Y
HN Search
Hacker News Search
new
|
comments
|
top
|
jobs
joshjob42
searching PlanetScale…
1.
▲
2.
▲
3.
▲
4.
▲
5.
▲
6.
▲
36 ms
·
31.
▲
by
joshjob42
1y ago
No! Don't you people know that's how you release Moloch the Corruptor?!
32.
▲
by
joshjob42
1y ago
If Vimium would work in Orion I'd be happily paying for Orion+ and Kagi but sadly it doesn't.
33.
▲
by
joshjob42
1y ago
I think the idea was actually that it would give people who like the taste/texture/experience of meat an option to skip meat, and lower the bar for people to eat a more plant based diet, not cater to vegetarians or vegans. I'
34.
▲
by
joshjob42
1y ago
There is no cost pressure on Waymo to reduce rates below Uber or Lyft as they are much smaller, they are price takers not price makers in the market. Lowering prices far below your competitors when you are supply constrained for the time be
35.
▲
by
joshjob42
1y ago
Well, Altman is also investing in Helion, which projects to get the price of electricity to ~$10/MWh, but for whom, much like solar, wind, and actual nuclear the cost structure is overwhelmingly dominated by capital costs and non-varyi
36.
▲
by
joshjob42
1y ago
I love Orion and it'd be my main browser if only a) Vimium worked and b) they'd add or enable web panel extensions ala old Zen, Vivaldi, or Edge. I'd pay $30/mo to use Kagi such an Orion.
37.
▲
by
joshjob42
1y ago
Well even for simple things there's a large difference. Say you toss a coin N times and observe heads x times. What is the probability of your next toss coming up heads? A frequentist arguably would say the question doesn't really
38.
▲
by
joshjob42
1y ago
I think I'll just use Julia, since it has great libraries for doing matrix operations and mapping loops etc onto GPUs and handling indices etc. And if you need some Python library it's easily available using the fantastic PyCall l
39.
▲
by
joshjob42
1y ago
It would be really amazing if you could attempt for those who've connected Pocket to pull the data of all those saved files etc into Reader. I realize it wouldn't necessarily work as some stuff is just gone, but it would be a nice
40.
▲
by
joshjob42
1y ago
I have only just begun my road down lisp, from starting to hack some elisp and working with some CL for fun. But structural editing is turning out to rapidly be my favorite feature, to the point that I'm working in my personal time on
41.
▲
by
joshjob42
1y ago
Did you use evil mode? I am basically just running Spacemacs and I don't often need to use ctrl and option most of the time by leveraging the modal editing.
42.
▲
by
joshjob42
1y ago
Interop with what? It has a very good FFI for C, PyCall, RCall, etc. work quite well. It is pretty REPL focused, but what is the thing you want that you don't have? Like better debugger support in VSCode or something?
43.
▲
by
joshjob42
1y ago
We have a constraint, say keeping warming below 1.5 or 2C, and we want to achieve it with as little pain as possible. You use a laptop, you probably play some video games, you see movies, you probably eat meat, you probably drive a car arou
44.
▲
by
joshjob42
1y ago
Is Coalton compatible broadly compatible with the features of CIEL? I've been interested in getting into CL, and CIEL seems like a very nice batteries-included way to do that. But Coalton is also quite interesting and brings some featu
45.
▲
by
joshjob42
1y ago
I mean, if you draw the scaling curves out and believe them, then sometime in the next 3-10 years, plausibly shorter, AIs will be able to achieve best-case human performance in everything able to be done with a computer and do it at 10-1000
46.
▲
by
joshjob42
1y ago
As someone who came from vim, I find evil to be great tbh.
47.
▲
by
joshjob42
1y ago
What is a pet in the relevant sense here? Let's assume you have everything in the Culture except the ASIs. So you have full blown Drexlerian nanotechnology, access to all the technology more or less possible out to the limits of physic
48.
▲
by
joshjob42
1y ago
For some reason this link isn't loading, but it's on https://archive.ph/Ks0xt
49.
▲
by
joshjob42
1y ago
I adored my Spark EV til it sadly died (fairly scarily, on a highway access road) one day. Chevy was never able to repair it and ultimately gave me a nice payout after paying for a rental for me for nearly a year. But if you sold the Spark
50.
▲
by
joshjob42
2y ago
Costs for a given amount of intelligence as measured by various benchmarks etc has been falling by 4-8x per year for a couple years, largely from smarter models from better training at a given size. I think there's still a decent amoun
51.
▲
by
joshjob42
2y ago
I mean, it's just doing STT, it's not trying to run inference on a frontier model, and if you have to connect to a remote datacenter vs your own machine that could add latency that slows down the experience. There are also tradeof
52.
▲
by
joshjob42
2y ago
Yeah if you're doing a big coding session up near the limits of context even with prompt caching each reply can be ~80¢ or something, so you can definitely blow through $200 on the API if you are using it heavily.
53.
▲
by
joshjob42
2y ago
https://ai-2027.com/research/compute-forecast In section 4 they discuss their projections specifically for model size, the state of inference chips in 2027, etc. It's largely pretty in line with expectations in te
54.
▲
by
joshjob42
2y ago
I think generally the expectation is that there are around 100T synapses in the brain, and of course it's probably not a 1:1 correspondence with neural networks, but it doesn't seem infeasible at all to me that a dense-equivalent
55.
▲
by
joshjob42
2y ago
"Oh no, I'm devastated. You bought my objectively 300k house for $1M bc I evaluated it that way. Whatever shall I do?" goes and rents a nice apartment in a large complex that's valued at $500M
56.
▲
by
joshjob42
2y ago
Actually they Miami would be #6. They are also starting to operate in Atlanta early next year along with Austin. So 6 cities across 5 states.
57.
▲
by
joshjob42
2y ago
Ah, Dirichlet Processes, such lovely things. Reading this paper, I was struck by how obvious most of the solutions were given my own background from grad school benchmarking quantum annealers and other classical solvers for spin lattices (m
58.
▲
by
joshjob42
2y ago
It's a processor and a screen. There's nothing about the physical device that prevents it from doing well at reading ebooks or comics. It's purely software. Forcing me to carry both a tablet for reading and a tablet for writi
59.
▲
by
joshjob42
2y ago
At the limits of what seems likely to be feasible with atomically precise manufacturing and chemical energy scale technology, you arrive at something similar to a somewhat slower Star Trek replicator, able to produce on demand in under an h
60.
▲
by
joshjob42
2y ago
An eink device not being great at reading books is silly. These devices are especially compelling for academics, but Remarkable has no links to reference managers like Zotero. It also is radically dependent on your computer or phone for get
More ›