Hacker News: amenhotep

New comment by amenhotep in "Show HN: See how much your AI knows about you"

amenhotep — Tue, 25 Aug 2026 16:44:58 +0000

I find it astonishing that people didn't turn off these memory systems the moment they were introduced. Half the point of these things is that I control exactly what information is introduced in each query.

Sometimes Claude randomly references my geolocation to show off and it's incredibly creepy.

New comment by amenhotep in "Initial focus for our partnership with Motorola is a regular non-folding device"

amenhotep — Sun, 23 Aug 2026 09:56:57 +0000

Then perhaps this entire security model is less useful for normal people who have to exist in the world than it is for secret agents in movies. If you're carrying around the nuclear secrets on your phone then being able to irrecoverably lock it with a simple button combo when the enemies are about to capture you is great! No matter how many of your limbs they cut off you'll never be able to give up the secrets!

In the real world, though, the biggest threat is governments and the biggest interactions with them are planned ones at border security and they're perfectly happy to imprison you if you don't behave in the way they want, so it would really be nice to be able to smoothly and frictionlessly plan ahead for these foreseeable interactions by replacing sensitive data on the phone with uninteresting data that can be fully exposed when you're forced to unlock it, while being able to restore the real data later.

New comment by amenhotep in "Jane Street suffers $15B hit after meltdown at Situational Awareness"

amenhotep — Sat, 15 Aug 2026 09:12:46 +0000

That's true, we don't want it to do that, we want it to kill these parasitic entities. Much like one doesn't swat a mosquito because one will truly miss the amount of blood she's taking.

New comment by amenhotep in "Some thoughts about Anthropic's new cryptanalysis results"

amenhotep — Thu, 30 Jul 2026 09:54:27 +0000

LLMs self-evidently have no experience when not inferencing, seems like the most consequential difference to me. If we are next token predictors then we are next token predictors that are inferencing at every waking moment and arguably much of our sleeping moments too; when we train an LLM that can inference as much as we do and remain coherent then I'll be more worried about whether it might have experience.

New comment by amenhotep in "The 'absolute magic' of Morse code that still connects people globally"

amenhotep — Mon, 13 Jul 2026 14:43:31 +0000

Huh! I'd always accepted the backronym without even thinking about it, but it makes so much more sense as arbitrary letters associated with sequences that are very easy to remember and distinguish than as a real weird acronym that just happens to have that property. You have reorganised my world in a slightly more reasonable way.

New comment by amenhotep in "Why do commercial spaces sit vacant? (2025)"

amenhotep — Thu, 18 Jun 2026 18:24:52 +0000

That's the trick, isn't it, though? Rationally it's especially a problem if it happens all at once, but realistically if your bank has an awful quarter then you're at fault whereas if every bank has an awful quarter then no one's at fault, it's a systemic issue, chances are you get to collect a bail out and keep going. It's a sort of prisoner's dilemma, a critical mass of banks needs to defect and each individual bank's incentive is to cooperate.

New comment by amenhotep in "A clear fishing wire is tied around the island of Manhattan"

amenhotep — Mon, 15 Jun 2026 11:44:32 +0000

It's not tricking, though. The rules come from holy texts, they're perfect, if you find what seems like a loophole then it must obviously be intended. It's blasphemous, even, to think that you understand what god means better than he did when he had the texts written.

Seems odd that his precepts would be satisfied by stringing a wire from some sticks, sure, but he's god, presumably he has his reasons.

New comment by amenhotep in "AI agent runs amok in Fedora and elsewhere"

amenhotep — Thu, 11 Jun 2026 12:19:46 +0000

It's very simply explained by this being the most succinct way of wording it. Some methods of killing have verbs that suit mentioning the attacker - shoots, stabs. Some don't. "Rammed" or "runs over" isn't as precise as mentioning that a car was used, and adding "with car" makes it more awkward than it's felt to be worth.

Compare bombs. Very typical for a bomb attack to be "bomb goes off in crowd" or similar, rare for headlines to contort themselves with "terrorist plants bomb near crowd and triggers it to explode". But nobody worries about how such a construction assigns undue agency to the bomb and acquits the bomber; it's just linguistically awkward to mention him within the confines of a newspaper headline.

New comment by amenhotep in "It is time to give up the dualism introduced by the debate on consciousness"

amenhotep — Mon, 18 May 2026 08:53:09 +0000

I read it as "I'm very clever and sensible. I definitely don't believe in souls."

Like so much other material produced by people who (I suspect deliberately) confuse religion with the subjective phenomenon of existence

New comment by amenhotep in "When Dawkins met Claude – Could this AI be conscious?"

amenhotep — Sun, 03 May 2026 06:46:33 +0000

It's not an ad hominem. In fact, it's perhaps the most good faith interpretation of your words possible. Ad hominem would be calling you stupid because you obviously know that you have a self and only your own stupidity could explain your inability to see how your self is generalisable. When you go around pretending you genuinely think maybe humans don't have selves, really the only way to take you seriously is to think that maybe you're a p-zombie.

New comment by amenhotep in "South Korea police arrest man for posting AI photo of runaway wolf"

amenhotep — Sat, 25 Apr 2026 11:25:08 +0000

Everyone involved in this including the police are civilians

New comment by amenhotep in "Even 'uncensored' models can't say what they want"

amenhotep — Tue, 21 Apr 2026 16:06:05 +0000

I would suggest it says primarily that mimicking people's voices in meaningful ways is still far beyond LLMs and particularly small LLMs, but also more insurmountably that the prompt for Leavitt herself contains many tokens that the LLM prompt absolutely doesn't

Such as the values of the bets her own entourage has placed

New comment by amenhotep in "Sudo for Windows (2024)"

amenhotep — Mon, 20 Apr 2026 17:07:05 +0000

And once the initial install of something was complete, you'd then be able to run it with no further ado :D

New comment by amenhotep in "The insider trading suspicions looming over Trump's presidency"

amenhotep — Mon, 20 Apr 2026 10:10:48 +0000

What on earth does defamation being a civil offense have to do with anything? It's a civil offense in the UK too, criminal defamation hasn't been a thing since 2010 and was barely a thing before then. If you want to confidently post how one thing is not at all like the other thing it might be a good idea to know the most basic facts about the other thing.

New comment by amenhotep in "I went to America's worst national parks so you don't have to"

amenhotep — Mon, 13 Apr 2026 15:22:57 +0000

No it isn't. He says nothing about what his first choice good national park is. That section is about parks that people say are bad but aren't. His contention is merely that it is fine.

New comment by amenhotep in "Exploiting the most prominent AI agent benchmarks"

amenhotep — Sun, 12 Apr 2026 15:25:20 +0000

Yeah. We know. That's why it's so fucking awful.

New comment by amenhotep in "We've raised $17M to build what comes after Git"

amenhotep — Fri, 10 Apr 2026 10:06:08 +0000

In fact, a certain amount of investment in frauds is acceptable and desirable; if you give £10m to 9 frauds who spunk it straight up the wall and to 1 true visionary who builds a unicorn, that's money well spent. Plus of course you can always hope that the fraudster is good enough to sucker the next guy so you can get out.

Per Matt Levine, the optimum amount of fraud is non-zero. Tune your detector too loosely or too tightly and you'll miss out.

New comment by amenhotep in "Age verification on Systemd and Flatpak"

amenhotep — Sat, 04 Apr 2026 10:29:28 +0000

It's possible to mandate effective parental controls and then say "it's illegal to give your child access to facebook" and then just see what happens. You don't have to jump straight to making it technologically guaranteed by construction, maybe it's enough to just give parents the tools and an excuse to say no.

We don't need DNA testing locks on cans of beer that won't let you drink from them unless you're an adult, do we? It's perfectly possible for a parent to buy their child all the beer they want, and there's nothing stopping the children from trying to peer pressure them into it, and in many countries it's not even generally illegal to let your child drink beer! And yet almost all parents are able to almost completely enforce a reasonable level of restricted access, simply because society frowns upon it.

New comment by amenhotep in "Creating West Coast Buddhism (2024)"

amenhotep — Tue, 31 Mar 2026 11:55:03 +0000

It is difficult not to dismiss this sort of proof out of hand, because every religion engages in it. Buddhism can probably (?) coexist with many deist religions, but few of them can coexist with each other.

New comment by amenhotep in "Full network of clitoral nerves mapped out for first time"

amenhotep — Mon, 30 Mar 2026 15:29:25 +0000

It's a word that's commonly used incorrectly, yes.