As the former CFO and a former SRE for Miraheze, I'm going to try to save the farm. I was uninvolved in the current problems, and at first was wondering what was going on so as to save my own wikis. The fundamentals at Miraheze are much better than they look; it's just that certain problems between volunteers got far out of hand, and with an all-volunteer force other people decided to leave to escape the drama.
There are several months of runway, all the servers are still up, and many volunteers are still around. All we need is more leadership that is ready to focus on our core mission of offering free, ad-free, volunteer driven hosting to wikis with a lot of options. With a little networking, the finances are a solvable problem.
I'm in talks with the current board to see if a group of current and former volunteers can take over, either by working in the current organization, or starting a new nonprofit and working from there.
I still think Miraheze could have a bright future, and that this was a series of mistakes that got blown far out of proportion to the actual problems that the wiki farm faces.
AMA. I was not involved in the current crisis, so piecing together what happened has been something of a puzzle to me, but I can try to answer most questions.
On "starting a new nonprofit": yes please, it's about time to have an entity with proper accountability.
Miraheze is mostly volunteer-run but it also needs a commercial side (such as selling premium features/support, and paying some people to handle them), so an obvious solution is a coop. I can help found a coop in Finland, where we have several long-standing MediaWiki devs, so that any EU person can easily participate. There might be other suitable EU jurisdictions but Finnish law is quite nice and easy:
https://www.finlex.fi/en/laki/kaannokset/2001/en20011488.pdf
Right now I am US based, and I'm thinking of incorporating in the US. Not because of any preference for the venue, but because of our limited resources and volunteer distribution. Honestly I'd prefer Canada, but as I am 2000 km away from there, it's not optimal for many of us. Anywhere in Europe would be better than being in the UK.
In any case I'm pretty sure we could get 501(c)3 public charity status, even though Miraheze missed out on charity status in the UK. If you could follow up on-wiki about co-op status as an option, I'd appreciate it.
On accountability: Members of the board need to come from the community and represent the community. Yes, it's true that no one tried to run for the board seats that were available, but people weren't encouraged to do so enough. I don't think the decisions were bad in and of themselves, but that not enough effort was made to get community support.
On paid premium features: I'm not opposed, but I want to see a business model that works without taking significant resources away from our free customers.
It's very difficult to set up a proper non-profit able to handle both donations and commercial services in the USA. You'll end up spending fortunes on laywers. It's much easier to use a jurisdiction where the law already does all the legal job for you.
That's a fair point but on the other hand I've been at Miraheze for almost 8 years and having such a huge operation rely on a few unpaid volunteers isn't realistic in the long term and people burnout, lose interest or both. There's also the anxiety of knowing that the fate of the farm depends on a yearly (or even more frequent) fundraiser.
These are problems that all small nonprofits face. I would love to pay people, but even having a commercial model doesn’t mean you escape questions about funding and staffing.
Particularly at this point, I'd like to see better communication between the board and wiki members/crats/admins. No news has been the rule with Miraheze except for planned service outages and donation requests. The way I found out about the closure announcement was from a Tweet that rides in with a 503 error message. I know it's early days, but a message about proposing to continue Miraheze would be useful on that platform, yes?
No communication is standard operating procedure right now. Volunteers found out this decision at the same time as everyone else, which meant no one had time to prepare.
We need to better communicate our plans and goals, and of course about downtime. But we also don’t want to flood average wiki users with sitenotices either. What kinds of communication would you like to see more of?
I will put out a sitenotice soon, but I need the outgoing board to buy-in first. Which I’m hoping will happen in the next few hours.
1. The shutdown notice would have been one; important news, instead of only posting on Twitter and other social media. After a few tweets like "30,000 wikis!" (not all active, ahem), this just turns into noise.
2. Decisions made by the board. A summary will do.
3. Current or ongoing technical issues. Or do we have to go through the whole of outstanding Phabricator tasks?
This communication can be accomplished by subscription. I would then like to see a notification/bell made with a capsule of what's being shared and its location within the Miraheze maze.
On other business, check out Lithuania as a possible location. It's a very tech-oriented country, and I've worked on a MediaWiki-based wiki based there. I don't know the specific server farm or company.
Yeah, it was a joke. That I backed off of right away. All I could think was that I'd like at least one voice who wasn't saying doom and gloom about everything near the top, so that we could possibly save things. Right now Miraheze is in an entirely unnecessary cascade caused by overreactions.
> ...long term sustainability problems, financial situation...
Weird, because https://meta.miraheze.org/wiki/Finance says they have more money right now than they had at the start of the year and twice as much as they had the same time last year.
> Owen and Alexander are fully committed to seeing the sustained growth of Miraheze
> Compared to where we were before 2019, Miraheze is in a much better financial position, has drastically improved its legal standing and compliance and has taken strides to improve the technological stack behind the project through the right levels of investment being open for technical volunteers to harness.
So, which is it? Are you shutting down or fully committed to sustained growth? Are you now unsustainable or better financially than ever before?
Nothing about this announcement makes sense to an outside observer who has no idea what "taking into account recent events" means.
My impression from reading their Discord server yesterday is this:
- Volunteers from Miraheze planned to create a paid wiki hosting service to supplement Miraheze called WikiForge
- Several of those volunteers left Miraheze to start a free wiki host called WikiTide
- The remaining volunteers were burnt out already and decided to call it quits
- There are some talks of handing off Miraheze to another group of volunteers
FWIW I had a horrible experience using Miraheze last year:
- September: I created my wiki
- November: Miraheze had an error with a drive and over 25% of their wikis were lost. It took them weeks to admit the extent of this error and that they had no plans to use a professional recovery service
- Early December: I recreated my wiki and rewrote it (250+ articles)
- Late December: Miraheze recovered the back-ups they thought they had lost and said they would merge the new and old versions of recreated wikis. Somehow, due to 'someone unplugging a hard drive', they lost all of the content I'd written since November.
I've since learned my lesson about making local back ups, but come on. It was clear there had been tons of technical and communication problems for a long time.
> earlier today Owen had assigned himself checkuser and suppressor rights for "auditing use of permissions" and left Raidarr a talk page message regarding an important email; sometime after that, Raidarr removed all advanced user rights from every wiki in which he held them before removing his local Steward group and making the log summary you're inquiring about. I assume it was the email or maybe a resulting conversation that caused Raidarr to resign; however, this is just speculation
in which the dissolution is discussed as an alternative to training volunteers on legal compliance with the UK Data Protection Act:
> OB clarified both proposals are mandatory under Information Commissioner’s Office advice – we need to ensure we train people appropriate and audit their access and usage of personal information inline with Data Protection Act 2018.
> TH raised a counter proposal of dissolution of the company.
My wiki also got affected by the harddrive. They send out a message with everything they knew pretty much immediately even to people not affected by the error and continued to update as they learned more. Not sure why you seem to think it took "weeks". For a volunteer-run service they've actually been quite on top of it and compassionate in my experience
You're right, it was 10 days, not quite weeks. It's not important but here are some of the miscommunications and apologies:
November 18: "The cloud server which hosts one of our database, db141, experienced a disk issue. As a result, a small number of wikis hosted on db141 are unavailable. [...] While cloud14 has been reinstalled, we will have to send the affected disks to professional data recovery. The earliest ETA for having wikis restored is potentially early next week. [...] Our number #1 goal is to restore the data on the disks affected so that wikis are restored versus using a backup which could be various days old"
November 24: An update to the situation was embedded in an unnecessary meeting on Discord: "Our Miraheze Meeting is starting now! Join us while we talk about everything Miraheze, such as recent policy proposals and we'll be providing an update on the db141 issue We'll answer any questions you have and will listen to any comments or concerns."
November 28: "I understand the frustration, and we have not yet given up hope on data recovery, so you can still wait if that is your choice. But I also don't want to give people any false hope. It is not looking super great for data recovery. I won't completely rule it out yet, and I can't say anything definitively. Venues will be opened shortly for requesting that wikis be recreated from scratch (with images at least still intact) if that is your wish [...] in the recent Miraheze Meeting it has been brought up that our communication has been less than ideal, and some feel some questions have been dodged and/or ignored. I have gone back and looked at our responses to some questions, and I can see how this could be assumed [...] I would like to clear one thing up now. Something that seems to keep being brought up is that we sent the drives to a data recovery service. This was not true [...] Owen currently has the drives and has for a few days. They are not at a data recovery service. This was due to some internal miscommunication within SRE, and is something that is being addressed [...] Once again we apologise for this miscommunication and we are working to resolve our internal communication problems [...] Indeed, over the last few days, it would appear that information was not properly being relayed within our own team which led to some breakdowns in communication. Clearly, different members had different insights and views on certain topics. This has led to a strain in communications and we are working to rectify that."
I don't think it was ever explained how the data was recovered.
My guess is that they used some of the XML dumps we of WikiTeam keep at https://archive.org/details/wikiteam , that not everyone at Miraheze was necessarily aware of.
It was your bog-standard disk crash as I understand it. Pros were (eventually) brought in for data recovery. But I also understand the backup for the wikis on that server (like mine, one of the dead ones) was on the same server, which doesn't sound like a very good plan. If a backup was stored elsewhere, the dead wikis would have been restored in a very short time. Lesson learned, I hope.
I got most all of my data restored, but there was a temporary problem of deleting and replacing files that the server programming thought still existed. That was fixed quickly.
> To clarify a bit on the reason behind this choice, the technical team behind Miraheze is currently 3 people (5 if you include two unprocessed resignations). Two of those three people (myself include) are currently burned out on the project and wish to step away.
There's also some discussion of possibly handing over the project.
I had figured it was not exactly technical burnout in running Mediawiki installs but regulatory burnout in following Euro GDPR/cookie banner/bullshit-of-the-now requirements. Some of them are easy but the rest phase in once you reach a certain scale of users (which it sounds like they probably have).
Yeah, speaking as a European tech lead; GDPR was a pain in the butt a few years ago when it was introduced - and all client websites had to be retroactively fitted with cookie systems, routines had to be implemented, stuff had to be learned, etc.
Once sites are planned with it in mind from the get go it's really no big deal, and huge for the end user.
For a wiki farm like Miraheze, GDPR/CCPA/etc compliance is fairly easy. MediaWiki collects fairly little personal data about users, and there are well-defined methods for the site operator to export or delete that data if a user requests that they do so.
Besides, that all kicked in years ago. It's not a new concern.
Cookies were regulated by the ePrivacy directive back in 2009. The case raised above is about someone who had access to see the users' IP address. Handling training and authorisations for people with access to personal data is an ongoing job, not something you do once and forget.
Certainly it is hard to make it gel on the money side.
From their frontpage:
> If 550 of Miraheze's 300,000+ users gave only £12 (16.35 USD) once a year, Miraheze would exceed its budget requirements.
So not for want of £6,600 / year, presumably?
Labour's of love like this can just burn you out. Building the right sustaining community is hard, but it also seems like they're too big to just fizzle out like this? A shame.
Miraheze is/was a wiki farm that provided a free Mediawiki-based wiki to pretty much anyone who was able to articulate a reason and fill out a request to be given one. This was invaluable in an age where the term "wiki" has become so debased that "a bunch of Markdown files in a GitHub repo" is, against all good taste and reason, considered a wiki. Miraheze was supported by donations, which is almost certainly why it's shutting down. There's no reason I can see that it shouldn't have allowed you to set up an annual bill-me cycle, but it didn't. And now it's going away, unfortunately. It would've be a good candidate for a pay-what-you-want model that incorporated usage-based pricing.
Another comment mentions that a lot of the volunteers on this projects have moved on to a similar project called WikiTide, and that's the "recent events" mentioned in the article. https://meta.wikitide.com/wiki/WikiTide does seem like they offer a similar thing (though, of course, Lindy's law applies regarding its reliability/viability).
The thing with that is that is is the same people that ran Miraheze that have forked, so the thinking behind it will be the same, and it largely has been — they've copied verbatim all of Miraheze's policies and stuff, so I doubt the systemic problemls experienced will be different.
Thanks for explaining. This announcement joins a long list of shutdown announcements where I find out a thing exists by its shutdown announcement, which neglects to explain why I should care about it.
They are a free wiki host run by a UK non-profit, pretty much the only free wiki host that has no ads whatsoever and also offers a reasonable selection of Mediawiki extensions. There are thousands, but some examples of popular wikis hosted on Miraheze:
- A ton of amateur wikis that migrated from Fandom (formerly Wikia) after a series of poor decisions by Fandom, such as the Polandball Wiki: https://www.polandballwiki.com/
- MOGAI Wiki, a continuation of some of the gender and sexual identities wikis from Fandom. There was a huge debacle where fandom tried to forcibly merge various superficially similar wiki communities, which caused a lot of chaos for LGBT wikis due to disagreement over xenogenders. https://mogai.miraheze.org/wiki/MOGAI_Wiki
It’s probably not just about money but also the fact you have to pick your administrators from the pool of people willing to deal with the interpersonal drama and legal headaches that come with those projects.
So that would be people looking for a captive audience and that’s typically for a reason.
I only knew about it because a popular mod Space Exploration for the video game Factorio has its wiki hosted on it. Actually even then I never went up to miraheze.org to look at the rest of the site.
media wiki's a particularly complicated thing to host and I don't think there's anything else quite like MiraHeze currently. It's been used by everything from video game wikis to casual conlangers to even some state governments so I'd say it's a pretty big deal
That's true only because mediawiki is designed for the infrastructure of a previous era, and massively overengineered for these kinds of use cases. Use any of the many free solutions for hosting a free site now. Use tiddlywiki in a single file for 99%+ of wikis. Or any linked markdown system. Making sites for small communities is easier than ever but harder to discover due to noise. That doesn't make mediawiki good.
Some time ago there was some huge fallout/fight visible on their Discord which I did not understand (it was all referencing some internal Wiki, mailing lists etc.).
That seem to be the "recent events" on first position in the announcement.
I hate how in search results Fandom is always above the better wiki about the same topic, that are independent or on better wiki farms like Miraheze. Fandom's SEO is brutal. They've slowly crippled the website and fucked over the regular users, but that doesn't affect them does it? Pinned at the top of Google there are always more clueless losers cascading in, they can't lose.
Worse yet: Fandom has a policy of not deleting inactive wikis, even if the community has deliberately abandoned them to move elsewhere, and will treat any attempt to advertise that another wiki has been created as vandalism and will block users (even wiki admins!) who attempt to do so.
This means that, if a community does decide to move their wiki off of Fandom, they have an uphill battle to "beat" their old wiki in search results. In many cases, Google will treat the new wiki as duplicate content and suppress it in search results unless pages are substantially rewritten.
They penalise duplicate content because 999 out of 1000 times it's spam. It's a bit of an issue forking any site, and one reason you might be better off just starting over.
There's an excellent alternative front-end for wikis hosted there called BreezeWiki.[0] There's a mirror available at the very convenient antifandom.com which allows you to easily view a better version of any Fandom wiki page you've landed on.
I wish mediawiki was more loved. It is an amazingly effective and intuitive piece of software. But maintaining, upgrading and managing it and the various extensions is a pain.
I suspect a rethink and rewrite could substantially reduce the cost of running such wili farms but is unlikely to ever happen.
TBH, I don't really see what the new settings stuff offers over the status quo. It feels like a complex abstraction layer to do what everyone was already doing already.
I'm not sure either. :D But having some explicit support directly in core hopefully means that we recognise wikifarm support as an essential feature and that over time there will be more collaboration to share good methods to run wikifarms. Multisite Wordpress also took a while to become good, IIRC.
I think fundamentally the current ext management system works well for Wikimedia and really huge farms (I would include Miraheze in that. MediaWiki is open source and there are plenty of non-Wikimedia foundation developers, but there aren't very many from the small wiki farms communities, so stuff aimed at them doesn't really get done.
Was doing some reverse engineering recently and wanted a quick wiki to put stuff on in case others wanted to join in and contribute findings about the same product. Looked around and Miraheze seemed to be the only real option for free wiki hosting (I tried another one first, don't remember what it was called, but it was bad).
Experience was pretty good adding the pages but then I went back 6 months later to try and consult my wiki. It had been deleted per an inactivity policy. Fair enough... they did warn me -- my own fault for not checking the email account I'd signed up with.
Anyway, I'm sad to see them go because it was a great service and I don't know of any replacement.
Yeah, NotAracham from Miraheze was really helpful and pointed me to this page: https://meta.miraheze.org/wiki/User:NotAracham/Wiki_Recovery... but I went through it and it doesn't seem like my wiki was captured on archive.org unfortunately. The .tar backups on archive.org only go up to 2021.
Confusing - what about the free github and gitlab wikis? And all the free wiki hosts? And tiddlywiki which has been freely hosted tons of places for 20 years or so?
Don't fall for the trap; those aren't wikis. Just because someone (or a company) uses the word "wiki" when describing something, that doesn't make it one.
Perhaps you don't understand what a wiki is, but there are hundreds or more implementations of every reasonable etymological interpretation of the term. If you have a particular narrow version of it, that's not universal or helpful.
I was actually specifically thinking of a MediaWiki instance. In this case there wasn't a codebase to version control so I didn't think of github and gitlab as options but maybe that makes sense. I don't know about tiddlywiki -- does that work for collaboration? It says "personal web notebook" so I sort of assumed it wouldn't have the multi user admin setup that mediawiki does.
Sad news. I started a wiki on Miraheze years ago, that became a fledgling community. It became inactive after a few years but I really appreciated the no nonsense, ad-free, vanilla MediaWiki hosting. They provided an important service for non-profits like mine.
There are several months of runway, all the servers are still up, and many volunteers are still around. All we need is more leadership that is ready to focus on our core mission of offering free, ad-free, volunteer driven hosting to wikis with a lot of options. With a little networking, the finances are a solvable problem.
I'm in talks with the current board to see if a group of current and former volunteers can take over, either by working in the current organization, or starting a new nonprofit and working from there.
I still think Miraheze could have a bright future, and that this was a series of mistakes that got blown far out of proportion to the actual problems that the wiki farm faces.
AMA. I was not involved in the current crisis, so piecing together what happened has been something of a puzzle to me, but I can try to answer most questions.