zbentley

31 ms·

511.
by zbentley 9mo ago
This was on the front page yesterday, which sounds related: https://news.ycombinator.com/item?id=46875228
512.
by zbentley 9mo ago
And heat and pressure. Negligible amounts in terms of the biosphere, but not in terms of flora and fauna in proximity to launch sites.
513.
by zbentley 9mo ago
If you use rot13 as your data encryption, then a ten year old can absolutely break it--not by knowing what "rot13" or "data encryption" mean, but by clicking a button in a one-click jailbreak app. On the other hand, if y
514.
by zbentley 9mo ago
> Why? Presumably to secure the company selling the device’s revenue stream. There’s a big difference between “any 10yo with $5 for an SD card can download a one-click app and jailbreak our projector” and “you have to be fairly technical
515.
by zbentley 9mo ago
Misinformation and mass hysteria suck, I agree. But if the amplification of the sky-is-falling-flavor-of-the-week braindead youtuber take can materially imperil financial markets, the stability of that system was always doomed. An “SEC for
516.
by zbentley 9mo ago
> you need a non technical person put the brakes on some technical teams. Can you expand on that? The general wisdom, true most places I and my peers have worked, is that non-technical business stakeholders are often the ones deprioritiz
517.
by zbentley 9mo ago
> Security has certain cost associated to implement it The article makes a strong case that, at least for minimum viable/ordinary security measures, the cost is $0. The projector in question wasn't missing features that would h
518.
by zbentley 9mo ago
> Data center in space could contribute to the EU's Net-Zero goal by 2050 How unbelievably crass. "Let's build something out of immense quantities of environmentally-destructive-to-extract materials and shoot it into space
519.
by zbentley 9mo ago
And 2.5s is best case . Signal strength issues, antenna alignment issues, and all sorts of unknown unknowns conspire to make high-integrity/high-throughput digital signal transmissions from a moon-based compute system have a latency m
520.
by zbentley 9mo ago
In addition to all the sibling comments explaining why this wouldn't work, the money's not there. A grift the size of Dogecoin, or the size of "free speech" enthusiast computing, or even the size of the criminal enterpri
521.
by zbentley 9mo ago
Similarly fair point. I think there's a wide regulatory spectrum available between "ban something outright" and "completely lassiez-faire noninterventionist approach". For stock buybacks, that spectrum includes (and
522.
by zbentley 9mo ago
> RIP anyone who had a penchant for "not just x, but y" though I felt that. They didn’t just kidnap my boy; they massacred him.
523.
by zbentley 9mo ago
Only that they are all quite new, but are frequently defended as if they are integral components of the country since time immemorial. As you say, on balance some of them are good/should be preserved, and some should be reconsidered.
524.
by zbentley 9mo ago
See also: DHS/TSA/ICE, credit scores, stock buybacks and self-dividends. All very, very recent systems in the grander scheme.
525.
by zbentley 9mo ago
Agree to disagree. I think that the nature of a given instance of a programmatic error as something that, once fixed, means it stays fixed is significant. I also think that if we’re assessing the likelihood of the entire SDLC producing an
526.
by zbentley 9mo ago
I meannnnnn… Pedantically, the “stolen valor” metaphor absolutely doesn’t fit here; you’re just griping about the “sudo brand” being used in another project’s title (which … citation needed, and so what? Is “doas” not committing theft but “
527.
by zbentley 9mo ago
Yes, because private companies aren’t subject to the PRA like federal agencies are. FWIW, I’m not sure that TSA ID verification does indeed have a PRA claim in the way the article asserts, and I am very sure that the PRA is a dumb law tha
528.
by zbentley 9mo ago
It’s a sticky issue. Without QI, it seems very plausible that many law enforcement departments would be seriously hamstrung by continual waves of legal action and thus cost taxpayers a lot more to operate effectively. Not only would many pe
529.
by zbentley 9mo ago
Poor people already cannot afford most legal services, and (other than a very small percentage of cases brought on a volunteer or nearly-volunteer basis at the goodwill of law firms) very rarely file suit without legal fees sought in restit
530.
by zbentley 9mo ago
Effectiveness and theatricality aside, that wouldn’t work: the risk that the TSA ostensibly controls for is primarily that of planes being used as weapons against non-passengers, and only secondarily passenger security/hijacking.
531.
by zbentley 9mo ago
how does the metaphor of stolen valor (in my understanding: claiming accolades or military credentials/decorations that one never received) apply to that project? I don’t know anything about the history here; it’s a genuine question.
532.
by zbentley 9mo ago
I think that GP meant that MacOS has a constant nonzero rate of platform churn. I might be wrong though!
533.
by zbentley 9mo ago
> the reality is that we are mixing instruction and data in the same context window. Absolutely. But the history of code/data confusion attacks that you alluded to in GP isn’t an apples-to-apples comparison to the code/data con
534.
by zbentley 9mo ago
Good idea! Unfortunately that requires classical-software levels of time and effort, so it's unlikely to be appealing to the AI hype crowd. Such a supervisor layer for a system as broad and arbitrary as an internet-connected assistant
535.
by zbentley 9mo ago
> with proper security controls on it That's the hard part: how? With the right prompt, the confined AI can behave as maliciously (and cleverly) as a human adversary--obfuscating/concealing sensitive data it manipulates and so
536.
by zbentley 9mo ago
That's not even slightly the same thing. A buffer overflow has nothing to do with differentiating a command from data; it has to do with mishandling commands or data. An overflow-equivalent LLM misbehavior would be something more l
537.
by zbentley 9mo ago
Oh, I don't think we should change the semantics of "rm"--not because reversibility is unimportant (shameless self-promotion: it is: https://blog.zacbentley.com/post/on-reversibility/ ), but because
538.
by zbentley 9mo ago
In my case, I’d use a ZFS snapshot. Many equivalent tools exist on different OSes and filesystems as well.
539.
Ingress Nginx: Statement from the Kubernetes Steering and Security Committees(kubernetes.io)
2 points by zbentley 9mo ago|0 comments
540.
by zbentley 9mo ago
I think you misunderstood the first half of my comment. Many of those things have an asterisk after “synchronous”, but in their default mode they often are in fact sync. Saying “technically commands are posted and later observed, and that’s

More ›