Jeff Duntemann's Contrapositive Diary Rotating Header Image

mediawiki

The Trouble with Wikis

Mar 14th, 2021
by Jeff Duntemann.
5 comments

A week or so ago I bestirred myself and installed MediaWiki on my Web host. I’d been intending to do that for some time, but (as my friend Don put it) my life was ODTAA for a bit. Installing it was a snap. My provider has something called Installatron that did the job, no issues. The software, of course, is free and open-source.

I installed it in part to become more familiar with the MediaWiki system. As usual, when installing something new, I went up to Amazon and checked for books on MediaWiki.

Unless I missed something, there are five.

Plus a few more in French, German, and Japanese. Furthermore, those five books did not all get favorable reviews. The title I was most interested in is now 11 years old and way behind the current release of MediaWiki. (I ordered it anyway, along with O’Reilly’s MediaWiki: Wikipedia and Beyond, which is even older.)

My first question was: Why so few books about software this famous?

The answer came to me slowly: Almost nobody wants to create/maintain/populate their own wiki. MediaWiki is famous for one reason: Wikipedia. I’ve seen a number of other public wikis, including Fandom.com, Conservapedia, Everipedia, WikiHow, Wikispecies, and WikiTree. There is a list on Wikipedia that eyeballs at about 80. Let’s be generous and triple that to account for wikis that Wikipedia didn’t list, and for private wikis. So, say, 250. That’s not much of a market for books. Even 500 installs would not float a print book.

MediaWiki’s online presence has a feature for creating a downloadable PDF version of the MediaWiki documentation, but it’s currently disabled. Sheesh.

Having gone crosseyed reading about it online, my conclusion is that MediaWiki is a bit of a hot mess. That said, I should tell you all why I even bothered: I want to create a wiki for my fiction, and especially about the Gaeans Saga, which includes the Metaspace books and the Drumlins books. I’ve done a little wiki editing, and have a couple of decent books on my shelf about creating content on Wikipedia. The trick to creating content on wikis is having a group of content templates and knowing how to use them. If you look at the page source for any Wikipedia article, the problem becomes obvious: The stuff is crawling with templates, and for the most part they’re templates that don’t come with the generic MediaWiki install.

I discovered this by opening an edit window for Wikipedia’s article on the star mu Arae, which in my Metaspace books is the location of Earth’s first colony. I loaded the whole wad onto the clipboard and dropped it into a new page on my MediaWiki instance. A few of the templates were present on MediaWiki. Most were not, and the article incorporated dozens. I went back and lifted the source for 47 Tucanae. Same deal.

Now, Wikipedia content is available under Creative Commons. Grabbing the articles is easy and legal. I soon found after googling around for awhile that grabbing the templates, while legal, is not easy. Some templates are actually contained in libraries written in…Lua. I have some sympathies for Lua, which strongly resembles Pascal. It made me wonder, however, why a formatting template needs to make calls into a code library. As best I know, this is something specific to Wikipedia, and is not present in the generic MediaWiki.

I like the overall look of Wikipedia. People are used to it. I’d like to incorporate that design into my own instance of MediaWiki. I wouldn’t need all the templates, though some would be damned useful. That said, I see no reason why some sharp MediaWiki hacker couldn’t gin up an installer for all of Wikipedia’s templates, no matter how many there are. Maybe such a thing already exists, though I think that if it did, I would have found it by now.

There are other projects needing my attention, so I’m going to set this one aside for awhile. Obviously, if anybody reading this knows where to find an installable collection of Wikipedia’s templates, give a yell.

The All-Volunteer Federated Encyclopedia of (Really!) Absolutely Everything

Aug 9th, 2020
by Jeff Duntemann.
6 comments

My regular readers will recall that I wrote an article in the June/July 1994 issue of PC Techniques, describing a distributed virtual encyclopedia that pretty much predicted Wikipedia’s function, if not the details of its implementation. My discontent with Wikipedia is not only well-known but not specific to me: The organization has become political, and editor zealots have various tricks to make their ideological opponents either look bad, or disappear altogether. Key here is their concept of notability, which is Wikipedia’s universal excuse for excluding the organization’s ideological opponents from coverage.

In one of the decade’s great hacks, Vox Day created Infogalactic, which is a separate instance of the MediaWiki software underlying Wikipedia and a fair number of other, more specialized encyclopedias. Infogalactic has a lot of its own articles. However, when a user searches for something that is not already in the Infogalactic database, Infogalactic passes the search along to Wikipedia, and then displays the returned results. I don’t know whether or to what extent Infogalactic keeps results from Wikipedia on its own servers. It’s completely legal to do so, and they may have a system that keeps track of frequent searches and maintains frequently searched-for Wikipedia pages in local storage. Or they may just keep them all. We have no way to know.

Infogalactic’s relationship with Wikipedia immediately suggested a form of federation to me, though Infogalactic does not use that term. (Federation means a peer-to-peer network of nodes that are independently hosted and maintained yet query one another.) The Mastodon social network system is the best example of online federation that I could offer. (It’s not shaped like an encyclopedia, so don’t take the comparison too far.) There is something else called the Fediverse, which I have not investigated closely. In a sense, the Fediverse is meta-federation, as it federates already federated platforms like Mastodon. For that matter, Usenet is also a form of federation. It’s been around a long time.

The MediaWiki software is open-source and freely available to anyone. There are a lot of special-interest wikis online. One is about Lego. (Brickipedia, heh.) For that matter, there’s one about Mega Bloks. Hortipedia is about gardening and plants generally. It’s a huge list; give it a scan. You might find something useful.

My suggestion is this: Devise a MediaWiki mod like Infogalactic’s, but take it farther. Have a “federation panel” that allows the creation of lists of MediaWiki instances for searches falling outside the local instance. A list would generally start with the local instance. It might then search instances focusing on related topics. The last item on most lists would be a full general encyclopedia like Wikipedia or Infogalactic.

Here’s a simple example, which could defeat Wikipedia’s notability fetish for biographies and a lot of other things: Begin a search for a given person (or other topic) with Infogalactic, which, remember, searches Wikipedia if its own database doesn’t satisfy the query. So if that search fails, submit the same search to EverybodyWiki, which doesn’t apply notability criteria to biographies. In fact, EverybodyWiki does what I suggested be done a number of years ago: It collects articles marked for deletion on Wikipedia, of which it currently has over 100,000. I’m tempted to post a biography on Wikipedia just to see if, when it’s deleted (and it would be) EverybodyWiki picks it up.

(As an aside: I just found EverybodyWiki a month or so ago, and I’m surprised I hadn’t heard of it before. It has more than just biographies and is definitely worth a little poking-around time.)

Now, the tough part: How would this be accomplished? I don’t know enough about MediaWiki internals to attempt it myself. There’s an API, and I’ve been surfing through the API doc. There’s even an API sandbox, which is a cool idea all by itself. Alas, there are remarkably few technical books on MediaWiki, and the ones I would be most interested in get terrible reviews. Given how important MediaWiki is, I don’t understand why tech publishers have skated past it. My guess is that few people bother to do more than custom-skin MediaWiki. (There’s a book on that, at least.) If the demand were there, the books would probably happen. If you know enough about MediaWiki modding, I’ll bet you could find a publisher.

I’m thinking about installing MediaWiki on my hosting services, just to poke at and try things on. Hell, I predicted this thing. I should at least know my way around it.

If you’ve done any hacking on MediaWiki, let me know how you learned its internals and what you did, and if there are any instructional websites or videos that I may not have encountered.