Exactly. LLMs are excellent tools for compressing the time needed to produce a given body of text.
Sure they can also be used for outsourcing the writing process and producing text that doesn't resemble your own voice, just as any tool can be used for multiple purposes. Tools aren't responsible for how particular humans choose to use them.
Your process is similar in spirit to mine, although mine is a bit more fluid. I don't think it's necessary to disqualify words and phrases based on provenance per se, but rather on whether they're genuinely the correct words for what I intend to say and how I intend to say it. That's basically the test I use; if I know that I've intentionally chosen every detail of the text, that I've maintained consistency with my own manner of writing and speaking, and that I've faithfully encoded my own bona fide thoughts, then as far as I'm concerned it's my writing and I see no need to qualify that based on whatever tools may or may not have been used.
> Better driver education and higher test standards would save lives too. So would banning alcohol. None of these solutions, including automated cars, will be mandated any time soon. It's not enough to have data, you also have to have societal buy-in.
Superman: I want to get rid of guns.
Mr. Terrific: For use around the world 24×7, you’d need to replicate the T-Spheres on a massive scale. Still, you could never get rid of all the guns.
Superman: But most. You know we could.
Mr. Terrific: Just because we can doesn’t mean we should. People would resent us.
Superman: And they’d be alive to do that. Check.
Mr. Terrific: Some would fight.
Superman: Let them come then, with sticks. Think of the lives we’d save. Check.
Mr. Terrific: Smoking.
Superman: What?
Mr. Terrific: Cigarettes kill far more. People keep smoking even though they know it’s killing them. Their families can only watch them die.
Superman: And?
Mr. Terrific: You want to save lives? Jail every smoker. Check. And imprison anyone who speeds. Traffic fatalities are huge. Check. Lock up everyone who leaves a dangerous dog unchained. Check. Finally, we kill anyone who doesn’t recycle. Checkmate.
Interestingly, I've been noticing almost the opposite issue. 5.6 Sol frequently starts its responses with "Yes" even when my prompt doesn't contain a yes-or-no question.
I've been noticing the same thing since 5.4- it starts with "Yes" yet I haven't asked a question. I think it might be related to the reasoning, like it's answering its own questions?
It's been happening for years, and it does seem to be entirely related to its background reasoning leaking out of the context and into the output.
The "It's not X, it's Y"-style repetition is another example of that: It argues with itself in the background, and the argument leaks out as if in refutation of things that you (the human) never had imparted into discussion at all.
Agreed, I think it's a mistake to treat "cable" as the entire product category here. It's a (dying) subset of the product category, not the whole category. Like internet and cellular service, SOTA television content as a whole hasn't ever really been cheaper than it is today.
What changed 50 years ago is the US government decided saturated fats were bad and complex carbohydrates were good, and began setting policy to rebuild the food supply and culture around that worldview. We're now living in the result of that population-wide experiment.
No, it was literally government policy. The free market didn't give us the McGovern committee or the following half-century of subsidies, regulations, and guidelines to promote low-fat diets.
Markets don't exist in a vacuum. In this case, it was substantially shaped by state-driven incentives. If nothing had changed, we wouldn't be having this conversation. It's a historical fact that something did change: https://en.wikipedia.org/wiki/Obesity_in_the_United_States#/....
I agree. If we're going the rally the industry to do the work, it should be the whole work in one shot. Any given project/infrastructure that implements both encryption and signing should adopt ML-DSA/SLH-DSA at the same time as ML-KEM, or at least in immediate succession.
My concern is that PQC is having a bit of a Y2K moment, and undercapitalizing on that sense of urgency may risk letting PQ signatures drag on for ages like IPv6. "We need $X engineering budget for PQC" is easy to understand, but "we need $X for PQ encryption now and $Y for PQ signing at some undefined future time" is murkier and may require getting into the weeds on cryptographic concepts and speculative CRQC timelines with non-expert stakeholders.
One of the biggest challenges with the signatures currently standardized is the signature + public key sizes. Demanding we hybridize both just maximizes the pain, and there's no real incentive for this.
Use ML-DSA-44. Don't combine it with other crap. It's good enough.
For KEMs, X-Wing (mlkem768x25519) is great, but ML-KEM-768 and ML-KEM-1024 are also fine on their own. Hybrids are the path of least resistance here, so I prefer them, but have no concerns over ML-KEM's security.
I wasn't implying that the two should be hybridized. I think both are great options to have in our toolkit. For example, in Cyph I chose ML-DSA for end user signing keys + certificates and SLH-DSA for code signing.
No worries, thanks for sharing that post anyway! Another post of yours[1] turned out to be a useful resource for me not too long ago, and the artwork is pretty entertaining.
To be fair, those aren't contradictory positions. I'd rather meth not exist, but given that it does exist, I'd prefer to let that revenue go to Big Pharma than North American ISIS.
(That's before even touching the can of worms of allowing the government to criminalize personal health choices, which feels like a glaring loophole in the Constitution to me.)
It sounds interesting from a hardware perspective, but yeah, IMO no one other than Apple has the luxury of shipping a PC with second-class Linux support anymore. If the Linux experience is anything less than perfect, it's DOA.
Also, USB-A in 2026? Really? That was already an automatic disqualifier for me at the start of the decade.
Agreed. These accounts of people having genuine emotional responses to LLM chats, even going as far as to spend tokens berating them, are very curious. I would be surprised to learn that SOTA models respond optimally to anything other than dispassionate problem-solving, or that scolding per se serves any productive purpose.
Of course we all swear at our computers every now and then, but for me it's always been in good fun. It's just a sarcastic joke that adds some levity and self-amusement to an otherwise arduous debugging process, not generally actual insinuation of malfunction (or malice) on the part of the hardware/OS/toolchain. I'd assumed that "half the job is cursing at the machine until it obeys you" was a big in-joke amongst the profession, but the LLM era seems to be exposing a divide in how tongue-in-cheek that statement really is.
Sure they can also be used for outsourcing the writing process and producing text that doesn't resemble your own voice, just as any tool can be used for multiple purposes. Tools aren't responsible for how particular humans choose to use them.
Your process is similar in spirit to mine, although mine is a bit more fluid. I don't think it's necessary to disqualify words and phrases based on provenance per se, but rather on whether they're genuinely the correct words for what I intend to say and how I intend to say it. That's basically the test I use; if I know that I've intentionally chosen every detail of the text, that I've maintained consistency with my own manner of writing and speaking, and that I've faithfully encoded my own bona fide thoughts, then as far as I'm concerned it's my writing and I see no need to qualify that based on whatever tools may or may not have been used.
reply