Y
HN Search
Hacker News Search
new
|
comments
|
top
|
jobs
jvanderbot
searching PlanetScale…
1.
▲
2.
▲
3.
▲
4.
▲
5.
▲
6.
▲
8 ms
·
91.
▲
by
jvanderbot
2mo ago
I paid for berkeley mono and don't mind. But I'm not saying anything negative about free alts
92.
▲
by
jvanderbot
2mo ago
Well there's a difference between "Generate all code with AI" and "Trust AI exclusively" And yes, in talking to friends in other areas of s/w, it's really eating the workflow. It does degrade muscle memory
93.
▲
by
jvanderbot
2mo ago
Berkeley Mono is an excellent font. Highly recommended if you still code. https://usgraphics.com/products/berkeley-mono And by the way, this website is fantastic.
94.
▲
by
jvanderbot
2mo ago
Even simpler: Can GPT-2 anticipate and build Gwen/Deepseek? I think the answer is almost trivially "no", so I wonder what changed?
95.
▲
by
jvanderbot
2mo ago
Came for: "A computer once beat me at chess, but it was no match for me at kick boxing." TFA was actually about leaps of intuition, sadly. One of the experiments I've heard proposed around here is to somehow create an LLM fro
96.
▲
Tell HN: Wife used Chat GPT to set up her new biz domain, email, and website
4 points
by
jvanderbot
2mo ago
|
2 comments
97.
▲
by
jvanderbot
2mo ago
Their profile specifically says they will not acknowledge follow requests, but they only share posts with followers. As such, not sure what the value of the top level link is.
98.
▲
by
jvanderbot
2mo ago
Love it. I confess I learned how to build constrained ILP/MILP problems by solving equipment loadout optimality for the game Highfleet.
99.
▲
by
jvanderbot
2mo ago
Reminds me of Zachtronics: Take a computationally difficult problem, reduce it to a game, and then score people against each other. Pure genius and nerd snipe marksmanship. Did you know multi-robot exploration pretty much as laid out in Exa
100.
▲
by
jvanderbot
2mo ago
No, but I bet this has a high likelihood of coming out in preliminary discussions. What most likely happens is a normie has no idea of how to build something that does nothing first - they just start describing the end state. I'm going
101.
▲
by
jvanderbot
2mo ago
Open any chat window and ask for a simple SPA with startup instructions. It's fine, it's fine, it works.
102.
▲
by
jvanderbot
2mo ago
> "The internet" Classic software engineer perspective. I have good friends who build their entire career between "ISP" and "Load Balancer"
103.
▲
by
jvanderbot
2mo ago
If only we could use that extra power to make methane for later power plant useage https://terraformindustries.com/
104.
▲
by
jvanderbot
2mo ago
Or, that the mathematical formalisms that model the limits of software performance are firm enough that barring P==NP, nothing much will change despite proofs of beautiful math.
105.
▲
by
jvanderbot
2mo ago
Right, and that's why it's preposterous to even imagine this happening. In no time you'd have an administration that was providing that DNA evidence to defense contractors for Top Secret clearance, background checks, etc. &qu
106.
▲
by
jvanderbot
2mo ago
Sure, but even worse, is the state-of-the-art: they use a cli agent and the "copy-paste, try to run, paste error, try to run" loop is called "Agentic engineering." Now we have to believe they're 10/10? I see no
107.
▲
by
jvanderbot
2mo ago
The point is that the deciding is the "higher layer." It's not that LLMs cannot, but LLMs as designed do not. They are not designed for free-ranging days off, they are designed to help us do things, and that makes them so
108.
▲
by
jvanderbot
2mo ago
Go ask an LLM to "do something with 1M tokens".
109.
▲
by
jvanderbot
2mo ago
I'm convinced this is why there's a bifurcation in old-hand/new-hand results: the old hands include the steps they'd do and the tools they'd use, and the new-hands never learned those tools. That'll get RL-d in
110.
▲
by
jvanderbot
3mo ago
The author of clean code (who makes his living consulting on how to write software) has pivoted to AI in a spectacular tweet about how to set up automated software development just in time to make his living consulting about how to set up a
111.
▲
by
jvanderbot
3mo ago
It depends on if we expect posters to be informative and include nuance.
112.
▲
by
jvanderbot
3mo ago
This comment would be much better without the second line
113.
▲
by
jvanderbot
3mo ago
I think you're missing a step --> Drier forests and more downed trees. All the lightning in the world isn't going to burn down a rainforest.
114.
▲
by
jvanderbot
3mo ago
A phone with a backup account, unlocked with duress pin makes sense to me. How hard would it be to admin-allow some apps to work across barriers. E.g., wallet apps work, airline tickets work, but email/socials do not. the issue is tha
115.
▲
by
jvanderbot
3mo ago
Do this long enough and you really believe it's good, after all the years of being promoted for strong-arming people. But hey, I do it. I prebake my own tools to work within the system, a new person joins, and I show them the tools as
116.
▲
by
jvanderbot
3mo ago
read the diff and time the query?
117.
▲
by
jvanderbot
3mo ago
https://x.com/jodavaho/status/2081416737883353417?s=20 (shame x is the easist way I have to post two images quickly) dots seem to fit bls pretty well. Just to bound the effect of the spike, that is a +10% jump, so
118.
▲
by
jvanderbot
3mo ago
Yeah those single points of failure/bottlenecks are prefect focal points for people with floating responsibilities and a good understanding of how to use claude. The system-sweeping features I still would ask to review, but you can
119.
▲
by
jvanderbot
3mo ago
If you drop the peakiest peak, the data is quite symmetric. Zoom in on 2015-2026 on bls chart. It's pretty symmetric.
120.
▲
by
jvanderbot
3mo ago
They cut out the peaky-peak (16%), and then fit a time-lagged smoother, it looks like. The dots are more like bls.
More ›