Meta Stack Exchange is where users like you discuss bugs, features, and support issues that affect the software powering all 158 Stack Exchange communities.

What is meta?
Here's how it works:
  1. Any Stack Exchange user can ask a question
  2. The community provides support, votes on ideas, and reports bugs
  3. Your voice helps shape the way Stack Exchange operates

After a lengthy discussion about the false positives the current system throws up, it was suggested that we should open a debate about whether or not the current system would benefit from being more open.

In my opinion, it would be valuable if the community had more access to how the good/bad filter was doing its job, so as to use the wisdom of crowds to improve upon it.

No one wants spam, but we also don't want to have to rewrite perfectly good questions to please an invisible overlord.

I don't think the algorithm should be made explicit, but to understand the kind of criteria used in its creation would be enough, I think, to benefit from the SE hive mind.

share|improve this question
Nope. I'm happy as it is. – Thursagen Jul 5 '11 at 12:30
up vote 18 down vote accepted

Two relevant points:

  1. We have more questions than we need. We can afford to be picky. Further, we can afford to throw away good questions that don't meet an arbitrary standard - similar to a restaurant that requires a tie. Sure, the tie doesn't mean anything more than some arbitrary requirement, however since the restaurant is booked solid, we aren't going to be hurt by this requirement. Remember, the person asking the question is requesting help - it's ok to make them work a little harder to get that help, especially if they are asking a questions which trip the rather trivial quality filter.

  2. We are not interested in giving users the tools that will help them get as close to the line without crossing over. If a user is asking, "How close can I get to the edge of the cliff without falling off", they are asking the wrong question. They should instead be asking, "What are the guidelines that will help me stay as far away from that huge gray area as possible, nevermind getting close to the edge" - and those guidelines are already presented to the users that demonstrate a need for them.

My experience with releasing details of the filters is as follows:

We have a "subjective question" alert, which is triggered by relatively simple word usage. When people hit it and asked a question on meta, we used to say, "It's probably due to this word, but here are the problems with your question title - try to make it less subjective.". Once Jeff released the actual regex for it, the discussions instead tended toward why or why not those words should be in the filter, rather than why a particular question was subjective or not.

In other words, releasing the filter caused the community to focus on the filter's faults, rather than causing the community to improve the questions.

As programmers, we can always find some fault with a heuristic.

But it's the wrong thing to focus on.

share|improve this answer

I think you're the only one who's ever found a false positive, if you indeed have. Disclosing the metrics would only encourage people to game the system by figuring out ways to work around them. The point is to prevent bad questions from entering our site. Thus, your suggestion is counter-productive.

As Bill implied in an answer to your previous question the standards that we have now represent an absolute minimum quality bar. If your questions are not meeting that minimum standard, we're not particularly interested in accepting them.

We have already documented some tried-and-true methods of improving your question here, and several other places around Meta. I see no compelling reason to waste any more time on this.

You're right in observing that you shouldn't be able to split the question up into multiple paragraphs and bypass the quality filter. But this kind of thing isn't worth optimizing. It's keeping out the worst of the worst, and that's all that it is really intended to do. If someone happens to skate by it by splitting a bad single-paragraph question up into multiple paragraphs, you can rest assured that the community will quickly close that question and all will be well with the world again.

Also, perhaps you missed it, but it looks like we've already fixed that particular bug. Whitespace is now stripped out of questions before they are run through the quality filter. Ahh, the power of regex.

share|improve this answer

There is no need to make the algoritm public; there are users who see which posts get automatically flag as low quality, and they can suggest which posts should not be considered as low quality, if they notice a recurrent pattern.

share|improve this answer

In my opinion, it would be valuable if the community had more access to how the good/bad filter was doing its job, so as to use the wisdom of crowds to improve upon it.

I agree completely!

But there is a problem.

We don't want users being able to game the system.

Why is that a problem? With this type of system, there are two components: the algorithm, and the underlying data. If someone has access to both components, the system can be gamed. So either we have to restrict access to our algorithm, or to our data, or to both. Our data, through the public data dumps, is (mostly) available to anyone already, and we can't exactly "take it back." Therefore, we have to restrict the availability of the algorithm.

share|improve this answer

You must log in to answer this question.

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