Y
HN Search
Hacker News Search
new
|
comments
|
top
|
jobs
mlyle
searching PlanetScale…
1.
▲
2.
▲
3.
▲
4.
▲
5.
▲
6.
▲
40 ms
·
781.
▲
by
mlyle
2y ago
If something bad happened to a mostly empty Russian shadow tanker in the Gulf of Finland, that impact is going to be mostly confined to Russia. i.e. past the major Finnish and Estonian ports. As long as we're all playing silly only-ki
782.
▲
by
mlyle
2y ago
I've not been talking about the reconstruction filter at all during any step of this discussion . Reading your comment more carefully, it seems you were trying to. I'm saying that if you oversample, it's easier to get appro
783.
▲
by
mlyle
2y ago
Your comment is mutually contradictory-- what is it, a brick wall (impossible) analog filter or a more gentle rolloff as things fold over? What you really do, these days, is you sample at a higher frequency; you can have an exceptionally ge
784.
▲
by
mlyle
2y ago
Anything close to N/2 is going to have varying magnitude that requires filtering and likely oversampling to remove. How close to the Nyquist bandwidth you can get depends upon the quality of your filtering. 44.1KHz is a reasonable co
785.
▲
by
mlyle
2y ago
Lots of missions use radioisotopic heaters, where you don't bother with the thermocouples and just have the material get warm and protect components which are vulnerable to low temperatures. That's the main reason why spacecraft d
786.
▲
by
mlyle
2y ago
Chips are made in a semiconductor foundry / chip fab. Printed circuit boards are made at a printed circuit board manufacturer / "fab house."
787.
▲
by
mlyle
2y ago
The molecular GRI work in the 90s was really proof of concept work that showed you could maybe build something like this, not something that looked a decade from approval. From what I remember, the early work spun off free radicals, had th
788.
▲
by
mlyle
2y ago
It's always a decade away because the progress promised shows up, and it improves quality of life a bit, but isn't enough and is very expensive. We've gotten a variety of synthetic insulins with different rates that help cont
789.
▲
by
mlyle
2y ago
Absolutely deterministic latency isn't a requirement for the desktop like it is for industrial control. If 99% of the time performance is faster, and the remaining 1% of the time isn't that bad, most people will take the trade o
790.
▲
by
mlyle
2y ago
Aside from infant morality, don't forget the massive death load from things like accidental death, famine, and maternal mortality. E.g. from Wikipedia, female life expectancy from age 15 in Britain in the 1400-1500s century was 33 year
791.
▲
by
mlyle
2y ago
An open source LLM that is being used for inference can't "learn from or remember" interaction patterns. It can operate on what's in the context window, and that's it. As long as the actual packaging is just the mo
792.
▲
by
mlyle
2y ago
I use Linux on mine. It's a bummer that the magic to send over 2 eDP tiled at 5k resolution was never open-sourced, so you can only send 4k and it ends up slightly blurry.
793.
▲
by
mlyle
2y ago
SSL/TLS/etc are libraries, yes. And the certificate store is an OS service. Ancient software has trouble talking to modern services; modern services and devices don't want to fall back to speaking the old versions because of
794.
▲
by
mlyle
2y ago
Mutagenic things are less of a concern when your lifespan is several months.
795.
▲
by
mlyle
2y ago
> > When you have some knowledge of the topology of the graph you can use different techniques to do better than djikstra's. > But that statement doesn't apply to the version of Dijkstra's developed in this paper rig
796.
▲
by
mlyle
2y ago
Sure, even when outer loss is pretty close to zero, it's conceptually not great. On the other hand, I get 400mbps over TCP-over-TCP connections, and can't connect in any other reasonable way. 400mbps > 0. Even tunneling in UDP
797.
▲
by
mlyle
2y ago
But, my connection speed is usually greater and my loss is much less to my VPN endpoint than to whatever services I am accessing though that endpoint. As a result it doesn't affect things much. Further, accessing it with UDP is not a
798.
▲
by
mlyle
2y ago
I'm not talking about towing being advertised as a feature. It's that choosing a vehicle based on something you need to do once every 5 years is not a great way to choose a vehicle. There's not even a single vehicle that I c
799.
▲
by
mlyle
2y ago
Sure. I'm not saying it's completely unreasonable. Here the person was saying "once in the entirety of your ownership". If it's really once in the vehicle's life, then you really should rent something else wh
800.
▲
by
mlyle
2y ago
I disagree. I've never had a vehicle that does 100% of whatever I'd want a vehicle to do. At some point we need to make tradeoffs and accept that we'll either have limitations or need to solve some problems in a different w
801.
▲
by
mlyle
2y ago
https://en.wikipedia.org/wiki/Solar_gravitational_lens > you'd have to get really far back from the Sun to resolve the image, no? Yah, a few hundred AU. > you are beholden to the orbital mechanics of your vi
802.
▲
by
mlyle
2y ago
Possibly, but the challenges to do so are immense. Using the sun as a giant gravitational lens seems much more tractable.
803.
▲
by
mlyle
2y ago
Some of the sources say that he sat in the incident meetings during troubleshooting and adjusted his attacks to avoid detection, too.
804.
▲
by
mlyle
2y ago
A lot of the original social media sources have been pulled, but this is what was alleged on social media: https://juejin.cn/post/7426926600422637594 https://github.com/JusticeFighterDance/JusticeF
805.
▲
by
mlyle
2y ago
In the article's case, there's evidence of actual malice, though-- sabotaging only large jobs, over a month's time.
806.
▲
by
mlyle
2y ago
Oh, I think your TLDR was valid. I just don't think his response to AGI doom is adequate. If cheap, equal-to-human intelligence showed up tomorrow, it would be hugely disruptive to the social order.
807.
▲
by
mlyle
2y ago
I think there's plenty of scenarios where AGI causes doom that's short of extinction but still super bad. I don't know what the probability of these scenarios are. LLMs are remarkably incapable but still seem to be pretty eff
808.
▲
by
mlyle
2y ago
Note that we got a lot of half-baked outcomes and military disasters with manual processes, too.
809.
▲
by
mlyle
2y ago
OK. So another low effort comment on a serious subthread.
810.
▲
by
mlyle
2y ago
You just copy and paste the url, so that the referrer is unset.
More ›