Casey retweeted
I deeply respect anyone who can update their views in the face of new evidence. The last four years have really taught me how rare it is for someone to be able to publicly change their mind.
Casey retweeted
"Depth Anything V4"
Depth Anything V4 reconstructs dynamic 4D scenes from monocular video using Riemannian Flow Matching over 4D Gaussian parameters.
Instead of forcing rotations, scales, and opacity through Euclidean space, it follows their natural manifolds, cutting invalid Gaussians from 12.3% to 0.01% and improving F1 from 0.762 to 0.806 over an otherwise identical MLP baseline.
Moving from frame-wise depth prediction to probabilistic 4D scene understanding.
alphaxiv.org/abs/2608.18388v…
Casey retweeted
Have you grabbed my #Claude spatial data interrogator skill yet? The goal of this project was to wrap up the most typical spatial data patterns I found in my Claude history to speed up new workflows!
𝐓𝐡𝐞 𝐬𝐤𝐢𝐥𝐥: janosovm.gumroad.com/l/claud…
More: youtu.be/gFsLup-_xrU
Casey retweeted
"The default architecture for agentic AI is increasingly a recursive loop with a credit card attached to it. Give the model a goal. Load a massive system prompt. Describe every tool it might possibly use. Inject a pile of organizational context. Ask it what to do. Execute the tool call. Append the result. Send the entire growing conversation back to the model. Ask it what to do again. Repeat until the model decides it is finished, runs out of context, hits a budget limit, or does something stupid enough that a human has to intervene.
That is the generalized agent loop. It is flexible. It is easy to demonstrate. It can be useful for genuinely open ended work. It is also an extraordinarily inefficient way to run most enterprise processes."
medium.com/@dorians/the-agen…
Casey retweeted
Spatial Graph RAG for the Physical World wherobots.com/blog/spatial-g…
Casey retweeted
InSpatio-World (inspatio.github.io/inspatio-…) looks so great, so our guy @oreki_galahad tested right away. Basically, the pipeline runs 3 steps: Use Florence-2 to generate video captions, estimate depth with DA3, and Run InSpatio-World v2v inference. Here is the result below. Amazing, but still has room for improvement in consistency. But generally I can dream a lot with it.
Casey retweeted
This post is economic illiteracy on full display.
It confuses valuation with cash, wealth with money, and production with redistribution, which is the basic economic error behind almost every grievance like this.
Musk’s “money” exists as ownership stakes in capital goods that coordinate labor, technology, risk, and time across millions of voluntary exchanges.
You cannot “pay Social Security” with factories, code, satellites, or future expectations without liquidating them, collapsing the very structures that generate income and wages in the first place.
Even if you could vaporize that equity into spendable dollars, you would fund a fraction of an actuarially bankrupt program for a brief moment while destroying the productive engine that sustains it.
Wealth is created through entrepreneurial coordination under uncertainty, not hoarded as a static pile waiting to be reassigned. Social Security’s problem is not a shortage of billionaires to harvest but a political design that ignores time preference, demographics, and capital accumulation.
Calling this “taken” is a refusal to understand value creation, price signals, and economic calculation.
The claim sounds moral because it avoids thinking, and it persists because it treats prosperity as loot rather than as a fragile process that collapses the moment you start eating the machinery.
No one is poor because someone else is rich. That framing assumes a fixed pie that never grows, which is the opposite of how markets work. If Musk were actually sitting on a giant pile of idle cash, which he is not, that would signal enormous voluntary demand for what he produces and it would be deflationary.
Idle money lowers prices because it is not competing for resources. You cannot eat money or build with it. You can only exchange it, and exchange happens when value is offered and accepted.
The federal government spends in a single year an amount in the same range as the combined net worth of all US billionaires. Confiscate every dollar, pretend you could convert equity into cash without market collapse, and the state is funded for one year. Just one.
The next year there is nothing left to tax and far less productive capacity to draw from. This is why the state is the worst possible steward of capital. It faces no profit and loss discipline, operates on political incentives, rewards compliance over competence, spends other people’s money, and optimizes for short electoral horizons rather than long-term coordination.
People default to this zero-sum thinking because it feels intuitive and morally satisfying. It offers a villain, absolves personal responsibility, and replaces economic reasoning with resentment dressed up as justice.
It persists because it flatters the believer while demanding no understanding of how value is actually created. It is a huge problem and they should be rebuked accordingly.
Casey retweeted
Last quarter I rolled out Microsoft Copilot to 4,000 employees.
$30 per seat per month.
$1.4 million annually.
I called it "digital transformation."
The board loved that phrase.
They approved it in eleven minutes.
No one asked what it would actually do.
Including me.
I told everyone it would "10x productivity."
That's not a real number.
But it sounds like one.
HR asked how we'd measure the 10x.
I said we'd "leverage analytics dashboards."
They stopped asking.
Three months later I checked the usage reports.
47 people had opened it.
12 had used it more than once.
One of them was me.
I used it to summarize an email I could have read in 30 seconds.
It took 45 seconds.
Plus the time it took to fix the hallucinations.
But I called it a "pilot success."
Success means the pilot didn't visibly fail.
The CFO asked about ROI.
I showed him a graph.
The graph went up and to the right.
It measured "AI enablement."
I made that metric up.
He nodded approvingly.
We're "AI-enabled" now.
I don't know what that means.
But it's in our investor deck.
A senior developer asked why we didn't use Claude or ChatGPT.
I said we needed "enterprise-grade security."
He asked what that meant.
I said "compliance."
He asked which compliance.
I said "all of them."
He looked skeptical.
I scheduled him for a "career development conversation."
He stopped asking questions.
Microsoft sent a case study team.
They wanted to feature us as a success story.
I told them we "saved 40,000 hours."
I calculated that number by multiplying employees by a number I made up.
They didn't verify it.
They never do.
Now we're on Microsoft's website.
"Global enterprise achieves 40,000 hours of productivity gains with Copilot."
The CEO shared it on LinkedIn.
He got 3,000 likes.
He's never used Copilot.
None of the executives have.
We have an exemption.
"Strategic focus requires minimal digital distraction."
I wrote that policy.
The licenses renew next month.
I'm requesting an expansion.
5,000 more seats.
We haven't used the first 4,000.
But this time we'll "drive adoption."
Adoption means mandatory training.
Training means a 45-minute webinar no one watches.
But completion will be tracked.
Completion is a metric.
Metrics go in dashboards.
Dashboards go in board presentations.
Board presentations get me promoted.
I'll be SVP by Q3.
I still don't know what Copilot does.
But I know what it's for.
It's for showing we're "investing in AI."
Investment means spending.
Spending means commitment.
Commitment means we're serious about the future.
The future is whatever I say it is.
As long as the graph goes up and to the right.
Casey retweeted
pithonIsHere redd.it/1o2j8eo
Casey retweeted
A nice free book to learn Linear Algebra for Machine Learning by interacting with vectors. Link in comment.






























