What is the goal of the low quality question filter?

Question

For a while now we have had a filter for new questions that doesn't allow them to be posted until they pass a filter that checks for certain mistakes indicative of very low quality questions. Recently this filter has been altered to be more strict.

What is the goal of this filter? When users end up hitting this filter what are we expecting them to be doing?

For a small percentage of users they will be capable of just spending more time and effort on the question to fix enough of the issues to pass the filter. Personally I imagine that, by and large, if they were capable of writing a question good enough to pass the filter it's unlikely they would have hit it to begin with, so most of these users won't be capable of editing the question into appropriate shape on their own. Do others feel that a significant percentage of users will be able to edit the question into appropriate shape on their own. Note that some users in this position may find a way of bypassing the filter through some transformation without truly improving the question enough to be of even mediocre quality.

Some users will give up and not ask the question at all, since they cannot get past the filter on their own. Is this desirable? Are we assuming that these questions are of such low quality that they're better off being lost entirely? Do we really want these questions to just not be asked if they can't meet the filter, or is this case an unfortunate side effect of adding the filter?

Some users will seek help from outside the site, asking a friend, teacher, going to some other site, etc. to improve the grammar and other problems allowing them to post the question here without being blocked by the filter. Do we expect this to account for a noticeable percentage of users? Yes, if this happened it would be nice, as it's taking the burden of improving the lowest quality of questions off of our users, but can we expect a significant percentage of users to go through these steps?

Finally, other users will post on meta saying that they couldn't get past the filter and asking the meta community to improve their question. Is this desirable? Is this preferable to having these questions just not be blocked in the first place and just be posted on the main site and be edited into shape (or to have comments indicating how the OP should improve the question)? If this is preferable, how so; what is gained from such a meta question that wouldn't take place if the filter didn't exist?

Answer 1

I think the goal of the low quality filter is to:

• Blocks garbage, i.e. nonsensical text. We don't waste time on people who don't need to ask question.
• Blocks poor writing style, so as not to waste people's time to try to understand the post and edit it into shape. (e.g. i haz this prlbem i want code to do this and that and a long wall of text without punctuation plzz itz urgent thx)
• Instruct people to format their code. (Or include their code into the question, in the case of jsfiddle)

Beyond this, I think it is outside the scope of the low quality filter, and into the community's (brutal) judgment. That is unless SE comes up with a filter that can analyze the meaning (semantic layer) of a post.

The role of the low quality filter is to reduce the community effort on tasks that can be automatically suggested to the poster. The automatic feedback needs improvement though; it is quite user-unfriendly, and does not suggest user what to do (I think give an extract of the FAQs is more user-friendly than just a red error message - people don't read FAQs).

I think it is OK to instruct user when they come to meta and ask for help with posting. If their post is craptastic or off-topic, we can save them from the shock of downvote and closure. If their content is good, but the presentation is a bit rusty, we can help them polish it a bit. I think it should be fine, as long as meta is not filled with these kind of requests everyday.

Answer 2

The goal of this filter is to provide some automation of the winnowing of questions that have zero chance of attracting even a semi-intelligent answer.

If we are to require nothing for someone to post other than being able to fill a web form - then there will be thousands of people that take advantage of that low bar to post things.

I think we can all agree that we are all for a filter that prevents posts like this (with the title being the text before the - and the body after the -):

• Q: Poop - What is is good for?
• Q: iOS - I want to make an app - what do I do? HELP PLASE NOW
• Q: EMACS or vi - Which is best for hadoop and go lang for making bespoke apps to attract VC dough?

The problem is that we don't have the data of all the thousands (millions) of failed attempts that didn't pass the filter, so most of us don't know how the filter works, what it does, what it's saving us from, what false positives are being thrown out with the poop. Since people are smart, those wanting to poop get better at pooping (especially since the filter is easily tested) and it's now time to implement some different approaches (or modified quality levels) to stem a very real quantity of very low quality information.

By saying no to some, we are saving more attention for measurably higher quality content.

I don't like filters, especially ones I didn't design, but I don't see a better option on the horizon than experimenting with tools to make the site better. From my interactions with the people that run the site, I would generalize that they are incredibly smart, aware of how their actions affect thousands of people daily, have the data to back up their decisions long before they make a code change and set up monitoring so that as the tool they designed interacts with the real world, they can and do check up on it to make sure it's actually working as intended.

On the flip side, there are some very terse questions that have provided some very awesome content for the site. Specifically, look over your site's greatest hits.

• https://stackoverflow.com/questions/greatest-hits

This list is filled with short, useful, detailed questions on all the original trilogy sites and whatever changes are being made will likely need to not have a high false-positive rate when it's fed a list of popular questions.

• What is the difference between "INNER JOIN" and "OUTER JOIN"?
• How to redirect to another webpage in JavaScript/jQuery?
• How do I install the Sun/Oracle Java SDK on Ubuntu?
• How do I reduce the desktop icon size in Windows 7?
• What port does SFTP use?
• Ping a Specific Port
• How to check if a port is blocked on a Windows machine?

I would expect that as long as the filter doesn't shut down questions of similar quality to these examples above, no one would object to the system being overly restrictive or something that a new user would be incapable of rising to the occasion to rephrase their question to pass the entrance exam for a new question to be submitted.

Do you have some specific examples of questions you could not ask due to the new filter or are you looking for a discussion to share how they're doing it (it's robots and neural learning patterns, big data analytics and VC dough BTW)? I've not seen a dramatic down turn in the availability of high quality questions due to the filter, so I'm not sure we've turned the filter up high enough yet, personally.