On Tuesday, I travelled by train to Sheffield to take part in the Yorkshire and Midlands Category Theory Seminar #37 meeting, where I would be speaking about my paper that compares partial map classifiers with Sierpiński cones in synthetic (domain/category) theory, which I summarised previously.

Kento Okura has made a lot of progress over the past week in getting Forester’s language server to the point where it can be used. The first editor that we will support is Neovim, which has good LSP support built-in. I think, however, that Kento had not realised quite what a huge amount of work it is to get a working Neovim configuration from scratch that exercises the features of the language server and actually works out-of-the box on other people’s machines. To address this problem, we will be providing a complete working configuration for anyone who wants to use it; experienced users of Neovim will of course prefer to set things up in their own way. Some preliminary code is available, but please stay tuned for further updates that take more advantage of the capabilities of Kento’s language server.

I had a very pleasant lunch in College with Patrick Ferris as my guest; we discussed many things, including the future of Forester and the importance of interop between different authoring tools on the World Wide Web. After lunch, Patrick and I wandered over to Espresso Lane where we had a coffee and a chat with Anil Madhavapeddy and David Allsopp.

Conspiring about the Open Web with my colleagues in the Energy and Environment Group here is making me feel scientifically alive again—there is much to do, and we intend to have fun doing it.

Not many people are aware that the Computer Lab has an old supplier agreement with Fastmail, which has persisted even after the (ill-advised!) transition to Microsoft Office 365 a few years ago. (There is a certain kind of person whom you can always trust to make poor and irreversible technical decisions, and argue for them on the basis of maintainability or security or liability or all of the above! Whenever you refute their technical arguments, there is always an unbounded source of further reasons why it is mandatory and inevitable that we enshittify our own infrastructure, at great cost of course!) Anyway, the savvier members of the Lab have been rocking Fastmail all this time while I have been suffering the constant outages and inconsistencies of Office 365, which is not only a horrible and unreliable product, but also interacts very poorly with standards-based clients other than Microsoft’s own unusable and bloated clients.

A month or two ago, Anil let me on the secret, and I quickly asked the Lab sysadmins to hook me up with a Fastmail account. The process was not completely trivial, as apparently nobody had asked to use this facility for a very long time. But in the end, Piete and Malcom were extremely helpful and I am now up and running with a Lab Fastmail account! PhD students, postdocs, and faculty are all entitled to use Fastmail if they choose, and I strongly recommend it. If you are one of those people, then you have access to this internal page, which contains the instructions for getting an account.

Although we will never be able to get professional services staff off of Office 365 and Teams (this is the sense in which such moves are irreversible), there is absolutely no reason why we have to use it too. I encourage everyone within the Lab to join me on Fastmail, which is extremely reliable and usable. And the more of us who depend on it, the stronger the insurance against forced enshittification in the future.

Fastmail is just the beginning. With Anil and other members of EEG, I am hoping that we can begin the process of taking back control of our internal infrastructure and making it work for us in the way it used to years before I arrived. I am spooked by recent proposals from University IT to drop talks.cam; it seems to me that taking over the administration and maintanance of such critical infrastructure would be a good fit for our capabilities, and I promise that we can do it for less than the millions that the University would pay a vendor to irrevocably enshittify this infrastructure.