At the same time, tags can be merged by moderators, so that all questions with the old tag are automatically adjusted to be marked with the new tag instead. This also eliminates the additional information, but does it in one fell swoop.
Synonyms are supposed to aid classification but actually kind of haphazardly remove information. This means that if at some later point in time google-geochart actually reaches a ton of users who want additional guidance, the old questions with that tag will require intervention from the developers (most likely) to get that information back and allow the information to be indexed properly. This is less than ideal.
Tag synonyms should be shown to people who can vote on them via a notification. Something like this:
This could be shown to users who have the score, or to prevent spam for tags with lots of users with the required score, only users who have the tag in their favorites.
The notification could be made monthly, or weekly, or one-time, but users with interest in that tag should at least know that something is being proposed so that they can move the process along.
Fix the Display
As mentioned above, the Synonyms link when you are browsing a tag doesn't even show you that there are synonyms pending. Adding that information would make it easier for people to vote:
This link would be hidden if there are no suggestions, and would cause absolutely no change to the layout of the page.
Improve Synonym Suggestion
Allow users to explain the suggestion reasoning. Currently the dialog for suggesting a synonym is rather minimalist:
We could add space for people to add a comment, add information on the tag wiki (excerpt with a link to full description) for the suggested synonym, and put it all in the same place allowing users to more easily determine if they are related. Then when a user wants to find details they don't have to do the legwork themselves minimizing the burden on them. It also allows the person suggesting to make links to things that explain why they are synonyms. Win-win for everyone involved. And less burden on the moderators.
Remove Silly Limitations
Let's say I want to propose that c# is a synonym of c++ -- now I don't have reputation in either tag, but I could just go and retag one of my previous questions with c++ and then I can vote on it anyway. In times when I'd like to use this 'trick' it shouldn't be necessary, in times when I shouldn't use this trick, I can use it anyway. The limitation is silly as a result (it doesn't help good behavior or prevent bad).
Remove the score limitation for trusted users over a certain reputation (5k? 7.5k? 10k?) that allows them to make synonym suggestions (but not vote on them without the requisite score in that tag). This will allow good-natured users who are driven by a categorical imperative to actually start suggesting synonyms for others to determine the value of (much like, say, suggested edits, or flagging to close).
Synonyms as a Temporary non-destructive Categorization Method
If Synonyms were easier to 'undo' as it were, we would be able to be a bit more loose with them since we wouldn't lose the ability to categorize more specifically, while still having all the questions in the larger group easily accessible. Tag Hierarchy/Trees have been discussed in the past, but this would be a far simpler way to handle it, which would allow synonyms to put together useful information (for instance, there are 13 tags related to google-maps), while not barring someone from searching for a specific subset (say google-maps-api-3, or from splitting that off later if it needs to be.
The first 2 suggestions are similar to this feature-request: Tag synonyms: show suggestions to people who can vote
The third suggestion was taken from my answer to Jon Ericson's question.
The fourth suggestion is taken from Can we allow 7.5K users to suggest tag synonyms without a score of 5 in the tag?
These suggestions are not mutually exclusive, nor are they exhaustive. Tag synonyms are great for grouping questions on related subjects together, especially when the small niche topics get little visibility on their own.
This isn't really a critical issue, it is more of an 'it'd be nice' sort of thing. And it'd save the following mods from the tireless work they do making SO a tidier place:
- Animuson (67 synonyms)
- Robert Harvey (48 synonyms)
- minitech (16 synonyms)
- BoltClock (15 synonyms)
Note: The above numbers are for SO only for the last 500 suggested synonyms -- these few, these proud, these moderators have approved far far more on other sites (including meta) and over a longer period I'm sure