By The Sword Linked will end this year. There are more detailed explanations below the cut, but the main reason is that I don’t have time to give the project the work it needs. All contents of the wiki will be permanently archived at Github and reusable under a CC-BY-SA licence. The live wiki will be deleted but I’ll give plenty of advance warning once I decide the date for deletion. Until then there will still be some additions to the wiki. I will also make some other datasets available separately at Github because it won’t be practical to import them to the wiki.

• The core of the project was always intended to be military units and their personnel, but this would need more manual data entry and record linkage than I can do. The inadequacy of existing data makes this project more necessary but less possible.
• Maintaining the software and server is a burden I could do without. When I do get significant spare time to spend on the project, I often have to spend it on upgrading the software instead of adding content. There’s a constant threat that a security hole will require an urgent upgrade at an inconvenient time.
• The wiki relies on extensions as well as the core Mediawiki software. There is no guarantee that all of these extensions will be maintained in future. Even when they are maintained, they often have annoying bugs. Sometimes an upgrade fixes one problem but causes a new one. The wiki also relies on an extension that I wrote myself to display dates. One day this could be broken by changes to the Mediawiki core, and I might not know how to fix it.
• Even if the software that the project currently uses worked as intended, it’s not quite right. The ideal software for this probably doesn’t exist. Semantic Mediawiki and Wikibase each have advantages and disadvantages that the other doesn’t have. In particular, I’ve found naming wiki pages to be a nightmare and would much prefer the Wikibase system of Q numbers, labels, aliases, and descriptions, but Wikibase doesn’t make it easy to embed inline queries or links to queries in the way that SMW does. Wikibase is also so poorly documented from a wiki admin’s point of view that it’s not worth even trying to use it.
• Many external links are now broken because of problems with other sites. The cyber attack on the British Library is still having a major impact after more than a year. The Ordnance Survey has withdrawn its old linked data service and apparently no longer provides a public resolver for TOIDs (this is also affecting Wikidata, and no-one has yet suggested a solution). The owner of the BCW Regimental Wiki died, and although the data has been recovered, the site has not reappeared (this would also be a potential problem with BTSL, but at least the data is easily recoverable and reusable from Github). The Internet Archive is sometimes unavailable because the servers are overloaded. Broken links seem to be an inherent problem with linked data.
• As I’ve probably said before, the project can’t achieve its potential without collaborators, but recruiting and supporting volunteers would add extra complications that I can’t deal with, and would make me less able to create and import more data.
• The wiki is a good way of pulling together information about specific named entities, but it’s not so good for extracting quantitative data for analysis. Being able to export query results helps, but the number of results and levels of recursion both have to be limited for performance reasons, and the SMW query language is not very powerful and has some inconvenient quirks. Extracting data from the RDF or wiki XML dumps into database tables is not straightforward either.
• Even after years of simplifying the data structures and reducing the scope of the project, I worry that I’ve created something so complicated that no-one else will be able to understand it.