Last Notes

- reply

on a tangent, I continue to be horrified by the habitual euphemisms that make English nigh-impossible to learn. TIL that "every position is fully funded" means "all positions are entirely unfunded and students or their professors need to bring or raise their own".

2026-05-18 18:23:23 CEST

- reply

ITYM https://DublinTechWeek.com

2026-05-13 08:33:25 CEST

- reply

the only way to enforce age verification of minors is through total internet supervision of everybody. And it won't help with any of the problems on the internet, because facilitating and effectively rewarding terrible behaviour optimises the revenue of the main players.
If you want a better internet, step one is to ban behaviour-based profiling for advertising, with no "consent" loophole. It makes surveillance ("tracking cookies") immediately illegal (no "legitimate interest" excuse, GDPR does the rest). That trashes the direct "more bullying and hate leads to greater revenue" connection, which immediately removes the incentive to put active effort into more vicious online spaces. This applies to Google, Meta, X, Reddit, etc.
Note that the "consent is not an excuse" part is critical. "Ask me later" just means that people are worn down into clicking "yes" eventually, just to get it out of the way, you know yourself the dark patterns, you know yourself the cookie consent forms, where "none of them" is tedious and must be repeated on every visit, but "yes please" is simple and eternal. Try changing that decision later.
Ban the surveillance. Yes, it's many companies' entire business model. That's too bad for them, should have tried being socially positive. Yes, many services will have to start some sort of subscription or pay per use model instead of being fake-free. That's OK, they'll also be disincentivised to enshittify.
It'd make the internet better for everyone, including children, who could continue to find and create so many positive communities online, instead of being blanket-banned.

2026-04-07 01:21:56 CEST

- reply

wenn du 100% im Google-Ökosystem bleibst, ändert sich nichts, richtig.
Wenn du das nicht tust - und das war ja Grundlage der Frage, was ist, wenn du GrapheneOS installierst - dann gibt es durchaus Unterschiede.
Die Ankündigung von Google ist natürlich inhaltlich richtig, sie ist aber natürlich auch unvollständig, denn natürlich gibt es auch Nachteile dieser Änderung. Nicht für Google, klar, aber für Benutzer und Entwickler, die nicht wollen, daß Google genau weiß, wann und wo das Telefon ein- oder ausgeschaltet wurde, welche Apps um welche Zeit für wie lange geöffnet waren und wieviel damit interagiert wurde, wo und wie sehr das Gerät zu welcher Tageszeit bewegt/geschüttelt - also herumgetragen - wurde...
Wer das alles nicht will, oder aus irgendeinem anderen Grund GrapheneOS installiert, der darf und muss wissen, was sich ändert.
Und das ist zur Zeit nicht abschließend zu beantworten, denn es hängt ab von Implementierungsdetails in den Play Services einerseits und von den API- und Store-Entscheidungen der App Entwickler andererseits.
"Nix" ist jedenfalls falsch. Bestenfalls "nix was diese konkrete Person beeinträchtigt".
"Lies die PR/Doku", egal ob die von Google oder die von GrapheneOS oder beide, ist abfällig und abwertend. Und auch falsch, weil ohne zusätzliches Hintergrundwissen beides keine Klarheit schafft.

2026-03-19 20:55:26 CET

- reply

what I love doing in cities that have them (this includes London, not sure about Edinburgh) is to ride one of the hop-on-hop-off tourist buses. They'll do a trip around a bunch of places, and tell you something about them, so you know what you're looking at, and one or two extra random facts. You can get off and investigate, and get on the next bus in the same route later. Or just stay on the whole tour, whatever you feel like on the day. For me, it's an excellent way to get to a bunch of the "must see, but don't know if I want to go, probably wouldn't make the extra effort" places in basically a day, with no own planning required.

2025-12-04 22:39:35 CET

- reply

no.
Specifically, none of these.

2024-09-12 02:05:48 CEST

- reply

you call the Ambulambulance when the Whaaaambulance would take too long, duh. :)

2024-08-17 14:24:08 CEST

- reply

cool. So we can merrily agree that I can modify and run an AGPL'd driver without necessarily making the modified source available, much like straight GPL or MIT?
I don't want to attempt to argue platonic ideal Free software dos and don'ts. The AGPL has different issues to other licenses, it doesn't perfectly achieve its goals, but also doesn't miss them entirely; it's possible to comply with it, maybe breaking git-specific workflows (meh) and some business models (WAI).

2024-08-17 12:46:14 CEST

- reply

The parenthetical is absolute. An AGPL video card driver does not provide network services in any reasonable-person sense, so that part of section 13 is not relevant. Combining that driver as an ML thing as your service is subject to the second half, but that's a completely unrelated topic.

2024-08-17 12:42:05 CEST

- reply

no, the preamble provides context in which the license is to be read. Just like the relationship between any involved legal or human persons provides context. Trivial example, if two people working for corp A modify some software for that corp, they're acting as agents and aren't (generally) individually on the hook for license compliance.

2024-08-17 12:37:59 CEST

- reply

Person B can run the unmodified AGPL licensed software without even accepting the license. This is also the case with any GPL software, explicitly.
If Person A under sec 5 made the necessary notification updates, they conveyed a compliant covered work to Person B. So neither is in violation.
That's maybe surprising. But if A and B aren't mutually at arms length, we're back to mens rea; if they are, then A's version is likely available separately anyway, so it's fine.

2024-08-17 02:04:59 CEST

- reply

someone else already commented on the bad faith proxy being an obvious mens rea demonstration, so that's a bit of a straw man.
But your GitHub comment is also wild, for two reasons.
1) If my fork is public and I thus provide modified source code, I have no users of my program, as I am not running it, so there is no need for a quine feature to exist (which wouldn't be usable anyway, because the program isn't running to, well, run it).
(cont)