William

What's the story mother?

[William @narrowfail]

When everybody tells you to buy and hold ETFs, is that a shoeshine boy signal?

[Charlie Bilello @charliebilello]

Gas prices in the US have moved up to $4.43 per gallon, their highest level since July 2022. The 49% spike over the last 9 weeks ($2.98/gallon to $4.43/gallon) is the biggest we've seen in the past 30 years.

[Claudia Sahm @Claudia_Sahm]

Interviewer: Do you think the tension between the White House and the Fed goes away when Warsh becomes Chair.

Me: No. It continues (form may change some) until the President backs off and lets the Fed do its job. And I don’t expect that.

[Citrini @citrini]

What are we thinking from here?

[Autism Capital 🧩 @AutismCapital]

Checking gas prices today.

[Charlie Bilello @charliebilello]

All 7 members of the Magnificent Seven are down on the year (-5% for Nvidia to -23% for Microsoft) and underperforming the average stock in the S&P 500 (which is up 1%). $MAGS

[William @narrowfail]

@PurpRoad @unusual_whales How about AA treasuries?

[Om Patel @om_patel5]

stop spending money on Claude Code.

Chipotle's support bot is free:

[ぺんぎんさん@C107新刊BOOTH通販はじめました @penguin2716]

これ作れば万事解決やね

[Kakashii @kakashiii111]

Nvidia is going to meet yet again the holy Beat and Raise this quarter, but what is going on behind the scenes in order to maintain it? Because if not, "the world would have fallen apart."

Just look at SMCI's Balance Sheet.

Accounts Receivable grew explosively 435% QoQ. There is one customer who is brand new according to SMCI's filing, or at least he has not accounted for more than 10% of SMCI's sales in the last two quarters. And he is responsible for 71.6% of SMCI's AR, which is almost $8B! And this brings us to Accounts Payable, where SMCI didn't even pay for this server order to its suppliers (!), bringing the Accounts Payable to almost $14B, a huge increase of 1074% QoQ (!!!)

And this is not the end. SMCI, while sitting on huge servers in warehouses, keeps stockpiling and building servers, now sitting on over $10.5B worth of Inventories, a growth of 85% QoQ (!).

As Jensen says: "sky high demand."

And this is just SMCI, without including Mag7's CIPs, the army of Neoclouds, and mountain-debted hyperscalers.

[vx-underground @vxunderground]

Bill Gates was friends with Jeffrey Epstein.

Guess who wasn't?

[William @narrowfail]

Hi @michaeljburry — curious what your take is on the dramatic rise in silver prices and recent reports of major financial institutions holding large short positions in silver futures. Do you see risks of a squeeze or systemic implications for big banks if those shorts unwind?

[William @narrowfail]

@Barchart Pump & Dump retail brokers 😹

[Mark Zandi @Markzandi]

The Bureau of Labor Statistics’ estimate of November consumer price inflation, which it released last week, is badly flawed. So much so, we constructed our own estimate of CPI inflation (courtesy @MattColyar). Inflation didn’t decelerate to 2.7% on a year-over-year basis in November as the BLS reported but instead remained unchanged at 3.0%.

[Alex Veremeyenko @alex_verem]

This paper from Harvard and MIT quietly answers the most important AI question nobody benchmarks properly:

Can LLMs actually discover science, or are they just good at talking about it?

The paper is called “Evaluating Large Language Models in Scientific Discovery”, and instead of asking models trivia questions, it tests something much harder:

Can models form hypotheses, design experiments, interpret results, and update beliefs like real scientists?

Here’s what the authors did differently 👇

• They evaluate LLMs across the full discovery loop hypothesis → experiment → observation → revision
• Tasks span biology, chemistry, and physics, not toy puzzles
• Models must work with incomplete data, noisy results, and false leads
• Success is measured by scientific progress, not fluency or confidence

What they found is sobering.

LLMs are decent at suggesting hypotheses, but brittle at everything that follows.

✓ They overfit to surface patterns
✓ They struggle to abandon bad hypotheses even when evidence contradicts them
✓ They confuse correlation for causation
✓ They hallucinate explanations when experiments fail
✓ They optimize for plausibility, not truth

Most striking result:

`High benchmark scores do not correlate with scientific discovery ability.`

Some top models that dominate standard reasoning tests completely fail when forced to run iterative experiments and update theories.

Why this matters:

Real science is not one-shot reasoning.

It’s feedback, failure, revision, and restraint.

LLMs today:

• Talk like scientists
• Write like scientists
• But don’t think like scientists yet

The paper’s core takeaway:

Scientific intelligence is not language intelligence.

It requires memory, hypothesis tracking, causal reasoning, and the ability to say “I was wrong.”

Until models can reliably do that, claims about “AI scientists” are mostly premature.

This paper doesn’t hype AI. It defines the gap we still need to close.

And that’s exactly why it’s important.

[Ben Carlson @awealthofcs]

We’re gonna do data centers in space and you’re bearish?

[Lyn Alden @LynAldenContact]

That awkward moment where investors ask themselves why they paid 55x earnings for Costco.

[William @narrowfail]

@soloun4chica Hay mucho lobby por impulsar la energía nuclear y los jóvenes la están consumiendo principalmente por redes sociales. Los deshechos son un gan problema. EEUU nos prestó un reactor nuclear y estuvo en Malvin norte. Nosotros no tenemos la tecnología para construirlo tampoco.