igor47

11 ms·

121.
by igor47 11mo ago
But dietary cholesterol is not linked to cholesterol in the blood vessels > We confirm from the review of the literature on epidemiological data, meta-analysis, and clinical interventions where dietary cholesterol challenges were utilize
122.
by igor47 11mo ago
Ugh what an amazing example of poor science communication. Literally the only useful numbers for battery advances are: * Energy density in kWh/kg * Cost in $/kWh Neither is mentioned in TFA. This is just PR fluff. Notably, the ra
123.
by igor47 11mo ago
I'm finding this hard to believe. He had a car with 70 miles of range? Was it a 2011 leaf? That car had about 70mi of range but a payload capacity of almost 1000lb. Was your luggage just bricks? Winter is more believable, cold weather
124.
by igor47 11mo ago
Super interesting, I'm just getting into DOM morphing libraries to try to patch around what I've found so far as the main limitation of the htmx approach: losing DOM on swaps. I started a discussion around some of the issues here:
125.
by igor47 11mo ago
Wait but, don't you render the jsx to html? So you you can still perform the diff on the rendered html? Also, how do you know what you rendered last time? When you do the diff between what's in the browser and what the server just
126.
by igor47 11mo ago
Curious if they're planning to support more complicated schemas. They claim to support JSON schema, but I found it only accepts flat schemas and not, for example, unions or discriminated unions. I've had to flatten some of my sche
127.
by igor47 11mo ago
Is "by life here" doing the heavy lifting in your comment? Because obviously immigrants haven't been knocked down by life here . I agree that folks have a pretty tough life generationally in America's rust belt. I disag
128.
by igor47 11mo ago
From TFA: > Because Homo sapiens went through a series of near extinction level events before eventually triumphing – periodically reduced to a few thousand surviving members clinging on in a handful of African enclaves – we have very li
129.
by igor47 11mo ago
I feel like you're up work is a hidden gem. I use it all the time, especially when I have well defined tasks. It's like a better but (much) more expensive LLM
130.
by igor47 11mo ago
> skeptics generally don’t care to push back They do care to push back though! It's just that there's much more of a market for wishful pseudoscientific bullshit than for careful, history and evidence based sceptical bashing
131.
by igor47 11mo ago
Selection/confirmation bias. Think of all the times the child told their mother some other random things which turned out not to be true. Those incidents don't stick out because children say nonsense all the time.
132.
by igor47 11mo ago
This is not true in a practical sense. There is a lot of petrochemicals still out there and our ability to recover marginal reserves keeps improving. I was really into the idea of peak oil when I was younger but it really hasn't pann
133.
by igor47 11mo ago
In particular, COVID led to a massive shift to remote work. On the one hand, commuting to work, especially in the US where most commuting is driving, could potentially save lives. On the other hand though, it could cause people to spend mor
134.
by igor47 11mo ago
Thanks, let me know what you think!
135.
by igor47 11mo ago
Using haiku 4.5 right now. I think I could get away with even smaller models. The prompt is here: https://github.com/igor47/csheet/blob/main/src/ai/prompts.ts and my tool use is here but it&#x2
136.
by igor47 11mo ago
Oh I'm super curious about your notmuch project. I've been wanting an email client that's just a tui on top of notmuch. Wonder if we can pair up? Ill ping you over email
137.
by igor47 11mo ago
I made an LLM-assisted DnD character sheet tracker! It's up here: https://www.csheet.net/ And the repo is here: https://github.com/igor47/csheet If you play DnD, I would love feedback! Feel free t
138.
by igor47 11mo ago
Yeah and my preferred tools (mise) are missing from the environment, and installing it requires arcane environment configuration and then the LLM spends 10 minutes just trying to get the environment set up... On every interaction
139.
by igor47 11mo ago
in my experience, it does tend to follow the plan pretty well.
140.
by igor47 1y ago
... Because currently, your daughter rides around on a Roomba? And this will allow her similar freedom of movement, but more fault-tolerant?
141.
by igor47 1y ago
My approach is to ask Claude to plan anything beyond a trivial change and I review the plan, then let it run unsupervised to execute the plan. But I guess this does still leave me vulnerable to prompt injection if part of the plan is access
142.
by igor47 1y ago
Yeah especially considering the next line is: > Constant exposure to the sun in orbit also means nearly infinite solar power Is it an infinite heat sink or an infinite power source?!
143.
by igor47 1y ago
Hmmm. Interesting prediction. I think even on social media, the consensus is still shaky, and social media is an unalloyed bad IMHO. I think personal cars impacted our lives in a deeply negative way but most people disagree. There is really
144.
by igor47 1y ago
Welcome to HN! I don't think you understood TFA. A) LLM prompting is categorically different from Fortan, not another version of it that's 'less precise'. B) 'upside' again is entirely different from craftsmans
145.
by igor47 1y ago
Yes! This happened in so many industries. Banking is my go to example, where we used to have local bankers making decisions based on local knowledge of their community, but then decision making was centralized into remote central HQs and th
146.
by igor47 1y ago
This tendency must get reenforced through RL in the training phase. It's very high profile when an LLM deletes the wrong thing, eg https://arstechnica.com/information-technology/2025/07/ai-co...
147.
by igor47 1y ago
There's no way the administration in power thinks that's worse.
148.
by igor47 1y ago
It doesn't matter is he's the only one. He's the one we're talking about right now. Comments like yours read to me as "well everyone is doing it" and I really don't like how this serves to normalize bad ac
149.
by igor47 1y ago
What does "the high percentage" mean? I donate plenty of money to both foreign and domestic non-profits and a pretty bad overhead is 20% in my experience. I generally aim for overhead ratios of 10% or so and have no problems findi
150.
by igor47 1y ago
Jobs aren't just business. Humans derive a lot of meaning from being useful and valued.

More ›