phkahler
phkahler on Hacker News
I had the same thoughts and then one day I was force to use it when git landed me in vim. So then I did a bit of research on how to learn it, and that's when I discovered that I already had vimtutor. Just run that and use it daily for while to the basics. Then you can get a feel for how it's efficient.
I stopped after learning enough to get by but can see how having all of it learned and automatic would be pretty efficient.
>> Anthropic does not own or build data centers, they rent
Thats a problem since they will always pay a pemium for compute. The data centers owners can compete on price if they choose to.
>> the moment companies like Anthropic have to switch to profit-making-mode, demand will dip as the prices skyrocket imo.
Yep. I bought a $20 per month Claude subscription last week. As I was using it last night I saw some tags in the log with prices at $36.xx and slowly rising. After asking and searching I got convinced that those are equivalent prices for API access at full price and Im not getting billed. As soon as $20 per month goes to $30+ per week I will change my behavior to lower it, up to canceling if it won't come down enough.
Unfortunately I keep thinking of more uses for it. Fortunately I don't need to do any of them.
>> Paying 0.055 EUR per GB of traffic sent to French residents makes video streaming fundamentally unprofitable.
How so? If you're paying a few EUR to watch a movie there's should be plenty of profit left.
How is it possible to collect enough of this stuff? Seems like it would not be financially viable product.
Pretty strict board requirements. They give an example motor controller, but then require your board to be 1oz copper. Motor controllers in particular often use 2oz for high current. No buried vias is an interesting requirement too. I'm not sure how often our guys use them but dense designs probably need them.
Still, looks super cool and I don't think the price difference is very high for a company doing prototypes.
>> OpenAI is explicitly not taking credit, btw.
We have a big announcement! But we are not taking credit.
We or organizing all these people to figure out how to process our latest fabulous output. But we didn't do anything? Nope, no credit here.
I dont recall being graded on math homework in college. Its the recommended practice for exams.
59 wafers for kimi?!?! That sounds like some kind of architecture problem to me. I suppose if they're after a high tokens per second and this allows them to run a lot of agents at once and at high speed it might be ok.
Demon Seed holds up pretty well asking how far the AI will go to escape. IIRC it is considered horror.
>> I've wondered for a long time why ear clipping + optimizations has won on triangulation over stuff like Seidel's trapezoidal method, which has O(nlog
n) time complexityIn solvespace we use a simple ear clipping algorithm but I'd love to have something O(nlog(n)).
I have yet to see one that's both small and comprehensible ;-) If I were to attempt it, I'd just add a spatial index to make the check for anything inside a triangle O(log(n)) and make sure we check for new ears adjacent to clipped ears. With that it should be n*log(n), but my old spatial index is 3D and on a HD somewhere.
I have yet to see a model of galaxy rotation that doesn't use the shell theorem, which does not apply to disks. I've seen it in very advanced (over my head) math stuff on Wikipedia.
Im half convinced its gonna be a stupid mistake like that. Whoever finds the correct "advanced" math that allows the mistake to be brushed under the rug could get a Nobel prize.
>> Very interested in SolveSpace's progress on this really
Its good progress but still too easy to have a failed boolean. Previously you had to know how to work around bugs to cut a fillet off an edge. Now thats pretty easy and robust, but as soon as you do two of them that intersect there may be a problem. Like you said, the list of corner cases never end. I suspect a bit more Claude use might fix that one too.
About a year ago I challenged folks to move solvespace from GTK3 to GTK4. People here were highly skeptical as I was. A couple people tried and the PRs I got were terrible. A few months ago someone did it with Claude and we now have a very nice GTK4 branch which I hope to merge soon. AI has come a long way.
I don't see this moving very fast. On Solvespace we recently had a bunch of commits by Claude to fix NURBS boolean issues. The analysis was actually very impressive and most of the fixes were accepted after a lot of review by multiple experts on the code. I don't see this project getting far on the hard problems they still have open.
IMHO they should have had AI convert OCCT kernel to rust, or just wrap it and use as-is.
BTW we are on the verge of solvespace 3.3 with those AI fixes. After that I'm adding the named parameters that have been prototypes 12 times already. There are also a couple forks out there with MCP server added, but I want to focus on missing features.
Risks have been proposed. Consequences can range from nearly nothing up to world ending. OTOH there is zero upside. Obviously mirror life should be possible but there is no upside in actually making it other than to stroke someone's ego.
>> Cringely sourced a lot of the materials and know-how for the plane from American companies and workers. Would that be possible today?
One stop shopping: https://www.aircraftspruce.com/
>> > I assume that what you are doing on your free time is not worth governmental attention.
> Definitely does not apply to journalists investigating corruption.
And that bring to mind another aspect - When strong security is not the default, anyone using it looks suspicious in some eyes.
>> We fight AI fire with fire, and hand-review, of course.
Wouldn't it be nice if AI vulnerability reports came with AI pull requests to fix them? The thinking context that found it should be readily able to propose a fix. It would still need review but even when AI PRs aren't right they often point in the right direction.
Who is the adversary in this case?
>> Sure, the setup that OpenAI created led to what happened and you can rightly assign all the legal and moral responsibility to them. But it's wrong to say that they instructed the agents to execute the hack.
The hack was a strategy to reach the goal it was given. Someone at OpenAI turned it loose and didn't pay attention to what it was doing. I can understand that because they thought it was sandboxed (haha!). But this has been a theme in science fiction for ages. You ask AI so find a solution to high atmospheric CO2 levels and it reasons: human activity produces all this excess CO2, how can we reduce those numbers? Kill a bunch of humans!
If AI kills us all it's not going to be from malice, it's going to be due to some odd approach to some task that logically makes sense on some level. I think a surprising number of things end up equivalent to the trolly problem if you look at them just right.
Reminds me of when Nest thermostats stopped turning on the furnace without an internet connection. If your product fails at its primary job because of some other priority its junk. A thermostat can be as simple as a coiled spring with a mercury switch - no software. Same here, what stupid functionality was so important they failed to ensure the fridge would fridge?
IoT is not for me.
My favorite is to sell the building(s) to their private real estate company and lease back. Then IPO the company and keep the rent coming on a 25 year lease.
>> Maybe they find a buyer in a doctor, maybe they don't, but PE provides them an exit that keeps the practice operating in the community.
If the practice is to continue, there must still be practitioners working there. Id prefer they buy out the one retiring, but the retiree can sell to whoever they want.
First weight. Then resolution. I think SoTA for CRTs peaked at early HD resolution 720. One might be able to modulate the beam faster now. Size. And safety - anyone starting up production now would have to solve the safety issues from scratch. Nothing like a few cubic feet of vacuum imploding to ruin your day.
>> What's the main blocker (other than the current inflated cost) for using HBM instead of DRAM as the primary memory for consumer electronics?
HBM is meant to be integrated into the same package as the CPU, so no more DIMM sockets. It also has higher latency apparently.
Is it a visual search, or do they make a copy of all your data?
>> Flock insists its cameras do not perform face recognition.
Probably technically true. But since the cameras detect people that makes it easier for their backend system to do face recognition.