Yeah sure, was trying to agree with you. Chess people talk about “engine moves” - superior moves that Stockfish at depth 40 will play but that a human could never come up with. seems inevitable that security research will get there eventually too.
This to me seems at about the limit of what people could develop: projectzero.google/2021/12/a-de...
No reason to believe that's an upper bound on the complexity or subtlety of possible attacks.
Looks like it's for lease, kinda makes me want to found a company just so I can sit in Noyce's spot
Amusing that “Tesla fire” exists as a news category separate from “car crash”
This is extra funny because I'm assuming the Pope is a Bulls fan?
Just fyi the Knicks were up 8 with five minutes left and then got outscored 15-6
View across the salt pond towards Moffett Field, where the restored Hangar One is visible
Got a little wet and muddy to do it, but visited my 11th Midpen preserve of the year: Stevens Creek Shoreline Nature Study Area
www.openspace.org/preserves/st...
I know @reckless.bsky.social will say chatgpt still has way fewer users than the iphone but I'm afraid the overwhelming revealed preference is that normal people really do want AI chatbots.
Agree that Anthropic with $30B in revenue is not pets.com with near-zero revenue
I don't think this is quite right - there is a lot of capex from:
- labs (openai, anthropic, etc)
- traditional hyperscalers (google, AWS, etc)
- neoclouds (lamba, coreweave, etc, etc)
and there are “startups” raising 9 figure “seed” rounds
So the personal news: I joined the RAP1 team at Rivian
rivian.com/newsroom/art...
“I wish my bike were cooler than all my riding buddies”
Genie: “you still have three wishes”
The Undertaker walking into WrestleMania
me walking up to the service counter to pick up my bike
“yeah it's the cervelo with purple tires”
A trail leading past a tree into a hilly landscape
Checking off my next Midpen preserve for 2026 - this weekend, Pulgas Ridge
www.openspace.org/preserves/pu...
You're not wrong, but also when I hear you for example talk to @crampell.bsky.social about tariffs on air, it's also quite different from how you guys post. Sanewashing adjacent at least.
Does this really match your model of, say, Google's behavior? They get high on their own supply and stubbornly refuse to get rid of products?
My original point is just that it doesn't make sense to assert that the whole tech industry is collectively throwing away hundreds of billions of dollars on a technology that no one wants. People want current AI products and are willing to pay a lot for them!
No? True, you probably can't get future allocations of DRAM, but you also haven't been able to get real physical DRAM for a while now. Someone really has been buying it and a lot of new data center compute actually has come online.
Does it look like AWS is crumbling under exponentially growing variable costs?
I'm making some assumptions but Amazon/Google/Microsoft break out cloud financials and if there was a huge jump in AWS revenue and profit from Q4 2024 to Q4 2025, I'm willing to go out on a limb and say it wasn't because people used more EC2 VMs.
This is just wrong. Inference HW gets more efficient, KV caching gets smarter. Products like Amazon Bedrock and GCP Vertex AI already have wildly positive operating margins. They're sacrificing near term cash flow to build even more and better infra.
Seriously, look at Azure/AWS/GCP financials.
Does this really match your model of, say, Google's behavior? If a product isn't getting traction they'll just keep throwing good money after bad?
Every data center near me (which to be fair basically means Santa Clara, CA) seems to be pretty well-staffed with private security? Much more than, say, banks or jewelry stores, at least.
I'll grant that saying one person is smarter than another is hopelessly vague and imprecise in the same way as saying one person is bigger than another
If the argument is that intelligence is just as fake as athletic talent then idk
It doesn't seem useful to me to deny that the “size” of a person is a valid thing to talk about.
Similarly the diversity of cognitive tests with high g loading seems to me to say that there's something real there, although yes we can debate what it is exactly
I very, very much agree that it is critical to keep intelligence separate from any notion of worth or moral weight.
But I don't think it's meeting the world honestly to deny that some people are better at cognitive tasks than others; i.e. “smarter,” in everyday language.
That's an argument I find intuitively appealing, but as I said, it really isn't what modern psychometrics finds empirically
Sure, but my point is that everyone knows what we mean when we say one person is taller than another. And I don't think it should be harder to accept the common sense meaning of someone being “smarter.”
