Open weights let you run a model. They do not tell you what trained it.
On Oct. 1, Gregory Kurtzer and Addison Snell explore why AI sovereignty requires verifiable training data and licenses, and how OpenWALDO is building that foundation.
Register: ow.ly/VtHZ50ZREr5
I have decided to relinquish my position as "twitter poool coating expert" and take up my new role as "twitter low flying military jet expert."
Now I can post images of vapor cones and call it breaking the sound barrier.
What is OpenWALDO and Why Did Greg Kurtzer Need to Create It? ow.ly/zbwg50ZIvML
Doug -thanks for all the fish- Eadline retweeted
Chemistry arguably won the 20th century.
Fertilizers helped feed billions. Polymers transformed medicine and engineering. Synthetic chemistry gave us pharmaceuticals, crop protection, semiconductors, batteries, dyes, materials—the infrastructure of modern life.
And yet somewhere along the way, “chemical” became a dirty word.
That’s fascinating.
Physics gave us nuclear weapons, yet people don’t buy shampoo labelled “physics-free.”
Biology gave us pathogens, yet supermarkets don’t advertise “biology-free food.”
But “chemical-free” sells.
There’s a story there about how an entire scientific discipline became associated in the public imagination with contamination, pollution and corporate wrongdoing - even while people live inside a civilization largely made possible by chemistry.
So how did that happen?
Part of the answer is that chemistry became a victim of its own success.
For most of human history, many of the things that killed us were obvious.
Disease.
Hunger.
Infection.
Contaminated water.
Chemistry helped make many of those risks smaller.
But the chemical revolution created new problems too.
Industrial pollution.
Lead.
Asbestos.
Dioxins.
Industrial disasters.
And unlike the benefits, many of these failures had names, villains and images.
A pharmaceutical quietly preventing an infection isn’t a news story.
A river catching fire is.
A polymer keeping blood sterile inside an IV bag is invisible.
Plastic washing up on a beach isn’t.
Fertilizer helping produce enough food for billions becomes normal.
Chemical contamination becomes a scandal.
Over time, something strange happened.
We stopped distinguishing between bad uses of chemistry and chemistry itself.
“Chemical” gradually stopped being a scientific description.
It became a warning label.
Then marketing discovered something incredibly powerful:
Fear sells.
“Chemical-free.”
“100% natural.”
“No nasty chemicals.”
“Clean.”
None of these phrases tells you whether a product is safer.
But they tell you exactly how you’re supposed to feel about it.
And once that framing becomes culturally embedded, chemistry has a serious problem.
Because chemistry is complicated.
Risk depends on dose, exposure, route, duration and a dozen other boring details.
Fear requires none of them.
It just requires a scary name.
So today we have reached a bizarre point.
We demand increasingly sophisticated chemistry to give us better medicines, cleaner energy, safer food, better batteries and more sustainable materials…
…while simultaneously treating the word chemical as something we want removed from our lives.
Maybe chemistry’s greatest failure wasn’t scientific.
It was communicative.
Chemistry helped build the modern world.
But somewhere along the way, we forgot to tell people that story.
Doug -thanks for all the fish- Eadline retweeted
Compilation of failures at World Humanoid Robot Games.
Doug -thanks for all the fish- Eadline retweeted
🦔AI companies have borrowed so much money this year that they're pushing up interest rates for the entire economy. Nomura estimates tech borrowing alone now equals 25% of what the US Treasury issues in bonds, five times more than last year. Bank of America says the surge has added about 0.3 percentage points to the 10-year yield. Bond managers are selling Treasuries to buy AI corporate debt instead because it pays more.
My Take
AI companies are now competing with the US government for the same pool of lenders, and the lenders are picking the corporate bonds. Alphabet's 30-year pays 6.4%. A Meta data center bond pays over 7.5%. At those rates, a 5.2% Treasury loses the fight for capital every time. That's one of the reasons long-term rates have stayed so stubborn even as the Fed tries to bring them down.
JPMorgan expects $5.5 trillion in AI infrastructure spending through 2030, and most of it will be borrowed. That borrowing raises the cost of money for everyone, the government, your mortgage, small businesses trying to get a loan. The AI buildout has reached the scale where it moves rates for the whole economy, and most people paying higher borrowing costs have no idea that a data center arms race is one of the reasons why.
Hedgie🤗
Infrastructure upgrade week. Fiber came to the neighborhood in June. Upgraded some infrastructure in July and BAM!
Doug -thanks for all the fish- Eadline retweeted
In case you missed it, the Chapel parallel language project is currently seeking new sources of funding. Without an influx of new resources, this September’s 2.10 release is expected to be our last for the foreseeable future.
chapel-lang.org/blog/posts/c…
The open-source Chapel programming language project is seeking new sources of funding—or other creative ways of sustaining the project—in order to keep the technology going. For details, please see our announcement on the Chapel blog: chapel-lang.org/blog/posts/c…
I'm not sure what to make of this
backyardbrains.com/products/…
On the one hand, ew cockroaches on the other hand, an army of indestructible little bug tanks, hmm...
I have decided to relinquish my position as "twitter space data center physics expert" and take up my new role as "twitter pool coatings expert." How about that delamination, don't think it is the peroxide per-se, possibly thermal cyling with poor adhesion to primer layer
Doug -thanks for all the fish- Eadline retweeted
My most recent claim is that "FP8 is all you need" for all scientific computing. Two new papers as well as my debate with my great friend Jack Dongarra are covered here: supercomputing.news/hpc/fp64…. (Revised) Part1 paper is on arXiv: arxiv.org/abs/2606.06510 Part2 is on hold at arXiv but you can access it here: dropbox.com/scl/fi/nlvojjpyw… Comments are welcome!
Wrote up the Satochi paper on HPCwire. I tried to pull the important points.
HPC Precision Wars: Satoshi Matsuoka Plants the Ozaki Flag
hpcwire.com/2026/06/15/hpc-p…
Doug -thanks for all the fish- Eadline retweeted
Is FP8 all you need? Satoshi Matsuoka of the RIKEN institute says that it is. Check out reporting from @thedeadline
ow.ly/2LZh50ZbY4L
#FP64 #FP8 #OzakiScheme #lowprecision #highperformance #HPC
Here's what happened at TPC26:
>Opening Plenaries led by Dario Gil (DOE/Genesis Mission) and Katie Antypas (NSF/NAIRR).
>8 breakout tracks
>4 hackathon teams
>7 hands-on tutorials
Watch the recorded livestream On Demand: hhttps://tpc26.org/register-for-tpc26-streaming
Doug -thanks for all the fish- Eadline retweeted
<<FP8 is All You Need - (Part 1):
Debunking Hardware FP64 as the #HPC Holy Grail>>
Prof. Matsuoka argues, based on a unified analytical model and a kernel-by-kernel audit, that the current HPC dogma regarding FP64 is wrong and that FP8 is all you need.
arxiv.org/pdf/2606.06510


























