joelthelion

4 ms·

31.
by joelthelion 4mo ago
How would you set the incentives, though? Almost by definition, it's hard to reward things that aren't visible. Note that there is also the flip side of the coin, people who spend all their time worrying about things that never ha
32.
by joelthelion 4mo ago
Thank you !
33.
by joelthelion 4mo ago
Do you pay for the model? Or run it locally? If so, which hardware do you use? Is it fast?
34.
Ask HN: What do you currently use for AI coding (personal or professional)?
2 points by joelthelion 4mo ago|5 comments
35.
by joelthelion 5mo ago
Isn't the database already one of the hardest piece of infras to scale? Why would you want to load it with additional long-running jobs?
36.
by joelthelion 5mo ago
Having control over the execution plan is super interesting ! This is a very common frustration when writing SQL. Do you think it would be possible to offer Prela as a direct interface to a relational database?
37.
by joelthelion 5mo ago
You're right, but what I'm saying is that solving the problem isn't necessarily the primary goal and these new abstractions can be valuable in their own right.
38.
by joelthelion 5mo ago
It's also a way to model the world and produce new useful abstractions. It's not just about solving problems.
39.
by joelthelion 5mo ago
What a wonderful world we live in. And to think that all of this is largely self-inflicted...
40.
by joelthelion 5mo ago
Or even sample from a distribution of variations for infinite possibilities ?
41.
by joelthelion 5mo ago
I've sold all my stocks. My reasoning is that if AI stocks go bust, they will take the global stock market with them.
42.
by joelthelion 5mo ago
Not really, how much of a public company are you when 5% of your capital is public ?
43.
by joelthelion 5mo ago
ICE Cars have been around for 100+ years. We have a good understanding of their achievable efficiency, including theoretical guarantees. LLMs on the other hand have only been around for a few years. Large technological breakthroughs are muc
44.
by joelthelion 5mo ago
AI isn't a necessity in the same way that EV / solar is, we can afford to take it slow and see what sticks. Also, EV/Solar are a much more mature technology than AI, so huge technological leaps are less likely.
45.
by joelthelion 5mo ago
I see it the other way around. In a few years, algorithmic and hardware improvements will likely make these huge datacenters mostly obsolete. Americans will get to enjoy their rusty infrastructure and polluted air, and Europe will have lean
46.
by joelthelion 5mo ago
I hope play money prediction markets are untouched. They are useful and fun, and as far as I know, dont have the downsides of real prediction markets.
47.
by joelthelion 5mo ago
Opencode can be surprisingly hard on the CPU (could be an issue when coding on battery or a weak remote VM), and uses a lot of RAM. A little competition is always welcome.
48.
Toto 2.0: Time series forecasting enters the scaling era(datadoghq.com)
4 points by joelthelion 5mo ago|0 comments
49.
by joelthelion 5mo ago
This rings very true as an industry. We are all extremely busy, producing billions of lines of code, and yet the software around us isn't really getting any better. What we'd need instead is concerted simplification. But that ain&
50.
by joelthelion 5mo ago
I agree it's rather basic but as stated in the article, its code is still longer than war and peace. There is still plenty of opportunities for security vulnerabilities in something of that size.
51.
by joelthelion 6mo ago
How does Warp/Openwarp relate to Opencode or Claude code?
52.
by joelthelion 6mo ago
The problem is "just". "Just" getting pytorch to work and to work well is a huge undertaking.
53.
by joelthelion 6mo ago
In my opinion there is also a deficiency of the models who should be able to say "I don't know" when asked for something unreasonable.
54.
by joelthelion 6mo ago
Can't wait for people to migrate to open tools (opencode/openrouter). This will unlock a lot of innovation. (I know openrouter is not open, but it allows competition and should be easily replaceable if needed)
55.
by joelthelion 6mo ago
And risk of loss of control on the software ecosystem.
56.
by joelthelion 6mo ago
> Is the wearable accurate enough to be sure that 3bpm is not a measurement fluke If the statistical tests show significance (and are valid), the answer to this question is yes. If you have enough data you can make strong conclusions ev
57.
by joelthelion 6mo ago
There's a difference between trusting Anthropic and trusting random mac owners.
58.
by joelthelion 6mo ago
Note that you don't need YouTube's permission to remove shorts from the homepage; there's a Firefox extension for that.
59.
by joelthelion 6mo ago
The good news is that, apart from the models themselves, we don't need much from these companies: - Use Opencode and other similar open-source solutions in place of their proprietary harnesses. This isn't very practical right now
60.
by joelthelion 6mo ago
> You are responsible for security. I saw good devs skipping basic SSH hardening and get infected by bots in <1hr. My go-to move when I spin up servers is a two-stage Terraform setup: first, I set up SSH with only my IP allowed, set u

More ›