Topic on User talk:Jurjendevries

Hi Jurjendevries, I'm Doug, a Steward on Miraheze. Stewards, together with Global Sysops, educate users about and enforce our global policies, on the one hand, certainly, but they also support small wikis with few or no active local bureaucrats and administrators with problems that may arise on wikis. Recently, a user, DarkMatterMan4500 reported to global functionaries in our #cvt channel on IRC that permanentfuturelabwiki had been inundated with a barrage of spam only accounts creating spam page creations advertising various spam websites, requiring constant monitoring, patrolling, and deleting by Stewards and Global Sysops alike.

As we are a small team hosting nearly 4,000 wikis across the wiki farm, we obviously do not have the scale of resources to track, monitor, and remediate all spam creations. So, as part of my responsibilities as a Steward, as you'll note from your wiki's ManageWiki log, I initially tried enabling ProtectSite as a temporary measure to limit page creations to only users in certain permission groups. Unfortunately, I was unhappy with that extension's design in that it does not permit the wiki protection level to be set to autoconfirmed and its logging facilities, as well as the fact I couldn't be certain if the temporary one month protection level I'd set had indeed been set correctly. So, as you'll note, I reverted my change, and opted instead to modify your (Main) namespace to require autoconfirmed permissions (an implicit user group granted by default to all users once they have had an account on the wiki for at least four (4) calendar days and once they've made at least ten (10) edits). Since the spam only accounts generally create fewer than three (3) pages and often only one page, this is a measure that would have minimal effect on legitimate users on the wiki, and would be remarkably effective at preventing spam from occurring on your wiki.

As always, you are of course free to revert this change, as this was designed to be an interim protective measure in absence of locally active wiki bureaucrats and administrators.

As you are the only active bureaucrat on this wiki, who was briefly last active on 28 October 2020, I'm opting to notify you of the actions taken by Stewards and Global Sysops in support of your wiki. If you have any questions at all, please don't hesitate to reach out to me either here by pinging my username ([[User:Dmehus|Dmehus]]) or on my user talk page on Meta. As well, for a current up-to-date list of global functionaries supporting your wiki, please see Special:GlobalUsers. :)

Thanks,
Dmehus (talk) on 05:57, 13 December 2020 (UTC)

thank you @Dmehus

No problem, and thank you for the prompt reply. :)

It was because I noticed a ton of spam pages that made me bring it to the CVT channel on Discord.

@Dmehus @DarkMatterMan4500 , Yes I can understand the reasons. Sometimes I did received a message with spam from the wiki. Mostly a small change to a page.

Is there a captcha option? e.g. https://www.mediawiki.org/wiki/Extension:ConfirmEdit with QuestyCaptcha ?

Jurjendevries ConfirmEdit is an extension that is installed as part of the mandatory wiki configuration on Miraheze, as I understand it, and, from the looks of it, it looks like it's part of MediaWiki core now. So, the spam only accounts seem to be able to solve the captcha challenge questions, I'm assuming? I do know spam farms do tend to employ humans to solve captcha questions for the automated spambots. If that's not the case here, then perhaps the captcha questions are too easy for the spambots to solve? At any rate, absent a stricter change to the global abuse filters, in which case we have to weigh the potential for false positives being ensnared by such filters, I'm not sure what more we should besides just leaving (Main) namespace limited to autoconfirmed or confirmed users since that seems to be working (i.e., the spambots aren't smart enough to edit in other namespaces first). You could also write a local abuse filter, if you have sufficient knowledge of the MediaWiki abuse filter regex that could warn, disallow, and/or block any spam only accounts that post an external link. Blocks could be short term (less than a week) to ensure any false positives are not indefinitely blocked. Something like that, perhaps? Let me ask Reception123, a Global Sysop, if he has any other ideas.

Jurjendevries, I've just had another look at the QuestyCaptcha module of the ConfirmEdit core extension, and it looks like that can be configured with your supplied challenge questions. As Reception123 is also a system administrator, he could assist you with doing the GitHub pull request(s) to configure this for your wiki. I'm not sure if we have any wikis using QuestyCaptcha, but we may.

@Dmehus Thank you so much for your investigation. Do you know if the spam is from bots or real human? If it's from real human then I agree it won't work. If it is from spam bots I think it will have a chance. If we don't know if it are bots or real human, we can have a try if this will solve the spam problems if it isn't that much work to do the github pull request(s).

Jurjendevries No problem, and I agree that I'm not certain, completely, whether they're human or bot spam only accounts. I suspect they're mostly likely bots, though likely controlled by one or more humans, or they're bots but the nefarious human operators behind the spambots employ a team of humans to answer challenge questions for the spambots. I don't think, though, that it could hurt to try. You can either do the GitHub pull request yourself, if you know what to do, or you can file a Phabricator task specifying the ConfirmEdit php variable configuration changes requested (which would be different from whatever the default configuration is). It would be helpful to also link to this thread, too. Hope that helps.

Thanks. I have created https://phabricator.miraheze.org/T6603 .