Y
HN Search
Hacker News Search
new
|
comments
|
top
|
jobs
blagie
searching PlanetScale…
1.
▲
2.
▲
3.
▲
4.
▲
5.
▲
6.
▲
37 ms
·
421.
▲
by
blagie
3y ago
As much as I agree with the high-level message, I disagree with the specific list. The list makes sense from the bottom, where the goal is to get stuff done. The list makes no sense from the top, where the goal is to make sure the right stu
422.
▲
by
blagie
3y ago
I have a successor, almost identical, but perhaps two generations newer. Upgrades: - Bigger, better dust bin - Slightly less dust makes it into the HEPA filter - Slightly better / cheaper roller design ... and similar details. A friend
423.
▲
by
blagie
3y ago
The "run over things" is almost a feature for me. I run the Roomba a few times a week, and I need to tidy up before. Mess never builds up. It means I keep my floors clean! The "gets stuck under furniture" is a bug. Howev
424.
▲
by
blagie
3y ago
I agree. However, I also don't think it's success can be attributed to Brooks either. He's not very good for much of anything, except making himself be known as a famous roboticist. The company was successful due to other fol
425.
▲
by
blagie
3y ago
FYI: Cheap Roombas seem better than expensive ones. I have a random one which bangs around the house for a long time, and the house ends up clean. Push a button, come back an hour later, and you're done. I like the simplicity.
426.
▲
by
blagie
3y ago
I have never successfully received compensation for lost or damaged bags. It's not that it's impossible. It's that the hoops required were always more than the compensation. It's theft.
427.
▲
by
blagie
3y ago
This is a git design issue. History should be at the keystroke-by-keystroke level, which aggregates into small, focused commits, which aggregates into large merges. In other words, good written records have layers and organizations. A histo
428.
▲
by
blagie
3y ago
Kind of. Here's the basic problem: 1) The BIG tools have SOME port to AMD. I am confident that one can run Mistral or SD in some framework there. 2) It's a port. If you want to do anything bleeding edge, development, niche, or sim
429.
▲
by
blagie
3y ago
16GB for $330. The price is about the same as an Arc A770 16GB, and about 66% of a 4060 Ti 16GB. That's potentially compelling. I'd love to see benchmarks and compatibility for PyTorch, llama, Stable Diffusion, etc. I assume it&#x
430.
▲
by
blagie
3y ago
My general impression is that a lot of companies are more than 4x overstaffed. Early Google had small, elite teams, whom it treated very well. That works really well. Late Google kept elite team salaries with bloated, mediocre devs. That wo
431.
▲
by
blagie
3y ago
Those are the most complex customers. They directly cost more than they bring in. However, they're also: 1) Thought leaders / influencers. How you treat them translates directly into blog / forum / etc. posts. 2) Innovat
432.
▲
by
blagie
3y ago
Outside counsel has one desire: Maximize billable hours. There is literally no incentive to have customers do well. I've seen a ton of frivolous litigation due to having legal strategy defined by outside counsel. Internal counsel, char
433.
▲
by
blagie
3y ago
I seriously doubt that. "net of everything" would include HVAC. * In the winter, my computer means I run my space heater less by every watt it puts out. It's free. * In the summer, the costs are several times higher. Every wa
434.
▲
by
blagie
3y ago
AoPS are among my favorite math books, but they're definitely not proof-based or particularly rigorous in terms of formalism. They do focus on complex problem solving, which is equally important. The key value-add of AoPS are interesti
435.
▲
by
blagie
3y ago
I don't think you get it. If you want to automate anything in Google drive, that's the modularity of permissions in Google APIs: https://developers.google.com/drive/api/guides/api-specific-... I aut
436.
▲
by
blagie
3y ago
It's not always bad security. a) I have apps with their own Google accounts b) I have devices with their own Google accounts It is (or was) a good way to go. Receiving an email from a scanner or sharing something with an app on Google
437.
▲
by
blagie
3y ago
I grant you a trophy for your extreme talent for reading too literally, and then stating the obvious. Here: U+1F3C6
438.
▲
by
blagie
3y ago
I don't have strong feelings about general internet access, but right now, I am probably spending $200k / year in tax dollars to keep him locked up behind bars. I'd be getting my money's worth. I would much rather he be
439.
▲
by
blagie
3y ago
> I think some ICs like that it's harder for people to say to them that they are driving in the wrong direction, even if it's still true. I think this is extremely cynical. Feedback makes for good organizations, and when I buil
440.
▲
by
blagie
3y ago
This is my experience. My productivity on coding is a lot hire remote. On many leadership / management-style tasks, it's a lot lower. Come to think of it, I'm wondering if that's where the split comes in. Upper managemen
441.
▲
by
blagie
3y ago
That's not the way legal cases went. Indeed, they went all over the place. That's the reason you see "IANAL" disclaimers all over the internet. Legal advice from non-lawyers can be problematic in many ways. Some jurisdic
442.
▲
by
blagie
3y ago
This basically means "Get sued." There is no clear legal, definition of "noncommercial," and courts have gone all sorts of different way on what constitutes commercial use. This is where CC NC licenses imploded. A lot of
443.
▲
by
blagie
3y ago
> Proof: if it was false, you could do what GPT4 does with 10 param model and a good prompt. This is oh so very much a strawman. There is rapid progress in AI. For my domains, the first useful model (without finetuning or additional trai
444.
▲
by
blagie
3y ago
Mostly ones related to text transformation (e.g. changing text style) and feedback (e.g. giving suggestions for how to improve text). A year ago, the ones I tried were useless and dumb. Right now, they work quite well.
445.
▲
by
blagie
3y ago
Thank you so much for that comment. I really appreciate the feedback. I do sometimes come off as arrogant. That's unfortunate, and in part due to my cultural background. It's helpful feedback. It's difficult to be downvoted o
446.
▲
by
blagie
3y ago
FYI: My Thinkpad has better build quality than any Macbook. My Dell was comparable. The key thing about Dell and Lenovo is they have a byzantine set of laptop lines, from plastic craparific up to rather good. A top-of-the-line Thinkpad will
447.
▲
by
blagie
3y ago
I think we are talking across each other. (1) You are correct that I have not read the docs or discovered everything there is. I have had HA for a few weeks now. I am figuring stuff out. I am finding the learning curve to be steep. (2) Howe
448.
▲
by
blagie
3y ago
> no matter how careful your prompt engineering is, you simply can’t use tiny models to do big complicated tasks. I can and do! The progress in ≈7B models has been nothing short of astonishing. > My experience with this stuff has been
449.
▲
by
blagie
3y ago
I can write a control system easy enough to do this. I'm kind of an expert at that, for oddball reasons, and that's a trivial amount of work for me. The "smart enough" part, I'm more than smart enough for. What'
450.
▲
by
blagie
3y ago
Honor to meet you! [Anonymous] founder of a similarly high-profile initiative here. > Creating a prompt to write JSON is possible but need quite an elaborate prompt and even then the LLM can make errors. We want to make sure that all JSO
More ›