dgroshev

31 ms·

361.
by dgroshev 3y ago
Yes, but collectively we understand that SO or Reddit are unreliable, plus it's usually one-way and we don't see people posting bad advice defending it just for us. LLMs, on the other hand, are extremely convincing, while being ev
362.
by dgroshev 3y ago
You are talking about most obvious hallucinations such as inventing a non-existent API. However, those are indeed trivial. What's not trivial is not using APIs that exist for your particular problem; using obsolete patterns; not recogn
363.
by dgroshev 3y ago
It's not just things that can be caught with tests. You wouldn't know if you got recommended an obsolete API, a dated pattern, or a charged foot gun, and if you're looking up docs for every suggestion I doubt that'd save
364.
by dgroshev 3y ago
…and the library might be obsolete, unmaintained, or outright a wrong choice for the task at hand.
365.
by dgroshev 3y ago
It's a really bad TA that has no self awareness of its limitations. You are well aware that it hallucinates, right? You probably have observed it yourself in domains you're familiar with. But how would you recognise a hallucinatio
366.
by dgroshev 3y ago
I used Copilot full time for about four months writing Rust and Python, and then stopped completely. Problem is, LLMs are really good at uninsightful, repetitive, boilerplate-y bits of code that shouldn't be there . The moment I get t
367.
by dgroshev 3y ago
Tsar Bomba was a one off and has no relation to weapons of today. Look up what's on top of Tridents, it's mostly W-76 of various kinds, all below 100kt, which is about twice more powerful than Fat Man when you consider that effect
368.
by dgroshev 3y ago
It's not just that, but also nuclear use would get more probable with less warheads before it gets impossible because there are none. Imagine there are only three nuclear warheads in the world, is the chance of their use higher or lo
369.
by dgroshev 3y ago
You can try pip install pillow for a good example of how it works. I suspect there's a strong survivorship bias here, as you'd only notice the packages that don't ship with wheels.
370.
by dgroshev 3y ago
At this stage [0], uncompiled native extensions are not yet a bug, but a definite oversight of the maintainer. They should come as precompiled wheels [0]: https://pythonwheels.com
371.
by dgroshev 3y ago
I don't know, but I imagine it would make testing substantially more difficult (right now they have fully tested images available), and it would be impossible to implement their distro update thing (you can update the entire .iso backw
372.
by dgroshev 3y ago
IME (x)sense is quite problematic when you start doing anything other than the bare minimum. I spent multiple days trying to figure out why it silently stopped accepting IPv6 delegation, or why does it spike latency for no visible reason un
373.
by dgroshev 3y ago
I was pretty confused what it is too and then I loved it. It's debian plus some shell trickery and CLI tools that let you configure debian and debian packages as a router from one large config tree using neat CLI tools (that support co
374.
by dgroshev 3y ago
The reason you feel that way might be that you are familiar enough with philosophy. After all, LLMs and ChatGPT in particular are are indistinguishable from productised Gell-Mann Amnesia. Edit: rewrote to be more neutral, sorry
375.
by dgroshev 3y ago
It would be great though if they stopped sponsoring the war in Ukraine. Yes it's unfortunate that their main HTS tape supplier is Russian, but pretending that it's not and just ignoring the issue (or obscuring the supply chain thr
376.
by dgroshev 3y ago
> Not true ... warming potency, bio sources (like fossil fuels), deforestation, energy use, animals' inefficiencies in converting plants into calories/protein It doesn't matter. If you run a fixed number of coal fired powe
377.
by dgroshev 3y ago
It's incredibly rude to not just "I'll have the last word and won't engage any further", but also to sprinkle some LLM dross on top. This kind of thing should not have place on HN.
378.
by dgroshev 3y ago
It is literally carbon neutral. Methane and N2O do not increase the amount of carbon in the ecosystem and do not change the climate in the steady state, unlike fossil fuels. Putting them in the same category is disingenuous. I don't un
379.
by dgroshev 3y ago
For comparison, US meat market revenue is about $125bn [0] and dairy is about $41bn [1]. Which means a rise in meat prices if subsidies are rebalanced towards veg, but not an orders of magnitude rise. [0] https://www.statista.com
380.
by dgroshev 3y ago
It is carbon neutral by definition. You don't add carbon to the biosphere by raising cows, slaughtering, and eating them, that's what "carbon neutral" is. The total amount of carbon in the cycle stays the same. It doesn&
381.
by dgroshev 3y ago
I often enjoy seitan and soy, but pretending they're the same is disingenuous. Moreover, it's pointless — meat production is by itself carbon neutral, and there is no fundamental reason meat can't be enjoyed from time to time
382.
by dgroshev 3y ago
I think those links often miss the forest for the trees. Meat production, disregarding energy (that has to become carbon neutral anyway) and in steady state, is carbon neutral. It might require more agricultural land, true, but equating pot
383.
by dgroshev 3y ago
There is effectively unlimited amount of carbon in the form of limestone, at the very least.
384.
by dgroshev 3y ago
The fallacy here is the assumption that we get "hammers" as libraries, and not some splinters of wood in one crate and a whole bunch of hammer heads in another. A "hammer" is useable on its own, it's a self-sufficie
385.
Two Visions of Environmentalism(slowboring.com)
3 points by dgroshev 3y ago|1 comments
386.
by dgroshev 3y ago
Google now does effectively the same with their in-search ads. Most of the time the entire space above the fold is just ads.
387.
by dgroshev 3y ago
It's not as clear cut really. Overall economic productivity didn't shoot up in the last decade despite the dramatic progress we had in software and hardware (e.g. [1]), and it's not clear that AI/ML will dramatically cha
388.
by dgroshev 3y ago
That's the main problem with Twitter Blue (other than its very bad reputation): it's just not worth the money. Instead of adding value to the subscription or making it cheaper, Elon just makes the rest of the product worse. It&#x2
389.
by dgroshev 3y ago
Current rate limit is about 20 minutes of use for me. Remember that it loads a couple dozen replies at a time if you click on a tweet, so 600 tweets is really not much.
390.
by dgroshev 3y ago
It's still LCOE, which is by definition focused on generating assets. It doesn't track societal costs or costs offloaded to other entities, like grids or households, that can be attributed to a particular generating technology bec

More ›