56

The ban process is very useful to preventing low quality posts - for the most part. One very large glaring issue with it is that users can bypass it at will by using a different account.

On a daily basis users hit the ban, and then either bypass it or take to meta to figure out why it happened and how to fix it. Their questions are a constant flow and have been for a long time, this should come as no surprise.

Being banned from anything is rather jarring. In this instance, I know that it is meant to be. However, I feel that if users feel more invested in their account then they will not bypass the ban. Bypassing the ban is a problem I think we can work on.

Users would feel more invested in their account if they were allowed to build more content before being entirely shut out. Being shut out so suddenly is the "cold turkey" reference I make.

"Cold turkey" describes the actions of a person who abruptly gives up a habit or addiction rather than gradually easing the process through gradual reduction or by using replacement medication -wiki

I think that question bans would be more effective as a throttle than as a cutoff.

For users who hit the ban, as opposed to a complete stop, begin throttling their posts.

Questions from this account are being limited. See the Help Center to learn more.

Have the limit be constrictive, and approach 0 but never hit it. Perhaps when the user first hits the ban limit them to 1 post per week. As the metric used for the ban progresses negatively, increase the week count. Using this approach, after another 4 posts if each one made the metric worse, then perhaps it would have taken then 1 week for the first, 2 for the second, 3 for the third, 4 for the fourth (assuming a +1 for each decrease in the metric score) -- meaning it took them 10 weeks to make another 4 low quality posts.

The benefit is that if they had rerolled their account, this would have taken them a mere 1 day to make another 4 low quality posts. Moreover, they feel they can still provide content at some point and also feel more invested in their account as it progresses.

This would also allow them a possible path to redemption at some point if they were able to begin to work well enough with the system to climb out of the throttle.

tldr;

Significantly throttle posts of banned users instead of preventing them from posting entirely.

  • 12
    The original premise behind question and answer bans is the notion that most users will never reform, and I still think that's probably true. – user102937 May 1 '14 at 17:53
  • 3
    @RobertHarvey - That could be true, but don't you think it would better to have those users be throttled as opposed to just rerolling their account and bypassing the ban? – Travis J May 1 '14 at 17:54
  • 3
    Well, most of the low-quality problems are caused, not by a small number of users asking a lot of bad questions, but a large number of new users asking a small number of bad questions. I'm not sure throttling will make any difference, in that regard, although it would certainly give people a (marginally) better chance at reforming. – user102937 May 1 '14 at 17:54
  • 3
    Oddly enough, we were just talking about something like this two days ago. I'll leave the details for someone else, but we're working in this direction. – Pops May 1 '14 at 17:55
  • 3
    @TravisJ Post bans are not just for an account, they're at a lower level. Most users that get post banned are because they're so lazy that they aren't even willing to Google their question first, or glance through the site's rules. Yes, some of them will be motivated enough to attempt to subvert the ban, but many won't, due to the exact same laziness that caused them to be banned in the first place. – Servy May 1 '14 at 17:56
  • 14
    We're having a meeting, coincidentally, about a spec I wrote to fix this right now. I'll answer this tomorrow. – Tim Post May 1 '14 at 18:00
  • @TimPost: I totally forgot to dig you your answer on MSO. I did so now: Question quality is dropping on Stack Overflow – Martijn Pieters May 1 '14 at 18:22
  • 1
    First decent FR on meta.SE, one of the best FRs in the past year. +2 (don't ask how) – user1228 May 1 '14 at 18:50
  • 1
    I don't think that this will lower the amount of low quality crap that comes in. Rather, the user will figure out that they're being rate limited, and bypass that, rendering the entire process moot. – fbueckert May 2 '14 at 2:20
  • 1
    @RobertHarvey Ahem!, I had a post ban on SO,the fist SE site a ever visited, and I hardly have gotten out of it, but I've now got 4k+ network rep. Users can reform, just give them a chance. Some people just need to learn the ropes, in order to contribute in the future. – GiantCowFilms Dec 31 '14 at 5:13
  • @GiantCowFilms: You're responding to a comment that's eight months old. – user102937 Dec 31 '14 at 6:12
  • 2
    @RobertHarvey True, I do that, since I guessed you'd still be here, no one that makes it to meta.se ever escapes. Also, its my opinion, which my actually be read by people in the future. I posted it because I felt that that opinion was needed for people who read this like me, 8 months after it started. – GiantCowFilms Dec 31 '14 at 16:53
  • YES PLEASE! I think this idea is amazing! – James Aug 16 '17 at 0:10
54

Question blocks worked well enough at the scale that they were introduced. Those that could be helped were able to get out from under them, those that couldn't generally just left. They're not working very well today.

Because they're so difficult to get out from underneath (voting is much heavier these days), people find it easier to just request account deletion and start anew, often repeating this cycle as long as it takes until they get the hang of it. A significant amount never get to that point.

We're fixing this, and it's not a single easy fix, but a combination of changes to the system, where we ultimately want to convert blocks to a sliding scale. Before we can get to that, this is what's in the pipe:

  • Warning users when they're getting close to a block, and when they do things that could potentially get them closer (e.g. deleting posts when close to the edge)

  • Controlling recidivism by making the fact that an account was either blocked or suspended persist after deletion

  • Coming up with a system that causes users to treat asking questions like any other resource that can be depleted - encouraging them to ask only when they must, and to make it count when they do.

I'd like to expand on two points, the first being recidivism. If you're suspended and request account deletion, you will be re-suspended for the duration of your suspension if you return during that period. For instance, if you're suspended for a year on Stack Overflow, return after 11 months, you'll still need to serve the final month of the suspension.

Additionally, if you're blocked at the time you remove your account and return, you'll be limited to one question per week until you can establish yourself as a contributor to the site.

I borrowed heavily from our very successful spam prevention system in order to make this work, and like the q-block algorithms, I can't go into a great deal of detail about how it works or people will just find ways to game it. We'll be talking more about this once it has been implemented, which we hope to see at some point next week.

The second is the sliding scale system which we desperately need to get working. It's going to require new users to show us that they care about the quality of the site before we let them ask a lot of questions. If you ask a question and do well, we let you ask more, and then more, and then ultimately lift all restrictions. If you bomb, you're slowed down significantly, but getting out of it is going to be easier than dealing with the restrictions you'll have if you just delete your account to get around it. It's this part that we're still working on, but Shog and I have been giving it a lot of attention lately.

In short, the current blocks have become self-defeating because most people will not be able to get out of them, which has inadvertently caused this super high rate of recidivism. The more people do this in order to ask questions, the less they care, which means the less effort they put into asking - the complete opposite of what we wanted.

It's a high priority for us, but it's going to have to roll out in stages, and we still have some interesting technical problems to solve when it comes to a total implementation. However, we're doing it, steady as it goes.

  • 7
    This looks very encouraging. – Lightness Races with Monica May 18 '14 at 13:49
  • 1
    If "voting is much heavier these days" is the reason for question bans being harder to escape the right way, via positive edits, then why don't we do something to limit that? e.g., editing a question causes its score to increase to no worse than -5. Then it doesn't matter if more people pile on with downvotes on the initial bad question. Getting back to positive numbers is still quite feasible. – Ben Voigt May 27 '14 at 23:28
  • @Ben I'm working on changes that better scopes recent activity in several time samplings. No firm details yet, but I think I've got that solved. Those that just get off to a bumpy start won't be punished like they have been, while perennial vamps should just find more fertile ground. I'll be posting about it next week. – Tim Post May 28 '14 at 9:40
  • do you plan for network-wide changes or only SO / Trilogy? – gnat Jun 4 '14 at 18:33
  • While question blocks are limited to only a few sits, suspensions are ubiquitous. I don't see why they would be partial to some sites in this regard. – BoltClock's a Unicorn Jun 14 '14 at 11:34

You must log in to answer this question.

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