S V

Math, Physics, CS and chill. Current: Having fun working on esoteric ideas. Previous: Ads FE TL @Google NYC, Asst. Researcher at MSR, CSE @IITKgp.

[S V @open_parens]

Einstein deconflated two notions of mass* - inertial and gravitational - and conceived General Relativity. Surprisingly, this kind of deconflation is often necessary to keep large software systems healthy over time. I explain it in [1] under "Evolutionary Conflations".

In my experience, the wider you read, the more the ideas from various fields converge. This is one such example and I can quote a few more. The human toolbox is vast but _finite_. So, pen and paper intelligence is likely to get automated completely and soon. Physical intelligence (like robots, building things) will come, but a bit later.

* Inertial mass determines the resistance to change and Gravitational mass determines gravitational force. That they are identical is not necessary (they are not identical with electric charges).

[1]: https://t.co/iptT6LIwD0

[S V @open_parens]

@A_G_I_Joe with a #masterclass on General Relativity in 1 hour. I also recommend his "Training sand to think" talk [1].

With the way he breaks it down, the building blocks appear so elementary as to be within reach of even today's models. Automated physics research is inevitable.

[1]: https://t.co/bZoQSmW1Kl

[S V @open_parens]

C++ should simply declare bankruptcy at this point and start again.

[♡ mari/cohe ♡ @noinconsistency]

C++26 reflection syntax looks abhorrent

[S V @open_parens]

@kkmaway @TheStalwart "We made Google dance". Seems like he lost focus after delivering this line. But, I wouldn't underestimate him.

[S V @open_parens]

Latency vs Throughput. Google is not set up for a _short_ pipe, but a _wide_ one. A mega wide pipe.

[tae kim @firstadopter]

Moonshot founder Yang Zhilin criticizes Google's bureaucratic corporate culture:

"Google can produce Transformer, but it can't produce ChatGPT. Its organization can't do it"

[S V @open_parens]

Excellent. I don't know what it is, but great engineering warms my heart like little else. It gives me a little kick to be part of humanity despite the obvious depravity of humans at large.

[ajaxthegreat @AlexanderJ91756]

We’ve been working extremely hard to make the strongest lightest and cheapest heat shield ever produced and today shows just how far we’ve come in creating the perfect heat shield for a reusable vehicle that will take humans farther than they’ve ever gone. I couldn’t be prouder

[S V @open_parens]

Very cool. Great teams are rare. Just a handful of great people coming together can do magic.

[Tito @giga__tito]

I am extremely lucky to get to work with every single one of these dudes. This is the hardest-working team I’ve ever been a part of!

Starship engine install team is the coolest team on the planet! 🚀🌎📍

[Zack @BLKMDL3]

Tesla Diner live-streaming Starship Test Flight 13 on their large screens! @SpaceX @tesla_na

[S V @open_parens]

Indeed. CSS is diabolic. Flutter sits on a much better Pareto frontier.

[nixCraft 🐧 @nixcraft]

"How does CSS work?"

New developer: Explains the CSS specification...

15-year frontend veteran: Who the f**k knows? Keep changing margin-top until it looks right 😅

[S V @open_parens]

The cost curve on rocket launches is unbelievable. In 5 years, we might be looking at 100$/kg to space and satellites will be built using off the shelf parts instead of custom light weight parts. Brilliant.

[Scientific American @sciam]

Starship demonstrated key abilities on its path to becoming a fully reusable transport for astronauts to the moon
https://t.co/RIc21C7I5K

[S V @open_parens]

UGC regulations have endangered the lives of male students of Unreserved category. Unless there is an official apology, a large part of their base will not vote for BJP. UP elections in Feb 2027 will reflect this reality.

[Brian Roemmele @BrianRoemmele]

BOOM! OPEN SOURCE MRI!

You can now 3D-print the core of an MRI scanner.

A machine that hospitals pay $1.1 million to $3.4 million for has been broken open. The OSI² ONE and its educational siblings deliver real images of heads and limbs for a fraction of the cost, using a permanent-magnet Halbach array, 3D-printed structures, and fully open designs.

This is not a toy or a simulation. Working systems already produce in-vivo images in Leiden, Utrecht, Berlin, and Uganda.

The magnet alone—396 carefully oriented neodymium cubes in a cylindrical Halbach array—costs about $1,370. A complete scanner lands between $28,500 and $68,000 depending on the console and coils you choose.

No superconducting magnets. No liquid helium. No specialized power infrastructure. It runs from a standard wall outlet and weighs roughly 150 kg.

The physics is elegant. A Halbach array arranges permanent magnets so their fields reinforce inside the bore and nearly cancel outside. The result is a usable 50 mT field strong enough for diagnostic-quality imaging of extremities and the head when paired with clever gradient coils, RF coils, and modern reconstruction. Spatial resolution reaches about 1.5 × 1.5 × 5 mm³.

The designs are modular: build the magnet first, verify and shim the field with a 3D-printer-turned-field-scanner, then add gradients and RF hardware.

The plans are public
Everything needed to replicate or improve the system lives in open repositories:

• Primary project hub and documentation: https://t.co/WgzAqQMF3T

• Full OSI² repositories (hardware, software, magnets): https://t.co/y42zuQDdPn

• Educational build focused on the Halbach frame, shimming, gradients, and student workshops (Utrecht / Lili’s Proto Lab): https://t.co/DTpxboUOkI

• Magnet-specific details: https://t.co/vlE7qYDLZZ

Hardware is released under CERN-OHL-W. Most software is GPL-3.0.

Where AI multiplies the impact
Low-field MRI has historically been limited by lower signal-to-noise and greater field inhomogeneity. That is exactly the regime where modern AI thrives.

Image reconstruction becomes dramatically better when deep networks trained on high-field data or physics-informed models denoise, correct for inhomogeneity, and push resolution beyond the raw acquisition limits. Real-time sequence adaptation can adjust gradients and RF pulses on the fly as the AI monitors signal quality.

Magnet design itself can be optimized by evolutionary algorithms or differentiable physics engines that search for better Halbach geometries or shim placements than human intuition alone can find.

Further out, local AI agents turn these scanners into autonomous diagnostic nodes. A small clinic or even a well-equipped garage workshop could run overnight scans, flag anomalies, and queue results for a remote radiologist—or eventually for a specialized medical model.

Synthetic data generation from the open designs lets researchers train robust models without proprietary hospital datasets. Robotics integration (patient positioning, coil placement, maintenance) becomes straightforward once the hardware is open and standardized.

In the longer arc of the Abundance Interregnum, this is the shape of things: sophisticated medical instruments that no longer require billion-dollar supply chains or national infrastructure.

A distributed network of open, AI-augmented low-field scanners could bring advanced imaging to places that have never had it, while simultaneously giving makers, universities, and small labs the ability to experiment, improve, and specialize the technology.

The plans are already on the table. The magnets are commercial off-the-shelf. The 3D printers exist in thousands of workshops. The AI tools for reconstruction and design optimization improve every month.

What was once the exclusive domain of major hospitals is becoming a community engineering project.

This is how abundance arrives—one open, reproducible, AI-extendable system at a time.

[S V @open_parens]

Yep. QE (fancy name for money printing) is the single worst thing to happen to the world. Bernanke, Yellen and Powell have destroyed value signals from the economy and that's why life feels like an increasingly difficult treadmill for many people.

[Ritesh Jain @riteshmjn]

This is probably my most important post.

The FED stole your future and there is no going back

"The system is rigged. The deep state does not want us to be free. The American dream is dead."

Statements like these conjure images of deep pessimism, a worldview where you have no agency, where you are merely a puppet dancing for malignant powers you cannot see or touch. We are not people who live in that camp. But sometimes, certain data points are so damning that they leave us no choice but to admit: something is seriously wrong, and it needs to be laid out in the open.

Every time I visit India now, I find people agitated. Even those in the top 10% of the income bracket, earning anywhere from ₹50 lakhs to a crore per year, feel like they are running on a treadmill that keeps accelerating. No matter how fast they move, it is never enough. At the ground level, the situation is far worse. It is the same story everywhere. In Canada, both partners in a household work full time and still fall short each month. In Australia, young professionals earn well and own nothing. In Germany, the middle class quietly shrinks. The geography changes. The exhaustion does not.

And the origins of this mess are not in New Delhi or Ottawa or Berlin. They are in Washington D.C. All of us are paying the price for a policy disaster handed down from ivory towers, by people most of us never elected and, frankly, never even saw.

Consider this: the U.S. money supply (M2) grew by 40% in just 2 years

*The Federal Reserve United States Money Supply M2*
January 1, 2020: $15.4 trillion
January 1, 2022: $21.6 trillion
A staggering ~40% increase
As of Mar-26, $ 22.6 Tn

( so they never reversed the increased money supply although Covid got over)

Unprecedented in the history of the Federal Reserve post-World War 2 era. (Source: FRED) This massive injection of liquidity created asset bubbles across the economy. Wages stayed stagnant. Those who owned capital benefited enormously. Everyone else got the inflation.

Most people have not yet identified the cause of their frustration, but they have begun to feel its effects viscerally. And that feeling, that the system simply cannot deliver on their aspirations, has become the quiet tailwind driving a very dangerous behavioural shift.

The more people sense that conventional paths are closed off, the more they reach for asymmetric bets, even knowing the odds are stacked heavily against them. The explosion of betting apps and prediction markets, Kalshi, Polymarket, Dream11 and their many cousins, are not trends. They are symptoms of a broken economy. The feverish rise in F&O trading and the massive uptick in exchange volumes are different expressions of the same underlying truth: when people stop trusting the system to reward honest effort, they start gambling on outcomes instead.

[S V @open_parens]

Google will reach the frontier. From my experience, Google is a beast at executing steadily. Structurally Google can't do 50% improvements per year, but not many can beat them at consistently doing 20% every year. Model development pace has been slowing and this favors Google.

Monthly cadence means they now have the full pipeline in place. I remember when I originally joined Google my product used to do weekly releases. We invested and moved to daily releases and it was so much faster to iterate on. A massive API surface and a massive UI but daily releases were smoother than weekly releases.

[The Transcript @TheTranscript_]

$GOOGL Alphabet CEO Sundar Pichai says they plan to approach a monthly model-release cadence after establishing Gemini 4 as the next platform baseline.

“With that, we are creating a baseline, on top of which you will see us rapidly iterate with subsequent model releases. Picking up the pace and releasing models almost at a monthly cadence is part of our roadmap, as we are building Gemini 4 as well.”

[S V @open_parens]

Glad to know that there are visual math people. I am one of them and have on occasion wished for more symbolic orientation. But I guess you are born/raised in a certain way and that stays.

[Jake Brukhman @jbrukh]

FYI, I studied graduate level mathematics at Courant Institute of Mathematical Sciences at NYU.

Being a visual learner in a symbolic field, in retrospect, greatly limited my progress.

If I had back then the AI teachers of math we have today, I suspect I would have gone way farther in pure.

[Guanghan Ning @quietnning]

Opus 5 reports 30% on ARC-AGI-3, ~4× the previous best model, ~20× its predecessor Opus 4.8. We tested it on Witness, our held-out suite of ARC-AGI-3-style interactive puzzle games. The leap doesn't transfer.

On Witness composites (same harness, same budget for every model), Opus 5 lands at 43.4 ± 3.2, which is a statistical tie with kimi-k3 (42.8 ± 1.9) and Fable-5 (43.8 ± 9.7). Ahead of Opus 4.8 (34.8), but nowhere near a generational jump.

The traces tell the why:
(1) On our most classic Witness-style game, Opus 5 states the hidden rules before its first action, then plays a byte-identical optimal solution in 5/5 seeds at temperature 1.0. Zero exploration. It already knows this genre.
(2) But on our most novel game (unusual mechanic combinations you can't pattern-match), Opus 5 regresses below Opus 4.8. Where rules must actually be discovered through interaction, the new model is worse than the old one.

That decomposition (perfect on templates, regressed on novelty) is the signature of “scaffold-then-internalize” training on genre-specific data, not a general gain in interactive abstract reasoning. Our benchmark can't tell whether that data was their in-house ARC-AGI-3-like corpus with ARC-AGI-3-specialized harnessing (likely thanks to [schema]? https://t.co/6vrofY9Zx1), or public Witness-genre corpus, or both, but it can tell the improvement isn't general.

Held-out evals only stay held-out while nobody's optimizing for the genre, and that clock is always ticking. It's ticking for Witness too, the moment we publish it.

[SciTech Era @SciTechera]

HUGE breakthrough reminder:

Scientists just built stable "Boron Graphene" 👀

'Researchers at Tohoku University realized a stable 2D honeycomb boron electronic system using the crystal LaRh₃B₂ (Lanthanum Rhodium Boride).

'The material behaves like the long theorized "boron graphene," but with much stronger electron interactions than ordinary graphene.'

'Using ARPES and STM/STS, the team discovered a quantum liquid crystal state, where electrons spontaneously align in one preferred direction.'

'They also observed a van Hove singularity near the Fermi level, a key signature of strongly correlated quantum materials and a feature often linked to unconventional superconductivity.'

'This breakthrough provides a powerful new platform for designing next generation quantum materials, exploring high temperature superconductivity, and developing future quantum technologies.'

[S V @open_parens]

I think INR will hit 110 in the next couple years. India has tremendous founders and potential, but it will remain potential, not realized, unless both education and justice system are reformed root and branch.

[🚨Indian Gems @IndianGems_]

"So what if the rupee hits 100 to a dollar? It's just a number."

— Dr Shamika Ravi, PM Economic Advisor

[S V @open_parens]

It's time to end reservation in India. #EndReservation India is facing severe threats from China and the US both economically and militarily. We can't have a divided India. Reservations cause resentment and division when a much better ranked student has to give up his seat to a rich student who gets half the marks.

[S V @open_parens]

@AvneeshKasture @sekar_vembu I suggest you learn rigorous mathematical thinking. Handwaving won't get you far with me.

[S V @open_parens]

The state has monopoly on violence (via justice system). No free market has addressed that. Beyond that, a blind adherence to free market is suboptimal. Capitalism also fails (e.g. the gilded age) and requires an anti-trust mediator imo.

Every system I know of has serious failure patterns. We need to subject any proposals to rigorous simulation testing.

[S V @open_parens]

I don't know if this is a deliberate strategy by China but if you are a committed technologist, producing goods cheaper over time will be your number one goal. That's the ideal of captalism. Looks like China is pursuing this ideal while US capitalism turned extractive and crony a long time ago. I don't have any connection to China and I love the idea of the US, but the public data suggests that China seems to be doing capitalism better than the US.

[Fireside Alpha @firesidealpha]

David Friedberg says China is deliberately commoditizing the knowledge economy because it already holds 20X our manufacturing capacity and is heading to 8X our electricity.

"I'll just zoom out for a second. Think about the strategy for China."

"If you think about the global economy of the last 50 years, the US has accrued so much value by being at the core of the knowledge economy, and effectively a services economy."

"Through the development of intellectual property, of IP, of knowledge, and the conversion of one bit to another bit, we've been able to derive trillions of dollars in GDP."

"Meanwhile, we outsourced manufacturing and created a sleeping giant in China, where they have this incredible manufacturing capacity..."

"All of technology ultimately leads to that simple equation: molecule conversion... Everything in our world is driven by molecule conversion..."

"And what's left is the value of the molecule economy. When you look at the juxtaposition of China versus the United States today, we have one terawatt of electricity production capacity in the US, and they're on their way to having eight."

"We have about 10 billion square feet of manufacturing capacity; they have 200 billion... So they have 20X the manufacturing capacity, 8X the electricity production."

"And I think that's the long game for China: over a two, three decade process, by compressing the knowledge economy and the services economy, commoditizing it completely, they are left holding all the value in the global economy because they can make stuff, and they can make it cheaper than anyone because they have the most electricity production."