What is meta? ×
Meta Stack Exchange is where users like you discuss bugs, features, and support issues that affect the software powering all 134 Stack Exchange communities.

Summary

Instead of immediately bumping a question, questions that qualify for special bumping behaviour shall be placed in a queue and sequentially auto-bumped by the system at a rate proportional to the site's current activity level.

This system satisfies all the given requirements:

  • The homepage will not be flooded with old retagged questions
  • All edits will still have the opportunity for peer review (although not necessarily immediately)
  • A retagger can now work at maximum efficiency at any time of the day
  • There would be no GUI additions; this is purely a back-end change
  • The system is well-defined, and is (what I perceive as) relatively easy to implement

and, most importantly,

  • Makes everyone involved happy! Except possibly Jeff Atwood. :D

Based on this question, several concerns were raised about the retagging that goes on, particularly on Stack Overflow. As I mentioned in that question, the current system does not allow those of us who want to organize to do our job as effectively as we could. I am proposing a system that allows us to be more productive members of the community while allowing full transparency of our actions.

Requirements

  • The homepage must be usable 24/7 (i.e., flooding of old questions should be minimized)
  • All edits should have the opportunity for peer review (as it is now), which minimizes the possibility of abuse of this type of system
  • Improve the editing efficiency of a retagger, particularly when the site is in a period of low activity (i.e., minimize self-edit-throttling)
  • Very lightweight as far as usability/GUI goes
  • As simple to implement as possible, so it gets finished sooner than 6-8 weeks from now

The System

For questions, edits and editors that meet a certain criteria (see below), the bumping behaviour of questions shall be modified such that instead of bumping immediately, a question shall be put into a queue (the Bump Queue), and at periodic intervals (the site's current Bump Rate), a question will be removed from the front of the Queue and bumped automatically.

Questions shall only be placed in the Bump Queue if all of the following criteria are met:

  • The edit is done on a question
  • The question has not been edited (including queued edits) within the last 7 days
  • The edit only affects the question's tags
  • The edit does not create any new tags
  • The editor has at least 3,000 reputation

Should these criteria not be met, the question shall be bumped immediately as is currently implemented.

If a question currently in the Bump Queue is edited again, the question shall be moved to the end of the Queue if the above criteria are met; or, if the criteria are not met, the question shall be bumped immediately.

The average Bump Rate of the Queue shall be tuned through community consensus on a per-site basis, and vary in real time proportional to the site's current activity rate (# questions asked + # of immediate-bump edits per unit of time).

For Stack Overflow in particular, I suggest an average Bump Rate of approximately 1 bump every 5 minutes, which allows for a daily bumping "capacity" of 288 questions (plenty), while maintaining an acceptable level of flood control (at peak times, the homepage would only have perhaps 5-6 old questions visible at once).

share|improve this question
5  
+1 for the small text at the top alone :) –  Billy ONeal Apr 15 '10 at 14:18
2  
it does seem rather complicated, but i like it. i'd be concerned about the queue never managing to empty itself, tho. –  quack quixote Apr 15 '10 at 14:54
    
@quack: As I said, the Bump Rate would be reached through community consensus. Considering the toleration of current retagging sprees, a rate of much higher than 1 bump in 5 minutes could be tolerated for periods of time. How many manual retags are going to happen in one day, every day? The Bump Rate could also be tied to the length of the queue, but I don't think that's necessary. –  Jon Seigel Apr 15 '10 at 15:25
    
How much retagging are we talking about? Can you quantify? Are you retagging 50 questions at a time? 10? 5? –  Jeff Atwood Apr 15 '10 at 17:21
    
Depends on the request. Sometimes you have to go in manually and make sure things check out. This one I did by hand was nearly 200 questions, but that's an extreme. Most of the time it's fairly straightforward, post a retag-request on here, but get impatient since a mod has to do the internal merge thing. I think this issue mostly has to do with the density of edits, not so much with how many are being done. Even with the captcha, I can plow through at a rate such that even during a period of high activity, it's clearly spamming the homepage. –  Jon Seigel Apr 15 '10 at 17:35
    
To answer your question directly, most of the ones I have on my to-do list at home are in the 30-70 question range. –  Jon Seigel Apr 15 '10 at 17:36
    
@jon if it's 200, that's wrong -- it should be handled by a moderator. We have tools to merge thousands of tags at once with no change to history or bumping. –  Jeff Atwood Apr 15 '10 at 17:37
    
Sorry, I should have given more context -- that one in particular needed to be hand sorted (at least in part) because no automated tool could separate out the questions into the proper tags. It wasn't a simple case. –  Jon Seigel Apr 15 '10 at 17:40
    
the mod tool can only merge one tag into another, or create one tag from another. it can't split one tag into two, perform conditional operations, add a tag, or handle anything more complicated than a 1:1 change. manually retagging 30-200 questions doesn't sound horribly off-base in my experience. –  quack quixote Apr 15 '10 at 23:31
    
Can someone run a query to try to predict how long things would stay in the queue? I retag for many reasons, but one is "dearie me, no-one who can answer this is going to see it with those tags" and I therefore want retagging to bump it. If it might take an hour or two that's fine by me, but three days? I'll do a content edit at the same time to get it some eyeballs. –  Kate Gregory Aug 22 '11 at 17:34

3 Answers 3

Seems a bit complicated. My solution:

  • give each question a "retagged" date that is set when a retagging (as opposed to an edit) takes place on a question more than a week (or whatever) old

  • provide a user preference tickbox "hide retags" which removes the recently retagged from the front page

share|improve this answer
3  
I think in practice, the second point would eliminate the aspect of peer review (which was a big sticking point in the other question). Most people, myself included, would choose to hide the retags by default because the system should be usable at all times for everyone. –  Jon Seigel Apr 15 '10 at 14:41
1  
@Jon But I never peer review retags, so why should I have to see them? And if even a retagger like yourself admits that they would like to hide them, how much peer review actually takes place? My guess is very, very little. –  nb69307 Apr 15 '10 at 15:01
    
The idea is to have the opportunity to peer-review all edits. I wouldn't hide the retags because I don't want to peer-review; I would hide them because of the flood control issue. A real-world example of this not working is exactly what happened in the other question: if your preference had been to hide the retags, you wouldn't have seen the question I edited, and never raised the edit as a problem (which, as I said before, was a valid concern). –  Jon Seigel Apr 15 '10 at 15:22
    
@neil: i review tag edits all the time. (and frequently curse users who insist on making an edit just to add a [windows] tag on a question that's already tagged with some version-specific tag.) –  quack quixote Apr 15 '10 at 16:18
    
@quack Fine - my solution allows you to carry on reviewing, but lets the majority of people who have zero interest in retagging ignore them. –  nb69307 Apr 15 '10 at 16:24
    
+1 Being a user who is not very interested in retagging (except in his direct domain), I find this a nice idea. –  Pëkka Apr 15 '10 at 18:21
1  
+1, but my guess is this won't get implemented because (I think) the home page is denormalized/cached, so it would require some pretty big changes to implement this as a user preference and show different results for everybody who opts out. That is unless it works like the "ignored tags" and doesn't actually remove them, just dims them, but that wouldn't be all that helpful. –  Aarobot Apr 15 '10 at 18:41
    
@Aarobot Actually the ignored tags are removed - it's a preference! But only after the cached page has been processed. Still, better than nothing.... –  nb69307 Apr 15 '10 at 18:46
1  
I didn't even know that there was a preferences page until right this minute. I suppose this could work like the ignored tags, then, although if somebody goes on a retagging spree than your front page is going to have 12 questions on it... –  Aarobot Apr 15 '10 at 19:07

How about only bumping the question when it is edited by it's original author or if the body of the question is edited? Seems much simpler. Essentially what this says is that we only want questions to get fresh attention if there is fresh information. If we're simply rearranging the deck chairs, we don't care. Granted, edits may simply be to improve the formatting, but I'm willing to tolerate some slop in the method in exchange for simplicity.

share|improve this answer
    
Would this behaviour only apply in certain situations? (i.e., old question, not a new user, etc.) –  Jon Seigel Apr 15 '10 at 16:07
1  
@tvanfosson: one of the major points that came out of the previous discussion was that tag edits should be peer reviewed, like other edits. the fact that they bump the edited question is a good thing. thus the suggestion under discussion here. –  quack quixote Apr 15 '10 at 16:14
    
@quack In the previous discussion, a very few people seemed to be of the opinion that retagging needed peer review. –  nb69307 Apr 15 '10 at 16:26
1  
@quack - IMO tags are mainly of interest to classifiers, not answerers per se, i.e., they're helpful but not the main focus. Questions should be bumped when there's new information, not necessarily new classification. My suspicion is that questions don't get answered, not because they're not tagged properly, but because they are (1) poor, (2) hard, or (3) an extreme niche (i.e., uninteresting) -- my bet is on (1) being the most common reason. In that sense, bumping a question that has no or few answers based on a retag will just bring a poor question back to the front page. –  tvanfosson Apr 15 '10 at 17:30

I have to disagree (but that system is a nice thought), I simply don't see the need. It feels a little bit blown for something that already works.

Mass retaggings by a user which would make this system necessary are highly discouraged, by the way.

share|improve this answer

You must log in to answer this question.

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