Should we require minimum reputation to continue asking questions?

Question

We are reevaluating this based on community feedback. The new heuristic to prevent questions being asked will probably be based on a combination of question closes, question deletes, question flags, and lack of significant upvoting, even at very low levels (0.20 etc) -- scoped to new-ish users.

Based on the discussion at

Should we cap reputation gained from questions at +2000?

We decided to go in a different direction. This is the new proposal.

1. After your first 10 questions, it will require 100 more reputation to ask every additional 10 questions. If you do not meet this reputation level, you will be unable to ask any more questions.

2. Bounties you have started will not be checked in step 1, so you can safely start a bounty without losing the ability to ask more questions.

3. If the users' historical questions did not meet the required reputation bar at the time they were asked, they will be anonymized and ownership from those questions removed. Yes, we will be rewriting history. (existing answers to those questions will not be affected in any way)

The net effect of this change is that we want to prevent people from asking dozens, hundreds, or thousands of mediocre questions (that is, they don't get upvoted). You will need to ask questions that get a modicum of upvotes -- or earn rep in some other way by answering questions or editing.

Examples of users who'll be affected initially on Stack Overflow

It can be difficult to grasp the effects of a change like this without taking a look at the sorts of users - and questions - it will touch. So here are a few users, arbitrarily picked from the thousands who would be blocked from asking further questions on Stack Overflow:

• ant2009 - 348 questions
• HollerTrain 187 questions
• Jitendra Jadav 83 questions
• user1111111 78 questions
• 5416339 57 questions
• StevieB 55 questions
• LukePet 30 questions
• HaterTot 11 questions
• Firefox 11 questions
• MangoTable 11 questions

Answer 1

First, I hate the thought of Stack Overflow becoming a Pay to Play system. Requiring people to feed into the system in order to use it for its primary purpose is exactly that.

In the 80's and 90's such BBS systems were called Elite - you had to submit some files to keep your account balanced enough to download files. While this isn't, strictly speaking, a form of elitism, it surely smacks of one.

Second, it's trivially easy to work around. Clear your cookies, clear your data cache, don't register, use a bogus email address, and you can ask as many low quality questions as you please. It will only hurt people who are earnestly trying to play within the bounds of the system.

Third, what the what? This proposal and the 2k rep cap proposal are fundamentally solving two different problems. If someone is getting too much reputation from questions, they will not run into problems with this proposal. If someone is submitting poor questions and not getting enough upvotes to continue asking questions, they won't run into the 2k rep proposal.

It sounds like you are either confusing the issues, or you still have a deep misunderstanding of what the root cause is.

For these reasons I strongly suggest you go back to the drawing board.

---

I'd like to see some data on how many users this would effect. In other words, find those users that have more questions than they would have been able to ask if the proposed system were in place. Then let's take a look and see if there's really a correlation that suggests those users shouldn't have asked the additional questions.

As an interesting data point, this system requires at least one upvote per question in order to ask additional questions. This assumes that EVERY reasonable question gets at least one upvote. As I describe in comments below, this doesn't seem to be the case.

Poking through a few of the high users I find several where 20-100% of their questions remain at 0 votes. Here is the list, the number afterward is the percentage of their questions which have 0 upvotes:

https://stackoverflow.com/users/29407/darin-dimitrov 20%

https://stackoverflow.com/users/50079/jon 25%

https://stackoverflow.com/users/34397/slaks 25%

https://stackoverflow.com/users/573261/richard-aka-cyberkiwi 33%

https://stackoverflow.com/users/413501/ladislav-mrnka 33%

https://stackoverflow.com/users/329769/brokenglass 100%

https://stackoverflow.com/users/328445/matthew-willis 30%

They are posting questions that are most assuredly not poor quality. Often, they are hard questions. In a way, this system doesn't merely penalize those who ask poor quality questions, but also those who are new to stackoverflow, are already experts, and are asking difficult questions.

I think we at least need more evidence to show that this proposal will actually address the problem, because at the moment I just don't see it doing what it's meant to do, and it appears that it could harm some aspects of the system that would be very damaging. Can we afford to turn away one new expert if it gets rid of a hundred low quality posts?

Answer 2

My big concern would be that it's actually the mediocre questions (being popular) that get the most votes, so this might not actually solve the problem. The best questions (being the most specific) often get the fewest votes -- not to be confused with the worst questions, which get negative votes.

Maybe I'm a bad example as I only have a few questions on SO, but it seems like it's hard to get 2 votes on every question. Does anybody have stats on how many people average 2 or more votes per question?

Also, I'm wondering how this policy would affect accounts like https://stackoverflow.com/users/39677/blankman, https://stackoverflow.com/users/104015/shore, and https://stackoverflow.com/users/43907/gold: how many of their questions would become CW?

I would actually propose that rather than preventing users from asking new questions when their old ones are crappy, require their questions to be approved like edits.

Answer 3

To know how valuable such a proposal is we really need to know how big the problem is. How many single users are starting accounts and blitzing out crappy questions?

I'll tell you what I see. My experience is twofold.

StackOverflow

I'm "just a user" on SO. ~20k points from answering things and a few good questions.

On StackOverflow the pace is very fast. Things fly by and questions, even good ones tend to get quite low votes. Only the "interesting" subjects really get attention but just because something isn't interesting to everybody certainly doesn't make it a bad question.

Under your rules somebody would need an average of +2 on their questions to get the points (slightly less with accepted-answer points). To me that seems quite high, especially for less-popular topics or questions that don't get picked up by the masses.

Ask Ubuntu

I'm a moderator on Ask Ubuntu. I've also answered a stupid number of questions there.

On Ask Ubuntu things are quite a bit slower but we do get our fair share of people pushing silly questions that are either duplicate, should be filed as bugs or are awfully written. However what I do notice is that these sorts of problems do not tend to all come from the same person. People hop in when they need something fixing and they don't tend to pebble-dash the whole site with nonsense.

Achieving a +2 average for "good" questions is easier to achieve but what's the point when I don't think we even suffer the problem you're trying to solve?

Joint issues

A limit may hurt legitimate questioners who just can't make something sexy enough to harness the attention needed for votes. Any solution to a problem that might not even exist needs to avoid punishing good, unlucky users.

You may say that a viable way to garner points is by answering something once in a while. For super-experts like me (ha!) that's not an issue -- I'm happy to grind my way through questions looking for points -- but is this really something we want to enforce on novices? If that's the only way to do things, I suspect people will just pebble-dash the answers sections looking for points.

And remember the problem with already have with withholding the commenting function: people will just post their junk where they can. If they can't post a new question, they might just jump on the most relevant question and post a "I'm having a similar problem"-non-answer. You need to be careful that you don't "solve" one problem by sweeping it to another section.

Counter-solution: Tackle the problem, directly

If this problem is real and you want to solve it, why are you putting blanket limits on things? To me there's one cause to the problems: bad users who don't write good questions.

Any question-limit should be based on past crimes. Target users with multiple closed questions (for bad things), negative votes on questions (not final scores -- we try to fix bad questions on AU), valid suffered flags, etc. Of course you can factor in their reputation but that's a tiny portion of it.

Using this sort of heuristic you can put people on probation where they essentially have to become better users or they start losing privileges.

Answer 4

1) Some people will just create new accounts on every 10 questions.

2) That make sense

3) If removing ownership from questions only for reps would probably ok, but if they literally lose those questions from their user pages that could be some issues.

Answer 5

No. There would be a risk of alienating users (specifically new users who are still getting the hang of this site). I understand it only takes an average of 2 upvotes per question, but that is not a completely trivial task; especially if you are new to the site and/or ask very specific questions in unpopular tags. I'm especially concerned for students, as they might not yet know enough to answer many questions, so asking questions would effectively be the only way for them to reach this 100 rep (unless you count asking them to copy-edit a few dozen questions).

Stackoverflow is supposed to be "100% free, no registration required"...not "your first 10 questions are free, but we demand something from you after that"...

Answer 6

Downvoting needs to be fixed first.

People have an aversion to downvoting; the data dump tells us that. Why? Because they lose reputation for doing it.

We should have a trial period of removing the penalty when downvoting questions; since there's data, it'd be easy to re-institute it (or wipe the votes from that period, as if it never happened).

Bad questions should be downvoted. Lots of downvotes on questions means the user doesn't ask good questions. The Rate limit shouldn't be on the number of questions you ask, but how well they do comparatively. Removing the penalty for downvoting questions will give people an honest assessment of that question's worth to the community.

Answer 7

A little questionable statement to start off:

SO, in general, doesn't seem to care about the quality of the questions asked as long as they are not complete crap.

There is no incentive to close-vote duplicate or mediocre questions as they generate lots of rep 'farmed' by many users. So those questions stay open and get answered

Is that changing now and is prohibiting users to ask questions the right way?

Users with 100+ low quality questions keep asking those questions because there are answers. And mostly they are answered very quickly and competently so there is no incentive for those users to stop asking questions and why should they? The System works for them!

The ones asking really really bad question will get enough feedback from the community as they don't get any answers. So they stop asking.

But that only happens for questions that are bad that they really get close voted. From my tags that are not that many, especially not the "easy to answer" duplicates.

So why 'punish' new users for using the system if it works for them?

If we, as in SO community, don't want people to ask duplicate questions why do we still "reward" them with answers? Why not change the system in a way that will make people clean up and show new users better what is an acceptable question and what is not.

For those users it's also not about reputation in any way (I'd assume). They come to SO get answers, not rep. Down-voting and taking away rep is not an issue as long as one gets an answer.

I don't think you will be able to prohibit people from asking bad questions. Well, you might be given that you have created SO and it's amazing but it's going to be a constant "trying to prohibit multiple accounts" battle that i don't think you can win.

To sum up: If people get their answers, they will keep asking questions. Change the first, not the latter

Answer 8

I like the idea, but if you change this to the following:

• every user has a free pool of 10 new (possible bad) questions (maybe even 5 for unregistered users).
• questions with at least one upvote (or possibly an accepted answer) are removed from the pool. So for each (approved) question you can add a new (possible bad) question.

This cuts the large amount of bad questions without hindering the genuine question askers. This completely locks users with 10 bad and closed questions, but this isn't a bad thing. Besides those questions are very often deleted and you can always delete them yourself (or flag for a moderator).

Answer 9

Any chance to enhance it with a time aspect? Like, for example:

1. After your first 10 questions, it will require 100 more reputation or one month longer registration to ask every additional 10 questions.

That slows down bad questions, but feels less restrictive to me.

Answer 10

What I like the best of this proposal is that, unless I'm reading wrong, reputation from questions counts towards unlocking further questions.

This means that you can continue asking questions without providing answers, so long as your questions are good enough (you need an average question score of +2 or better).

Here's a naive query that wrongly ignores bounties, suspensions, etc (one could calculate reputation directly, but that would ignore reputation caps and other bonuses):

• What'd happen if you needed reputation to ask questions?

For exampe, ooo has a ton of reputation spent on bounties (2.8k of which on a single day!). It will be interesting to see, however, what the rewriting history process entails. Did ooo have enough reputation when he asked his eleventh (twenty-first? thirty-first? etc.) question to ask any other? Would that "CW" all other questions, clearing associated reputation and bounties spent with reputation he then didn't have?

Fun. Fun. Fun. Fun.

Answer 11

LACK OF MODERATION IS THE PROBLEM

The whole reason askubuntu is having this problem in the first place is not that there are stupid people asking stupid questions. This happens on all the SE sites. The problem is the lack of moderation.

LACK OF MODERATORS = LACK OF MODERATION

The reason there is lack of moderation is that there is a lack of moderators. On other SE sites you get reputation quickly which makes you want to get involved more. This site is opposite. Because it is too hard to get involved, checking for duplicates and editing stupid questions to fix them etc. no one's reputation can increase.

LACK OF REPUTATION = LACK OF MODERATORS

The reason there is a lack of moderators/people with enough rep to moderate is because there currently is no way to increase your reputation here. The standard for what is a good question and what is a bad question is way too high. I've asked several questions that get no votes, no comments and no answers and they were good questions.

IN SUMMARY

So askubuntu is in a downward spiral unless you can do the opposite of what this question suggests. New users need more privileges and incentive to get involved.**

Answer 12

I disagree with the proposal. I think that implementing it is going to very quickly highlight the other problem that has been noted lately: the downward trend in voting_{(1, 2)}.

Some people simply don't have the knowledge to answer moderately difficult questions, or the speed to win the FGITW race on the easy questions. Combine this with the (perceived) lack of active voting, and the fact in can be difficult to even get views on your question_{(1, 2)}, and a n00b can end up in quite a difficult position, so you may get an explosion of new accounts (with an explosion of orphaned/abondoned accounts) as mentioned by @YOU.

So it seems there are two distinct problems:

• quality. A good tool already exists for dealing with quality problems: moderation. Maybe the answer is to introduce a Darwin™ flag - that is a flag for questions that really are no use to mankind and evolution needs to be unkind to them. Questions with enough of these flags can be devolved over time, till they get pruned or deleted after a number of weeks or months. This means the asker probably gets an answer, and the low quality question will automatically disappear after a while. Edit: Or instead of the question automatically being pruned after time it could appear on one of the 10K+ mod lists, and it can be deleted from there.

• reputation. If the problem is about users earning too much rep for a large amount of questions, then that has already been discussed, and a rep cap on questions or a throttle based on question & answer ratio would do the trick.

Maybe the best way to deal with quality is to let the community decide what is suboptimal, you just need to figure out the way to activate the community to do it. Down-voting by itself doesn't solve this issue, as frequently the askers of the bad question have low rep anyway, and a negatively scored question can still potentially stick around forever.

In any case, there is only a small amount of users really abusing the system - let's not introduce new convoluted rules just to bring them into line when the proper use of existing tools could do the same job.

Answer 13

I'm not at all fan of anything that "rewrites history" -- especially something that breaks your own attribution and licensing rules. I think it will better to just grandfather these questions in somehow.

I also really like Arjan's time-based addition (upvoted), so I'll repeat it for emphasis.

Update (another take):

I think Stack Overflow has undergone a fundamental shift from it's early days. Most users — especially new users — now only visit the site when a Google search points them to a question. Moreover, they only ask a new question when the one Google showed them didn't quite solve their problem. Put simply, the typical user never even sees the list of active questions waiting for answers. Stack Overflow has now been around long enough that many of these same people now have many questions each. This is normal, this is natural, and trying to change it just won't work.

Stack Overflow continues to succeed because of the sheer quantity of traffic Google sends it's way: enough of those people continue to need additional help beyond what the original search result provided that new questions continue to feed into the system, and enough people decide that they want to start answering questions to keep that part of the user-base happy. Of course, this is just my hypothesis; but I think most will agree it's probably not that far off.

The problem we have now is that some (probably not the majority, but enough) of the users described above consistently ask enough low quality questions that it's creating significant background noise, especially in certain tags. Perhaps some of them even bypass google and decide to go straight for the "Ask Question" button, but I don't think you can profile for them in that way.

With all that in mind, I think the proposed solution is on the right track. I certainly hope that good questions do tend to end up with a higher score than bad ones, at least on average. However, I think the numbers are still off. It's hard to earn 100 rep per 10 questions, even with good questions, because not enough people vote on questions. More importantly, the chosen numbers are completely arbitrary. What we need to do here is use the available data to determine more precisely how many votes per question on average you need, rather than just setting it at 2 arbitrarily.

So go to the data dump do something like filter to only questions by users where >70% of their posts are questions. Then sort these users by votes-per-question (perhaps limit to questions in the last year). Go through these users and find a real, data-backed point where you feel like the users above this line should be left unmolested, and the users below this line need throttled. Then use that ratio to build your requirements.

On the other hand, if we go look at the data and it turns out that there is no clear-cut point at which we can separate bad questions from good (or, at least, less bad), then we'll no we need to go back to the drawing board. If that happens, I think the best place to start is finding a way to improve the scoring, so that there is a better distinction for these questions.

Answer 14

Sounds like a good idea, but I'd be interested in seeing some hard data first:

• How many users will retro-actively be affected by this?
• What are some of the questions that will be affected, is their quality noticeably different to ones that aren't affected?
• Is this StackOverflow only, or all StackExchange sites?
• DOES IT AFFECT ME?!! :)

Answer 15

No

What made SO so rich so fast was the possibility of asking questions, and the answers that came with them, of course. I don't think we should put a cap on what makes SO work.

There's already a system for pruning bad questions: downvoting.

There's already a system for pruning duplicate questions: flagging/closing as duplicate.

Reputation is at best a gimmick, and we make too much of it. If big askers with a large rep score are such a big problem, cap their privileges to answer-earned rep only. Or, make upvotes on questions worth 1 point.

Just don't put a cap on what makes SO rich with knowledge.

Answer 16

Counter-Solution

Instead of forbidding low-rep users from asking additional questions, maybe put some checks to affect the rate at which they can ask questions. If a user with no rep asks a question, they must wait n hours/days before asking another question. The more rep they earn for that question or that they have in total, the shorter the wait until they can ask again (and for users with lots of rep that wait would be ~0). This could serve to slow the rate at which poor questions are asked while not being too harsh a punishment for new users who have difficulty earning rep.

I'm just throwing this out there as a less severe alternative. I'd like to see what the community thinks; frankly I'm not sure what I even think of this solution, but I want to try to find a solution rather than just pointing out flaws in proposed solutions.

Answer 17

Ah, if only we could just get new users to honestly respond to a pre-screening question like, "Do you intend to post scores of poorly-written questions that suck time away from other users and result in answers that will never be useful to anyone but yourself?"

Aaaanyway, I like this idea. It doesn't go out of its way to "punish" users who got stuck on the wrong side of an arbitrary metric. It doesn't place onerous burdens on (or grant scary powers to) the editors, voters, or moderators. It doesn't solve a "big city" Stack Overflow problem at the expense of budding Stack Exchange sites, and it could probably be adjusted easily on a per-site basis if necessary.

...And it does directly address a frustration that has been expressed by concerned members for a long, long time: users who take, take, and take a bit more without ever giving anything back. What's the simplest way to stop users from asking dozens of not-very-useful questions? Stop them from asking any questions at all until they've made those already asked more useful... We already do this for users who post terrible questions, but at some point it's not enough to just avoid being terrible - if you're gonna stick around, you should be trying for good.

But more importantly, it emphasizes something that's often overlooked: voting on questions is very important. Even when it feels like a waste of time, it still adds precious, delicious data to the system... In lieu of that handy "I ask lousy questions" profile checkbox.

Answer 18

One significant side effect is that in order to ask a new question, users will delete their old questions. Since 60% of the questions being asked today receive NO upvotes, even though only a fraction of those are really low quality, we would be forcing users to delete reasonable questions when they want to ask new questions.

Answer 19

Apart from “historical questions”, I see no problem with this.

I think more thought needs to be given into what happens with “historical questions” when the user would now not have had enough rep to ask them.

I think it would be lower risk just to leave historical questions alone for now and let any user that don’t have enough rep to ask questions due to their historical questions, have “10 free questions” as a one-off action.

Answer 20

Based on looking at badp's query in his answer, I was questioning whether this was going to address a significant portion of the mediocre questions being asked on a regular basis. The users near that top of his list don't seem to be poster children for the problem facing the site.

However, I took a look at some of what I think are lower-quality questions today in my favorite playground, the [iphone] tag. It's safe to say that the [iphone] and [android] tags are littered with bad questions, due in part to the gold rush in mobile development. If I examine the last few hours of questions, I see these badly formatted, repeat, or otherwise mediocre questions:

• how to plot a scatterplot using numbers in core plot (13th question of this user)
• How to store multiple text messages in array using plist (14th question of this user)
• How can i stop playing movie? (2nd question of this user)
• login user able to see all registered user of openfire xmpp server from iphone xmpp client…How to implement such functionality !!! (119th of this user)
• Verifying a Receipt with the App Store (38th of this user)
• Time components separate from other in Objective-C (2nd of this user)
• get key value from .strings file (19th of this user)
• What is code for localization to pass the string file iPhone Simulator? (18th of this user)
• How to check whether username and password have been saved in keychain (3rd of this user)
• iphone error in delete function (1st question of this user)

Some of these could be salvaged with a little edit work, but you get the idea. I'm not trying to poke fun at these users, just provide some examples of questions that need work.

From this list, 6 of the 10 would have been prevented by this policy (assuming those users did not just create a new account as a workaround). I was surprised at this, but it does seem that the majority of the bad questions in this tag are coming from repeat askers.

Browsing through the questions over the last day, I don't see any false positives (good questions that would have been prevented by this filter), which also surprises me.

Now, the 4 out of the 10 bad questions that were from new users aren't targeted by this. Perhaps a question from a new user that scores highly enough from some bad question heuristics (use of "plz", no uppercase letters, etc.) could be placed in a review queue before posting, like low-reputation edits currently are. I know that we have the /review page, but I wonder if some sort of queue that's as easily accessed as the edit review page could work, with similar approve / deny / edit functionality. Then again, the sheer number of questions coming in might not make this feasible.

Answer 21

This site requires both questions and answers to survive. Experts provide most of the answers, and non-experts provide most of the questions. It's a balance.

It's wise to practice moderation in all things, including moderation. -Jeff Atwood♦

Removing too many questions will offset the balance. We need to remove bad questions, not just all questions of any user that doesn't meet certain heuristics. IMHO, the underlying problem here is lack of voting on questions. Bad questions need to be determined as those that are downvoted, not just those that aren't upvoted. To accomplish this, people shouldn't be as discouraged from downvoting as they currently are. Downvoting is a better punishment for asking bad questions then simply being told "You're not good enough. No more questions for you."

I think the mechanisms behind downvoting need to be rethought.

I support Adam Davis' proposal to allow one free downvote for every two upvotes that a user casts.

Answer 22

Is the situation really that bad it requires such extreme measures? I think that's the question we should ask first.

Such action is like amputating hand or leg to save someone's life... we will lose lots of existing members and lots of potential new members who won't like the idea.

However, I do see the benefit - those who will stay will put "better" questions (or in other words think twice before posting), and the overall quality of the contents will improve.

So, are things really that bad right now? From what I see - No.
True, we're swamped with tons of questions that many times repeat themselves and many of them are bad quality but so far the community is handling this well: bad questions get downvoted and purged after some time. Duplicates are closed. The good questions get upvotes and good answers.

To put some visualization, the forest of Stack Overflow is getting bigger and bigger with more and more small trees that hide the stronger and bigger trees that actually sustain the forest, but those big trees can still be found especially if the forest rangers will nourish them and paint them with bright colors. :)

Bottom line: don't implement such thing yet - it's way too early in my opinion.

Answer 23

Instead of 100 rep for next 10 questions from users, I think it is good to have something similar to flagging system.

i.e. To start with, every new user can ask 10 questions. But after those 10 questions, he has to earn some reputation (may be every additional 15 rep points) to ask additional questions.

All these constraint will be available to user until he reaches 500 rep points. After that there will be no limit on asking questions from that user.

Answer 24

I have mixed feelings about this. On one hand, I like the idea of limiting bad questions, but on the other, a strict limit may be counter-productive, specially in the case of very specific (good) questions, which seems to be the main concern by looking at the previous answers.

One alternative is to require positive votes only to those questions that exceed a certain view threshold. Throwing random numbers, questions with less than 20 views and no downvotes may not count towards any limit.