Tom Marchesello, Ghost Architect at MAGNET. Build analog & digital tools for human superintelligence. Author of Ghost Math & Science Books. Creator SundayPunk.
@desitechsuppoat@MargoinWNC@grok do a financial analysis of the real estate impacts of price and supply and mortgage company and government programs that funded the buyers. How did this happen so quickly when American families could not afford a house or qualify for a mortgage in this same timeframe.
Welcome to a NEW GOLDEN AGE OF SCIENCE.
The Trump Admin just launched the biggest set of science initiatives in decades: $6B+ across government, industry, academia, & philanthropy. 🚀
@DataRepublican Hey i see that you were stepping away from some work with the government. Would it be possible for me to fill in on some of that? I’m a builder and I would be happy to do some of that work.
Google Deepmind just broke chemistry..
They open-sourced a model that can invent completely new biology from scratch.
It’s called “AlphaProtein Novo”
It's an AI pipeline that designs brand new enzymes for "new-to-nature" chemical reactions that have never existed in any living organism.
to put this in perspective..
standard protein ai models (like alphafold) answer the question "what shape will this protein have?". ap novo answers the question "can i make a protein that does this specific chemistry?".
the results from the paper are actually terrifyingly good:
the ai designed an enzyme to synthesize piperidines, which are a critical building block used in pharmaceuticals and materials.
this designed enzyme inverted natural biases to achieve near-perfect regioselectivity.
it produced the chemical building block 99x more often than the competing natural product.
it also designed an enzyme to hydrolyze DEHP, which is a pervasive plastic environmental toxin.
this plastic-eating ai enzyme was 14-fold more active at 90°C than at room temperature, and survived in 75% acetonitrile.
meanwhile, natural enzymes completely failed and were entirely inactive under those exact same extreme conditions.
these new ai-generated designs are up to 60% smaller than natural equivalents.
because they are smaller, they have a much higher mass density of active sites.
i am speechless.. this is the holy grail of synthetic biology.
we don't have to wait millions of years for microbes to evolve enzymes that eat our pollution or build our drugs.. we can just compute them on a gpu today. the ai's success rate in finding functional designs actually matches or surpasses traditional natural sequence screening.
and deepmind open-sourced the whole thing on github.
The OpenAI folder is a same-day prestige dump, not a solved stack. Score it as attention-grade preprint slop: high packaging, unproven mathematics, no machine, no paycheck.
What it is. openai/math went up on 6 October 2026 in one anonymous commit. OpenAI says an unreleased internal model was posed about 4,000 problems, at roughly three hours of ChatGPT Pro thinking per kept result, then filtered to 722 manuscripts in 372 families. The blog is a release note. The Institute for Advanced Study advisory group was consulted on how to publish, not on whether the proofs are true. The README locks a public history so later corrections can still point back at tonight. That is an arXiv-style date stake.
What is not proved. Results are at different stages of verification. Many have no Lean file. OpenAI says the unformalized ones may be wrong. The Re(s) > 11/12 zeta writeup was human-edited. The CM Hodge claim is an exception to the normal procedure and is not in the Lean catalogue. The catalogue status line is “Partial progress.” Comparator configs are staged, including one for the 7/8 zeta claim. Nobody has posted the line that matters: “Your solution is okay.” Their August ten-proofs repo claimed sorry_count: 0. This one does not.
Headline discount. Riemann is a 7/8 zero-free half-plane, not the hypothesis. Hodge is rational Hodge for CM abelian varieties only. Birch–Swinnerton-Dyer is a leading-term formula at Selmer corank 0 or 1. Navier–Stokes in this folder is universal computation in a forced flow, not the Clay smoothness problem.
Category count, 372 families.
Bucket
Families
Share
What they filed
Arithmetic
31
8%
Number theory only
Algebra
32
9%
Algebra 18, group theory 14
Geometry
98
26%
Algebraic/complex 36, differential 29, topology 18, convex/metric 15
Calculus
43
12%
Analysis 16, PDE 16, functional analysis 11
Trigonometry
0
0%
Empty
Other
168
45%
CS 40, combinatorics 37, probability 29, math physics 25, operator algebras 19, dynamics 12, logic 6
Geometry is four subjects, not one: polynomial shapes, curved smooth spaces, topology, and convex bodies. The only lattice in it is a triangular-lattice energy claim. Physics is a garnish: Vlasov–Maxwell existence, Heisenberg magnetization, XY/BKT, Haldane gap, Bose–Einstein condensation, a photon-number inequality. No fusion, no MHD, no photonics, no flight.
Where it points. Not a random punch, and not a hidden engineering stack. The empty regions are the tell: almost no applied numerics, cryptography, learning theory, or ordinary differential equations. What survived is a famous named statement sharp enough to call a resolution or a counterexample. Three vectors: prestige names at the door, formalizable yes/no conjectures in the middle, a thin physics layer so the dump can be read as “the model reached physics.” Direction is inward, toward their evaluation story.
Scores, 0 to 10.
Test
Score
Why
PR / feed / podcast
9
Famous names, big count, same-day fight story
Date-stake / priority claim
8
Version history is explicit
Checked mathematics
1
Staged Lean, no public kernel pass
College-preprint junk
8
Named conjectures, uneven writeups, company warning on the rest
AI slop risk
7
Closed model, abridged “reasoning,” human edit on a headline
Job / money / hire value
1
No spec, no solver, no repeatable measurement
Airplane / coil / instrument
0
Nothing leaves the page
Hidden-stack evidence
2
Gap is real; purpose is not shown
Overall: 8 as a distraction package, 1 as a body of qualified work. A later clean Lean build would rescue that one statement and still not make a machine. Flush the parade. Keep the folder only as a list of claims to audit.
The US government will give your startup up to $1.55M and take 0% equity.
@NSF's America's Seed Fund is open again after Congress reauthorized it in April.
Three tracks:
- Phase I: up to $305K to prove the idea
- Phase II: up to $1.25M to build it
- Fast-Track: up to $1.55M, both phases in one go
Who can apply:
- US-based startups with 500 or fewer employees
- 51%+ owned by US citizens or permanent residents
- Not majority owned by VCs
- All the work done in the US
How it works:
- Send a 3-page Project Pitch first
- NSF invites the best ones to submit a full proposal
- Next full proposal deadline: Nov 4
Phase I funding rates have run about 10 to 20%. It's work, but nobody takes your cap table.
Application link 👉 https://t.co/f5CcZoC1bf
Bookmark and tag a tech founder who needs this.
🚨🤖 NOPE! ABSOLUTELY NOT!
I’m starting to think that movie “Rubber” wasn’t so far-fetched after all!
That coyote better get an Acme anvil and drop it on this ‘Roadrunner’ quick!
Writer: Jamie
“I am announcing the formation of the Super Intelligence Force (SIF). The Super Intelligence Force is tasked with coordinating the effort of the Federal Government to ensure that America continues to lead the World in Super Intelligence, which many say is bigger than the Industrial Revolution, and the Internet, and will protect the interests, and improve the lives, of all Americans.” - President DONALD J. TRUMP 🇺🇸
That’s because H-1B coders wrote the ATS resume screening system and American resumes are screened out to trash and never make it into HR. Then if they slip into HR they purge the American resumes out of the database pool. The H-1B coders then share the Requirement to a head shop who Ai writes a 86% to 98% match of the words to a fake resume. The American resume never gets more than 49% match.