Re: What was the Japanese Word? ...
I don't know the word and I can't find it in my dictionary but 袖下 (under the sleeve) might be what you're referring to? Interesting to think there might be an onomatopoeic word for it!
2691 publicly visible posts • joined 8 Nov 2007
IANAL either but I think that you're slightly wrong. AIUI, it's about validity of evidence: if it's gathered illegally then it can't be used against the accused.
I have heard of cases of prosecutors doing such a thing and then using a "parallel" chain of evidence that supports the claims without becoming tainted by the original evidence gathering processes... but that's a whole different kettle of fish.
They can go with the TiKTok model... pick a random "winner", shower them with tokens (or views, in the case of TikTok) and hope that one of those "winners" will hit the jackpot and produce some amazing "AI-"generated content that will turn them into proselytes. All those other users just weren't "AI-"ing enough, doncha know!
1. Users set up an identity that can be used with ZK proofs (eg, a directed graph with a Hamiltonian cycle only known to the user)
2. Users authenticate this identity with a government agency (along with documentary proof of identity), which provides a "known to us" certificate
3. User requests some sort of permission/certificate linked to the presented (and authenticated) identity
4. Government (or their proxies) issues a certificate saying something like "if this user can produce an authentic "known to us" cert, they're over 18"
5. At point of use, the person anonymously (and with ZK) proves that they have the given permission to the service provider
No need for third party databases. Just PKI (like how browser certs are managed) and ZK proof mechanisms.
Or even go with something like fucking Kerberos. The technology we're talking about is ancient.
https://karpathy.github.io/2026/02/12/microgpt/
Covers the basic idea of how LLMs work including tokenisation, "attention" and how networks are trained to work in "chatbot"/"Eliza" style. It does assume that you know at least the basics of multi-layer neural networks, but you don't need to know much and those basics have been around practically forever so at least most people here will be able to follow it. It also covers how the largest LLMs are basically no different from the ~200 line Python code presented; the only differences being number of parameters and different optimisation strategies.
It should be clear from both the code and discussion in the linked article that talk of "consciousness" (emergent or not) or "gender" or "personality" with respect to LLMs is (and no doubt shall continue to be) a category error.
Minor pedantry alert... the Japanese is "yurusenai"...
The English translation is close enough to "I can't forgive you", so that's cool. An alternative translation might be "intolerable!"
We've had the long-ish period from that guy from Sun (I think it was) saying that privacy is basically dead to the current time where breaches like this have especially become criminals stealing from other criminals. William Gibson nailed it... the future is here, but it's just not evenly distributed.
I wish I could give a cheery outlook and say that everything is going to be all right, but I'm sort of equivocal on that. At least I can offer some wisdom from Leonard Cohen: "There's a crack in everything... that's how the light gets in." ("light" is kind of cheery, no?)
Stay safe, be well, y'all.
It's not only that. If you want to be a "scofflaw" (as per the article) in Japan, you either strike out on your own (high risk, high reward, particularly attractive to western immigrants) or roll in with the existing local grey/black economy. (or do some other lower-level shit like promoting bitcoin or selling pornography). And if you're a non-western immigrant, you just keep your nose down and do your regular (healthcare, restaurant, combini, etc.) job so that you can eventually gain citizenship (and besides, you're not going to be able to change jobs without putting your visa at risk).
This foreign attempt at setting up a new grey economy is destined to fail.
THE UNIONS FOREVER! DE-FEND-ING OUR RIGHTS. POWER TO THE FACT-RIES! THE WORRRKERS UNITE!
"Down with the blackleg," surely? (I can't believe I had to apply the strike tag and all)
(Yes, "There is power in a factory, power in the land, power in the hands of a worker" but you are misquoting)
Unfortunately, not all apps implement shift-insert in the same way. I would love it if all apps pasted what you have just highlighted (ie, do the same as mouse middle click), but Firefox, for one, pastes the contents of the main copy/paste buffer.
At least shift-insert works the way I want in terminal programs. It's particularly handy for mouse-free copy and paste between an emacs buffer and a terminal. Still very useful if I have to use mouse to select text in other source apps.
I am rather curious as how the taishoku daiko perform their task. In person? Can imagine it would be excruciatingly formal.
Not only in terms of business etiquette, there's also a higher level of language formality to deal with. Sounds like a difficult job all right.
Apologies JP speakers if I have commited a "the hoi polloi" on "the taishoku daiko."
No, it's not a "the hoi polloi" error on your part ("hoi" literally translating to "the?"). Japanese has no definite or indefinite articles, so when translating from Japanese to English it's OK to add one or some other surrogate word ("these", "such") as appropriate.