Hacker News: cavisne

New comment by cavisne in "Google Discloses $94.1B in SpaceX Stock, Marking 6% Stake"

cavisne — Sun, 26 Jul 2026 21:26:10 +0000

How do you think googles stocks are set up?

New comment by cavisne in "Australian energy retailers must offer three hours of free daytime electricity"

cavisne — Wed, 15 Jul 2026 04:05:22 +0000

Australia has a fake residential electricity market, until this year they had to directly offset electricity bills.

This year is the only year that prices have gone down.

New comment by cavisne in "CarPlay Is Additive"

cavisne — Fri, 03 Jul 2026 04:12:26 +0000

For me it’s about the waste. Engineers have taken the time to build a screen into the car with a good viewing angle, that works well with glare, doesn’t obstruct the drivers view.

Then the software side just makes the whole thing useless with a terrible UX that will never be updated.

CarPlay is a great solution because the non safety critical stuff (music, navigation) gets offloaded to a competent software company.

New comment by cavisne in "The curious case of low-protein diets"

cavisne — Sun, 07 Jun 2026 20:25:42 +0000

When you include the thermic effect of food protein is closer to 3 calories per gram. Ie it burns about a calorie to digest a gram of protein

New comment by cavisne in "Higher usage limits for Claude and a compute deal with SpaceX"

cavisne — Thu, 07 May 2026 02:03:34 +0000

Probably the best source https://jax-ml.github.io/scaling-book/

New comment by cavisne in "Higher usage limits for Claude and a compute deal with SpaceX"

cavisne — Thu, 07 May 2026 02:01:34 +0000

Definitely not going to be leasing it at a loss. GPU's are sold out, Anthropic will be paying a significant premium.

New comment by cavisne in "Agent Skills"

cavisne — Wed, 06 May 2026 05:23:43 +0000

The problem with all these skill frameworks is when you are deep enough into AI coding to write one - your brain just cannot handle actually putting effort into it. These are all written by AI, and the authors skim read them.

It’s better to do the maybe 2 hours of human effort the authors put into them yourself for your own workflow.

New comment by cavisne in "Google plans to invest up to $40B in Anthropic"

cavisne — Sat, 25 Apr 2026 05:13:42 +0000

Does not seem that complicated. OpenAI basically had to do a lock-in deal with Microsoft/Azure at the time, and they pioneered this circular funding hyperscaler deal structure so there were some rough edges.

Anthropic (all ex Open AI) knew the negatives of the deal, so they made a slightly better deal with AWS, not a full lock in. They also grounded it in hardware from the start, ie. being the flagship customer for Trainium, the flagship customer for external usage of TPU's.

New comment by cavisne in "OpenAI is walking away from expanding its Stargate data center with Oracle"

cavisne — Tue, 10 Mar 2026 03:00:32 +0000

This feels a bit overdone. OpenAI has had problems with every compute partner they've ever had. It's just not a solvable problem, who would they go to to allegedly get next-gen chips quicker?

New comment by cavisne in "OpenAI is walking away from expanding its Stargate data center with Oracle"

cavisne — Tue, 10 Mar 2026 02:57:07 +0000

Definitely not true. Meta is building tents for GPU's

New comment by cavisne in "Overdose deaths are falling in America because of a 'supply shock': study"

cavisne — Sun, 11 Jan 2026 19:00:47 +0000

Channel 5 did a good piece on "Tranq" which lets China skip the Mexican part of the supply chain as they can mail the finish product straight to the US.

Since the article suggests there must have been a change in china to cause this it seems likely they just moved from fentanyl to tranq.

https://www.youtube.com/watch?v=925wmb-4Yr4&t=1623s&pp=ygUPY...

New comment by cavisne in "Nvidia to buy assets from Groq for $20B cash"

cavisne — Wed, 24 Dec 2025 23:27:05 +0000

I would assume this is a very well paid position, and with basically nothing to do.

New comment by cavisne in "At least $9B billed across 14 Medicaid services in Minnesota may be fraudulent"

cavisne — Sat, 20 Dec 2025 04:58:59 +0000

The state government already failed with Feeding our Future and nothing has changed. The judge who "suggested" the payment of 250m is still serving his term, Jacob Frey (who's aide was indicted as part of it) is still in office, despite a literal vote rigging attempt to primary him.

This should be handled at the federal level, even just to reduce the chances of another jury bribe attempt!

New comment by cavisne in "At least $9B billed across 14 Medicaid services in Minnesota may be fraudulent"

cavisne — Sat, 20 Dec 2025 04:38:33 +0000

There is plenty of fraud to go around but most of it is sophisticated and distributed (ie. lots of individual fraudulent PPP loans from somewhat real businesses exaggerating their expenses). The difference in MN is the fraud is jaw-droppingly stupid. Just read the Feeding our Future wikipedia page [1].

The fraudsters knew they were under suspicion, still got money from the state ,sued the state to force them to pay more - claiming racism, won, a judge forced the state to pay them 250m, blatantly spent the money on sports cars (again - knowing they were being investigated), got caught, tried to pay off the jury, the person who tried to pay of the jury stole the bribe money.

The investigation was wrapped up some time ago, but it did not get a lot of attention for obvious political reasons.

[1] https://en.wikipedia.org/wiki/Feeding_Our_Future

New comment by cavisne in "At least $9B billed across 14 Medicaid services in Minnesota may be fraudulent"

cavisne — Sat, 20 Dec 2025 01:29:44 +0000

It’s not 50% of all Medicaid spending. Minnesota spends about 18B a year (federal + state) on Medicaid). This is an alleged 9b of fraud over 7 years in specific services, so about a billion a year. An organization Minnesota knew was likely fraudulent at the time was still able to steal 250m in a year during COVID so a billion a year doesn’t seem too far fetched.

New comment by cavisne in "A quarter of US-trained scientists eventually leave"

cavisne — Mon, 15 Dec 2025 23:39:10 +0000

“40% move to China”

This is the plan not a coincidence. China pays huge “grants” to their citizens to come to the US, get educated, work in big tech/science, then bring it all home.

New comment by cavisne in "Gemini CLI tips and tricks for agentic coding"

cavisne — Thu, 27 Nov 2025 16:56:39 +0000

I think this guide is mostly AI slop. The Google docs section is full of "they might have support for this" "you could do this", "maybe they will add this MCP". How is that a guide?

New comment by cavisne in "LAPD helicopter tracker with real-time operating costs"

cavisne — Sat, 22 Nov 2025 18:54:44 +0000

Down from a massive peak in 2020/2021 when cities tried the "lets not enforce crime" approach. Still elevated from pre 2020 levels.

New comment by cavisne in "Gemini 3"

cavisne — Wed, 19 Nov 2025 04:39:20 +0000

After ChatGPT accidentally indexed everyones shared chats (and had a cache collision in their chat history early on) and Meta build a UI flow that filled a public feed full of super private chats... seems like a good move to use a battle tested permission system.

New comment by cavisne in "Study identifies weaknesses in how AI systems are evaluated"

cavisne — Sat, 08 Nov 2025 18:13:02 +0000

https://wuu73.org/blog/aiguide1.html

You can get a lot of free usage out of the models.