It's been over 6 years since I moved in here and I've only just realised that the reason the lights in the extractor hood don't work is that nobody had ever installed any bulbs in it
Notices by Matthew Garrett (mjg59@nondeterministic.computer)
-
Embed this notice
Matthew Garrett (mjg59@nondeterministic.computer)'s status on Thursday, 25-Jun-2026 15:58:30 JST
Matthew Garrett
-
Embed this notice
Matthew Garrett (mjg59@nondeterministic.computer)'s status on Monday, 22-Jun-2026 16:09:54 JST
Matthew Garrett
The El Toroito specification for bootable CDs was named because it was originally sketched out on the back of a napkin at an El Torito restaurant. If you had to name a spec you'd written after the bar or restaurant you came up with the idea in, what would its name be and what was the actual name?
-
Embed this notice
Matthew Garrett (mjg59@nondeterministic.computer)'s status on Wednesday, 17-Jun-2026 14:42:44 JST
Matthew Garrett
LLM-written code is, unsurprisingly, in the absence of human review frustratingly annoying to modify by hand
-
Embed this notice
Matthew Garrett (mjg59@nondeterministic.computer)'s status on Friday, 22-May-2026 10:44:34 JST
Matthew Garrett
People on the "LLMs mean all bugs need to be patched immediately" train: are you replacing your ops teams with LLMs as well or have you just forgotten why Patch Tuesday is Good, Actually
-
Embed this notice
Matthew Garrett (mjg59@nondeterministic.computer)'s status on Saturday, 16-May-2026 17:35:43 JST
Matthew Garrett
It is very funny that there is a guy who spends a lot of time talking about his commitment to facts, and who also keeps making references to me swearing on a bible in court despite (a) that not happening and (b) the transcripts making it clear that didn't happen
-
Embed this notice
Matthew Garrett (mjg59@nondeterministic.computer)'s status on Monday, 11-May-2026 23:03:20 JST
Matthew Garrett
DRM is pretty obviously something that inherently removes user freedom without benefit, and decrying it is entirely reasonable. Hardware identity and state attestation *can* be used for DRM, but can also be used for other purposes that improve things for users (like Signal verifying that it's communicating with a genuine enclave before disclosing any sensitive data), and attacking the technology rather than the ways it's used seems short-sighted
-
Embed this notice
Matthew Garrett (mjg59@nondeterministic.computer)'s status on Monday, 11-May-2026 23:00:18 JST
Matthew Garrett
People will complain that a technology can be used to oppress user freedom while contributing to free software that gets used in literal weapons of war
-
Embed this notice
Matthew Garrett (mjg59@nondeterministic.computer)'s status on Monday, 11-May-2026 23:00:17 JST
Matthew Garrett
I do entirely understand the idea that functionality that can be used against users (even if it can also be used to enhance user security) is bad, I just don't understand why people will simultaneously make that argument and support the idea that a software license that says "You may not use this software to murder people" is incompatible with the ideals of free software
-
Embed this notice
Matthew Garrett (mjg59@nondeterministic.computer)'s status on Friday, 08-May-2026 01:50:13 JST
Matthew Garrett
@Suiseiseki your storage is on a device that runs free software? I would like to learn more.
-
Embed this notice
Matthew Garrett (mjg59@nondeterministic.computer)'s status on Friday, 08-May-2026 01:18:51 JST
Matthew Garrett
The downside of full-on self-hosting is, of course, that your server will inevitably choose to have weird hardware issues while you're in another country
-
Embed this notice
Matthew Garrett (mjg59@nondeterministic.computer)'s status on Thursday, 30-Apr-2026 08:07:04 JST
Matthew Garrett
A bunch of time trying to figure out why I was getting 500 errors any time I tried to type a username into Mastodon and it turns out Elasticsearch believed the index was corrupt but on validating it it seems like it wasn't and just deleting the "corrupt" tag in the index was enough for things to start working again? I'm sure this will all be fine
-
Embed this notice
Matthew Garrett (mjg59@nondeterministic.computer)'s status on Monday, 27-Apr-2026 17:40:02 JST
Matthew Garrett
Speaking of Unicode, I think many people would be surprised just how strong the pushback against it was even in the early to mid 2000s. UTF-8 adoption in Linux was a fight.
-
Embed this notice
Matthew Garrett (mjg59@nondeterministic.computer)'s status on Wednesday, 22-Apr-2026 15:38:34 JST
Matthew Garrett
Garrett Hardin: I described the tragedy of the commons as a warning against a system that drives people to consume common resources for personal benefit
AI companies: Good news, we have implemented the tragedy of the commons as a service -
Embed this notice
Matthew Garrett (mjg59@nondeterministic.computer)'s status on Monday, 20-Apr-2026 05:44:40 JST
Matthew Garrett
Palantir telling their foreign customers that their priority will always be to defend the USA is certainly a choice
-
Embed this notice
Matthew Garrett (mjg59@nondeterministic.computer)'s status on Monday, 20-Apr-2026 05:28:15 JST
Matthew Garrett
@glyph I understand your point and to me it does feel like there's a real difference that I'm not expressing terribly well. Words have a meaningful impact on how the story lands, and that just doesn't feel true for most code? In general I want code that clearly communicates the functional goal, not code that seeks to accentuate that through style.
-
Embed this notice
Matthew Garrett (mjg59@nondeterministic.computer)'s status on Monday, 20-Apr-2026 05:28:13 JST
Matthew Garrett
@jwz @glyph Fair point, and also obviously commit messages play into this. If LLMs are tending to churn out people's comments I think my argument ends up massively weaker.
-
Embed this notice
Matthew Garrett (mjg59@nondeterministic.computer)'s status on Sunday, 19-Apr-2026 11:47:04 JST
Matthew Garrett
@glyph I think I've covered why the plagiarism bit feels less true to me for code than for other fields, and I don't think the error prone aspect of it matters for the cases I'm thinking of. The world burning and economic destruction and loss of human skills are certainly a consequence of how these things are currently deployed but it's not inherent (at least, not to anywhere near this scale), and having it be an immediate "no" rather than "Is there an ethical way to do this" feels rough
-
Embed this notice
Matthew Garrett (mjg59@nondeterministic.computer)'s status on Sunday, 19-Apr-2026 03:54:38 JST
Matthew Garrett
@p If you're doing something other than
var++
then you're doing something wrong. Code is instructions to a machine. The description of what that code does may be creative, if the actual implementation is then you are almost certainly in a bad place.
-
Embed this notice
Matthew Garrett (mjg59@nondeterministic.computer)'s status on Sunday, 19-Apr-2026 03:16:07 JST
Matthew Garrett
Free software people: A major goal of free software is for individuals to be able to cause software to behave in the way they want it to
LLMs: (enable that)
Free software people: Oh no not like that -
Embed this notice
Matthew Garrett (mjg59@nondeterministic.computer)'s status on Sunday, 19-Apr-2026 03:16:05 JST
Matthew Garrett
@radex See I fundamentally don't believe that code should be copyrightable and also me 30 years ago did not produce code that was suitable for professional use but it fixed my problems anyway