Maybe he means that LLM will hit a ceiling glass or that the "right" approach will give equivalent with less training/less intensive compute requirements ?
And the store does not use facial recognition and/or checking id to know if the potential buyer is a kid ?
The only (huge) difference for me is the scale of the verification and how data are stored.
> And the store does not use facial recognition and/or checking id to know if the potential buyer is a kid?
They can just not serve cigarettes. In addition I think it's also insane to compare cigarettes, which are purely negative, to free internet usage which is massively net positive.
I think I know where they're coming from as I used to have a similar wrong model. I thought strength = more muscle cells and endurance = just better heart/lungs to deliver oxygen and clear waste like CO2 and lactic acid.
Turns out muscle fibers mostly grow bigger rather than more numerous, and there are different fiber types (slow-twitch vs fast-twitch) that adapt based on how you train. So for the same muscle, an Ironman runner and a guy doing heavy low-rep squats will develop different fiber characteristics: you can't fully max out both.
I'm simplifying, but learning this changed a lot about how I understand exercise at the biological level.
My understanding is that anabolic steroid are somehow close to what you're thinking about? It's just that as anything taking a simple shortcut , it comes with unwanted effects
I also agree on "for personal things we don't need SaaS" and I would say do we even needs self hosted in the sense of a central server.
By that I mean, could we have like for firefox , heavy clients but with client to client sync. The goal is to not need to have a always online machine while still solving the "i prefer if my emails are copied both on my laptop and my phones" . Especially as nearly all my devices are often if not always on the same LAN
Firefox sync clearly requires a central server. For any kind of peer to peer syncing to work you must have the machines on at the same time and accessible. And then there is the issue of NATs, including CGNATs. To work reliably these almost always have to have some kind of relays anyway (Tailscale's DERP, Syncthing also has relays).
For the experience an average consumer expects, you at a minimum need a central short-lived cache.
Yes sorry I meant firefox not for the way its sync , but in the way its a heavy client you install.
As said for me most of my devices will be at some times during the week in the same nat so that no centralized server even short lived should be needed. And for personnal use, I only care if the device I have on me is the one with latest data especially as for most use case I'm the only one reading/ writing , so eventually consistency is not an issue
Yes what would be better is a "libsyntching" that i can plug to a software so that it does not require additional brain power, i.e install Note app on device A and B, pair them once, fire and forget.
Thanks a lot, I really first thought "404" was just a geek reference and not the actual code name !
I have some very good friends which are Chinese but are not able to read English, do you mind if I do a AI translation, and if you can check it to see if it translate what you're trying to convey ? (I propose that as I think it would be too much to ask to ask to redo the text in Chinese)
Edit: haha I see you actually did the reverse ! Do you mind sharing also the original CHinese script ? That would also help me with my own mandarin learning !
I did write and publish this story in Chinese first. You don't need an AI translation for them; the original text exists and has been quite popular in the Chinese corner of the internet.You can search for it using the title:《我在404长大》
I definitely appreciate the style of the HN English article, but I think the browser-translated version possibly gives a bit more context to some of the story.
e.g. This is the English version
"We would clutch candy wrappers in our hands, giggling endlessly. The teacher would scold us for disturbing the nap, but we Hid behind our parents, still laughing."
This is the browser-translated version:
"I kept giggling when I saw her, and she giggled too, and we kept laughing with small sugar paper during our lunch break. When my parents came to pick us up, the teacher criticized us for being undisciplined, and we still hid behind our parents and giggled."
I do not wish harm to befall you, but is it that because of CCP censors that you removed it from the Chinese version? Did they ask you to remove it or did you do that proactively?
When I first write this in Chinese, they didn't censor anything, just let me published without question.The first guy wrote this been called by phone, they said it's secret, but nothing happened later. Before I published here I add some part, for example, the context of the famine, in Chinese version, people know what I talked about.
I find it hard to imagine otherwise. HTTP codes are based on the server return code system used in FTP, first published in 1971, where each of the three digits had a specific role and the values simply counted up from 0-9 as different meanings were assigned. HTTP is a little looser about the syntax, but it's the same general idea. Given the scheme, something was going to be code 404.
reply