Y
HN Search
Hacker News Search
new
|
comments
|
top
|
jobs
Our_Benefactors
searching PlanetScale…
1.
▲
2.
▲
3.
▲
4.
▲
5.
▲
6.
▲
4 ms
·
31.
▲
by
Our_Benefactors
4mo ago
There were layoffs yes. The solution is to pursue something that on a small scale is net negative financially? Get out of here.
32.
▲
by
Our_Benefactors
4mo ago
> get a pot of land and grow your own food Rejecting industrialized society is actually very expensive
33.
▲
by
Our_Benefactors
4mo ago
This quote is from a viral interview between Jensen Huang and Dwarkesh Patel, the latter made some points during the interview that were very condescending and betrayed a deep misunderstanding of technology.
34.
▲
by
Our_Benefactors
4mo ago
> The attitude of the industry is that it’s up to you to read data sheets carefully and test. Is this backed up by court precedent? This seems like you could easily claim damages due to a differently speccd part. I’m not doubting that’s
35.
▲
by
Our_Benefactors
4mo ago
This is give an inch take a mile type thinking. The arguments against data centers are based in emotion not reason, so it’s effectively a fools errand to try and placate them.
36.
▲
by
Our_Benefactors
4mo ago
We the people. In places where there are significant water shortages, water hungry data centers are not being built.
37.
▲
by
Our_Benefactors
4mo ago
> datacenters use a TON of water and electricity We have a shortage of neither.
38.
▲
by
Our_Benefactors
4mo ago
This is the top comment: “ And ask if AI is so great why hasn’t it designed resource efficient data centers? And ask what the data centers are actually doing. Bitcoin mining, anyone?” Says all you need to know about the competency level of
39.
▲
by
Our_Benefactors
4mo ago
They do, it’s called CXMT and they are making RAM under contract for Corsair
40.
▲
by
Our_Benefactors
5mo ago
I prefer my OS to not require “skill” (there’s no skill left in the age of AI anyways, just wasted time)
41.
▲
by
Our_Benefactors
5mo ago
It’s fairer to call them factions. Also a misuse of the term “FUD”. They also don’t have a parliament, they have the “National People’s Congress” which doesn’t function as a parliamentary body.
42.
▲
by
Our_Benefactors
5mo ago
It’s possible (likely, even) to have a chip fast enough for inference, but not fast enough or with enough memory to do meaningful training runs. Like the current DGX spark.
43.
▲
by
Our_Benefactors
5mo ago
This is the best way I can explain why ai-driven layoffs can be logical. Take a baseline as 5 devs costs 5 salary units and produces 5 output units. AI does something like “for 1.25 salary units you get 1.5 output units” so it’s costlier on
44.
▲
by
Our_Benefactors
5mo ago
> This keyboard is not for you if you do not want to spend an hour or more to set it up and time in the future to adjust a key that stops working well. Do not use any beam spring keyboard in a mission-critical environment. Yeah, I’ll pas
45.
▲
by
Our_Benefactors
5mo ago
Holy horrible UI
46.
▲
by
Our_Benefactors
5mo ago
As someone completely outside the 3D design world who always thought of AutoCAD as the gold standard - really? What program would be used instead? Please enlighten me.
47.
▲
by
Our_Benefactors
5mo ago
Azul and Set are both good. Can’t Stop is another classic “push your luck” game.
48.
▲
by
Our_Benefactors
5mo ago
“What the fuck did you just do?” -> when I need to undo recent changes “What the fuck did you just fucking say about me, you little bitch? I’ll let you know I have over 300 confirmed prompts” -> when the AI argues with me and I need t
49.
▲
by
Our_Benefactors
5mo ago
Claude has 1M context window for the enterprise. 128k feels like a toy in comparison.
50.
▲
by
Our_Benefactors
5mo ago
It runs like shit though in terms of tokens/second and still has a reduced context window. Vs a single claude prompt can easily get into 300k tokens without breaking a sweat. I want local AI to be a thing but the hardware isn’t here ye
51.
▲
by
Our_Benefactors
5mo ago
AI has well and finally killed the idea that outsourcing saves cost. Local AI, sure, but not outsourcing. Been there, done that, doesn’t work.
52.
▲
by
Our_Benefactors
5mo ago
No, I think he’s the spiritual leader for Catholics (not humanity)
53.
▲
by
Our_Benefactors
5mo ago
> But it's a lot more work for still a pitiful amount of money It’s really not a lot of work and if $2000+ for a few hours work is pitiful, I envy your financial position. List on Craigslist at bottom market prices (you’ll still com
54.
▲
by
Our_Benefactors
5mo ago
Of course, you should never sell the car to the dealer and should always make the effort to sell private party, which will often get you 50% or more greater than the dealers best offer.
55.
▲
by
Our_Benefactors
5mo ago
His lane is Catholicism.
56.
▲
by
Our_Benefactors
5mo ago
It’s because it’s not true, there’s no evidence for it that passes the sniff test. No lab is “shipping a worse model once they’ve got you”. People have a bad few days and blame the model providers instead of stepping back to fix their workf
57.
▲
by
Our_Benefactors
5mo ago
Too true. The idea that you could play instants during your opponents turn caused many a disagreement, especially over combat timings or unintuitive but real mechanics like “stack combat damage, sac mogg fanatic”
58.
▲
by
Our_Benefactors
5mo ago
This is similar to how VS system worked, it was a short lived superhero CCG in the mid 2000s, you could play up to one card per turn facedown as a “land”, but you still drew from one deck.
59.
▲
by
Our_Benefactors
5mo ago
And now the cost has increased again, because you need to field enough simultaneous launches to get 10k in the air at the same time. It doesn’t mean you need 10k launchers, but you still need a lot
60.
▲
by
Our_Benefactors
5mo ago
> could probably have emps get it themselves, and reimburse They can’t track token use this way. Also it’s a massive violation of the model providers TOS.
More ›