Log in
If you're old enough, you might remember using floppy disks, either of the 3.5-inch or 5.25-inch variety. They didn't hold much and often you had to have many disks to install one program. Don't be misled by Fits on a Floppy's retro-tech name: it is most definitely not about 20th century data media. It's about compactness and comprehensibility. Fits on a Floppy describes itself as "a Manifesto for Small Software," and as we read it, we found ourselves nodding in agreement, right from the opening line: That is certainly the impression of this author, and it is not just us. We are irresistibly reminded of the Red Hat developer's six waves of industry BS that we recounted in February. Like any eternal verity of the computing industry, there's even an XKCD comic about it, if you needed any more persuading. XKCD's own internal citations, both about voting machines and indeed about the use of the blockchain, reinforce the message. Randall Monroe spells it out: That sounds about right. And parenthetically, anyone who says that they can improve anything with either blockchain or AI is no more to be trusted than a schoolteacher who gives no homework. A year before that Red Hat engineer talked to us about waves of industry BS, Dutch consultant Bert Hubert talked to The Reg about digital sovereignty – and he feels similarly. In 2024, he wrote A 2024 Plea for Lean Software in tribute to the great Niklaus Wirth, who passed away earlier that month. In our own obituary for Professor Wirth, we mentioned the 1995 paper that inspired Hubert: A Plea for Lean Software (this is a PDF of a scan – but we have posted a more readable plain text version here). One of the lasting effects of that paper is what is now called Wirth's Law: Sadly, that seems to have been its main impact. As a usable demonstration of his 2024 "plea," Hubert offered a working example, which he explained in Trifecta Technology. It's a web image-sharing tool, implemented in under 2000 lines of code. There's also a page for the Trifecta app itself, which comes in as a 1.7 MB compressed Docker file. (With some clever disk formatting, you could get that on a 1.4 MB floppy.) There are, as that anonymous Red Hatter observed, so many different layers of unnecessary complication and plain old marketing lies in modern IT that it is now hard to even keep track of them. One point of the Fits on a Floppy idea is that if you impose an artificial limit on project size, merely by keeping it very small, you will be forced to keep it very simple. That simplicity is the goal here, not fitting on 1980s physical media. You might react with scorn when you hear the idea that in the 2020s, anything useful could fit into under 1.5 MB. When even a leading tool to write an ISO file onto a USB key is a hundred times that size, it sounds absurd. But it really is not. The mind behind the manifesto is developer Matt Sephton, and he offers 18 tiny but useful apps that he's written to prove his point – plus a screensaver which we feel sure is an hommage to Berkeley's classic Flying Toasters screensaver. Others are still making useful single-floppy-sized apps today. We wrote about the revival of the Dillo web browser, and at last year's FOSDEM, the project lead was handing out floppies with the latest release. The whole app, on one diskette. Drew DeVault's Hare programming language is still in development, but when it reaches version 1.0, he plans to sell copies on a floppy: Another tiny modern language is the Janet Language. It's not quite so small, but its just over 2MB download could fit onto the 2.8 MB floppies that were used in later IBM PS/2 models and the NeXTstation. The real point here is about the readability and long term maintainability of compact, even minimalistic code. It's a similar point to that made in Dave Gauer's Ascetic Computing essay, which we cited and linked to when looking at OpenBSD 7.9. Small size and simplicity is what Fits on a Floppy is really talking about, not about physical media. He explicitly spells it out for the hard-of-thinking: Bert Hubert too returned to this theme when he wrote a piece On Long Term Software Development. At this year's Open Source Policy Summit, we saw some pundits pontificating that to escape the US cloud, the answer was that Europe needed its own companies running their own datacenters running Europe's own domestic cloud. This is so manifestly Getting The Wrong End Of The Stick that it put us in mind of Wolfgang Pauli's famous line: "That is not only not right; it is not even wrong." The way to escape a broken model that was a bad idea in the first place is not to make your own sovereign version. All you're doing is locking yourself in your own personal cage. The smart answer is to discard the broken model, and go back to an older, simpler model where organizations own and store their own data on their own servers. As ever, the KISS Principle is one of the best guidelines. It's Occam's Razor in reverse: the best solution is the simplest possible solution. If the problem is that you are trapped in someone else's cloud, then don't switch to another cloud and risk it happening again: get your private data out of the cloud altogether. Just Use One Big Server. Hire some grumpy old techies with grey hair (or none) to run it – there are plenty out there, but ageism keeps them out of work. At the smallest and most local end of this scale, then one useful guiding principle is to just keep the tools as small as you can possibly make them. It's an artificial limit, but that does not lessen its validity. It's not the only way. It may not even be the best way. But it's a way, a simple, clear, obvious way – and there's nothing to prevent anyone finding their own different path to radical simplicity. The PC rose to greatness running on two 360 kB floppy drives – hard disk drives only came along later. Tools like Lotus 1-2-3 redefined business management running on one 360 kB disk, with a second 360 kB data disk in drive B: – and this vulture is willing to bet that some spreadsheets built on such machines, long since converted to Microsoft Excel, are still running multinationals, and indeed nations, today. Compared to that, 1.4 MB is luxurious. ®
