2xdog retweeted
Replying to @2xdog
I wish I could screenshot my post in another chat with your post.
+1 same page
2xdog retweeted
Flashback to June 14, 2024.
At the time, frontier AI models have zero fluid intelligence. They score around 5% on ARC 1. They can't reliably add two numbers. They make laughable logic errors every other prompt. Reasoning models don't exist. Coding agents don't exist. Claude Code doesn't exist. "Code gen" is a complete joke. Scaling up next-token prediction based on web text data is hitting a wall.
What if a time traveler told you that, by November 2026...
• AI models would have human-like fluid intelligence, thanks to a new paradigm: test-time induction (modern reasoning models).
• Not only would ARC 1 be saturated, but unreleased benchmarks 10x harder (ARC 2) and 100x harder (ARC 3) would be saturated too.
• Nearly all code would be written by AI, and models could one-shot complex codebases flawlessly.
• AI models would be solving multiple Millennium Prize problems.
• OpenAI's valuation would be 15x higher, and Anthropic's would be 85x higher.
And on top of all that...
• Nvidia would still hold ~80% of the AI compute market.
• Nvidia would have grown its revenue and earnings 4.5x.
• Nvidia would still be growing 75% a year.
• Nvidia would have even higher operating margins (67% vs. 65% back then).
Based on that information, straight from the future, would you have bought NVDA?
If you answered yes -- congrats. Today, 2.5 years later, you'd be down 5% on your investment (even more after inflation).
2xdog retweeted
Anthropic: The Committee of Public Safety IPO
Anthropic went public today.
It’s the first company built by “effective altruists” to go public. Thus, I’m certain most investors have no idea what they’re buying, or how dangerous the company’s charter is, not only to investors, but to humanity.
Effective altruists believe that capitalism is a flawed system because it relies on self-interested actors competing in markets. They believe they’ve created a better system, a way of creating value that’s based on doing what’s best for the “common good,” not greedy shareholders.
Anthropic is a Delaware “public benefit corporation.” Its charter assigns it a purpose: the responsible development and maintenance of advanced AI for the long-term benefit of humanity.
Its directors are required to balance three things: the financial interests of shareholders, the interests of everyone materially affected by the company’s conduct, and that purpose.
What in the heck does that actually mean?
It means that the company’s directors can do whatever they want. They aren’t accountable to the shareholders. They’re accountable “everyone,” which effectively means they’re not accountable to anyone.
A director who sacrifices your returns for the long-term benefit of humanity is deemed to have met his duties so long as the decision is informed and disinterested. Anthropic’s own prospectus concedes there is limited judicial precedent on how courts will judge that balancing. Thus, Anthropic’s board is completely unaccountable to its owners.
And… who selects the board…?
A Long-Term Benefit Trust holds a single share of Class T stock. That share carries no economic rights and no expiration date. It gives the trust the power to elect and remove four of Anthropic’s seven directors: a majority of the board. The filing describes the trustees as financially disinterested,with backgrounds in AI safety, national security, public policy and social enterprise. The men and women who pick the majority of the board of a $1.5 trillion company have not a dollar of their own at risk in it.
And the other directors?
The seven co-founders hold a single share of Class F stock through an entity called Founder LLC. That one share carries 50.1% of the vote on nearly everything else shareholders decide, including the election of the other three directors. The filing says that voting power will not be subject to dilution. Counting their own shares, the founders control about 61.3% of the vote. Each has pledged to give away 80% of his or her stock. A founder keeps a seat in the LLC so long as he holds just 2% of his IPO-era shares.
One of the most powerful technology companies that’s ever been created is going to be controlled by a group of completely disinterested directors who believe they know what’s best for the common good.
What could possibly go wrong?
Here’s what one of the founders of the world’s largest altruism movement believes about the near-term future, thanks to technology:
“Man at last will begin to harmonize himself in earnest… He will try to master first the semiconscious and then the subconscious processes in his own organism, such as breathing, the circulation of the blood, digestion, reproduction, and, within necessary limits, he will try to subordinate them to the control of reason and will… in order to reduce the fear of death to a rational reaction of the organism towards danger… Man will become immeasurably stronger, wiser and subtler; his body will become more harmonized, his movements more rhythmic, his voice more musical... The average human type will rise to the heights of an Aristotle, a Goethe... And above this ridge new peaks will rise.”
Dario Amodei, the chief executive of Anthropic (ANT), believes exactly the same ideas.
In 2024 he published an essay called Machines of Loving Grace. In it he wrote virtually the same words: weight, physical appearance, reproduction, and other biological processes will be fully under people’s control. He predicts the next 50-100 years of biological progress in 5-10 years, and a doubling of the human lifespan to 150.
What’s the problem?
That first description of the coming technological revolution (above) was written in 1924 by the leader of the world’s first effective altruism movement, Leon Trotsky. It’s from the closing pages of his book Literature and Revolution. Yes, the communists were avowed altruists. They sought to create a “perfect man” that would always act for the common good. “The average human type will rise to the heights of an Aristotle, a Goethe, or a Marx.”
Trotsky built the Red Army. In September 1918 he ordered the families of officers who defected to the enemy be arrested, so every officer knew his wife and children were hostages. Then he hung them all. For the greater good, of course.
These kinds of revolutions always devours their own. Stalin had Trotsky expelled from the party in 1927 and deported from the Soviet Union in 1929. In 1936, a Moscow show trial sentenced him to death in absentia. In May 1940, Stalinist gunmen sprayed his bedroom in Mexico City with machine-gun fire. Then, on August 20, 1940, Ramón Mercader, a Soviet secret-police agent who had worked his way into Trotsky’s household, drove a mountaineering ice axe into his skull as he sat reading at his desk.
Effective altruism, like communism, began with an appealing idea: give to charity using evidence, not sentiment. Its leading thinkers concluded that because the people who will live in the distant future vastly outnumber the people alive today, the most important moral work is preventing events that could wipe out that future.
Of course, if you accept that logic, then virtually any present sacrifice can be justified by the trillions of hypothetical lives on the other side of the ledger. It’s a preposterous idea. And it reveals a profoundly naïve false belief in their own judgment.
In March 2022, the economist Tyler Cowen asked Sam Bankman-Fried, then effective altruism’s most famous donor, whether he would take a bet with a 51% chance of doubling the Earth and a 49% chance of destroying it, and keep playing double or nothing. Bankman-Fried said yes. Eight months later his exchange, FTX, collapsed. A jury convicted him on seven counts of fraud and conspiracy. A federal judge found Bankman-Fried stole $8 billion and sentenced him to 25 years.
It’s arrogance to the level of insanity to believe that you know what’s best for everyone. It’s arrogance beyond the level of insanity to believe you have the right to other people’s property because you know what’s best for them.
Look at Anthropic's charter. Is there a meaningful difference between what Sam Bankman-Fried did and what Amodei has done with the structure of Anthropic? I don't think so. These people believe they know what's best for everyone. And they're going to do what's best for the "common good."
And where do these ideas always lead? To totalitarian systems of control.
In 1762, Jean-Jacques Rousseau wrote that society rests on a general will, and that whoever refuses to obey the general will shall be compelled to do so by the whole body. Thirty-two years later, his most devoted student, Maximilien Robespierre, told the National Convention: Terror is nothing other than justice, prompt, severe, inflexible; it is therefore an emanation of virtue.
Robespierre’s Committee of Public Safety arrested about 300,000 people. Roughly 17,000 were tried and executed. Thousands more died in prison or without trial.
The Bolsheviks replaced the general will with the dictatorship of the proletariat. Same idea: a vanguard that could see the good the masses could not.
Mao Zedong’s Great Leap Forward promised to move China into industrial abundance – for the common good. Planners in Beijing set the targets, and the peasants starved. Demographers estimate the famine of 1958 to 1962 killed somewhere between 15 million and 45 million people.
You might object that these were fanatics with guns, and that the people running Anthropic are scientists with data centers. But, it’s only a matter of time before they will create a reason to start killing people.
In the early twentieth century, American scientists, reformers and philanthropists embraced eugenics: the idea that society could breed out disease, poverty and crime with the help of science. It called itself evidence-based. It was impartial and long-termist, concerned with the welfare of generations not yet born. It was all for the common good.
The Supreme Court ruled on it in 1927. By a vote of 8 to 1, the justices let Virginia sterilize a young woman named Carrie Buck. Justice Oliver Wendell Holmes, the most celebrated legal mind of his era, wrote: Three generations of imbeciles are enough. Somewhere between 60,000 and 70,000 Americans were sterilized against their will.
C.S. Lewis best explained the unique danger of these kinds of ideas. "Of all tyrannies, a tyranny sincerely exercised for the good of its victims may be the most oppressive... those who torment us for our own good will torment us without end for they do so with the approval of their own conscience."
Alas, there is no such thing as the common good.
There are only goods, plural: the particular wants of particular people, which conflict constantly.
You want a cheaper car. The autoworker wants a higher wage. The retiree wants a safe yield, and the founder wants your savings for his start-up.
No committee – and no data center – can add those wants into a single number called “the good” and then maximize it. The wants are not knowable in advance, even to the people who hold them, until they have to choose.
But you cannot choose for others. You can only choose accurately for yourself.
That is, acting in your own best interests is a vital part of the system.
Friedrich Hayek spent his entire career explaining why self-interest is integral to a functional free market. The curious task of economics, he wrote in 1988, is to demonstrate to men how little they really know about what they imagine they can design.
The knowledge that runs an economy is not understood by one man or group of men. It is not contained by any machine – not even a superhuman AI.
The knowledge that runs an economy is local, tacit and scattered. It surfaces only when people with something at stake make choices.
That is how what is “good” gets discovered: one transaction at a time, by people who bear the consequences of their choices.
Adam Smith wrote the most famous sentence in economics: It is not from the benevolence of the butcher, the brewer, or the baker, that we expect our dinner, but from their regard to their own interest. The merchant, Smith wrote, intends only his own gain, and he is in this, as in many other cases, led by an invisible hand to promote an end which was no part of his intention. Then comes the line that every American should know: I have never known much good done by those who affected to trade for the public good.
Anthropic’s shareholders are about to learn this lesson the hard way. And I hope they're the only victims.
2xdog retweeted
Bitcoin is dead right now. It fell 56%.
Let's check back in 2-3 years during the next rising liquidity cycle and PMI cycle.
Replying to @LynAldenContact
Bitcoin is dead right now. It fell 70%.
Let's check back in 2-3 years during the next rising liquidity cycle and PMI cycle.
I still think we can look for dead cat entries but need to be very careful here.
At risk of sounding like Ari Paul here but… Kitchen sink/Lou Pai delta is still low-ish (I’m going to say 15%) but it’s up from like 2% before this and it’s total Armageddon if that’s the “why now” on these affairs.
Replying to @jbooger91 @dimndimn
We are here now because the hyperscalers ran out of fcf to fund this and credit markets have revolted. Some Coreweave bonds traded over 17%.
Why did the credit guys revolt? The bullish reason for that is we actually want to fund this but we cannot have 3%+ real rates. The bearish reason is they cannot underwrite circular financing deals. I’m not sure what the truth is but even the existence of those financing deals is an indication of not enough capital.
But going on the outcomes of DC ballot measures nationally and what happened in Texas, it also appears we don’t want to fund this. There is precedent for this. We have slowed previous technological breakthroughs due to political or safety concerns in the past - nuclear is the best example.
So as I wrote in Feb in response to Citrini’s fast takeoff fiction that briefly crashed the markets, both financial and political constraints are mitigating the reality of that scenario. The pessimists are the new optimists.
It’s not like dotcom where the majority of investment was capital-light and we were able to overbuild for the future. It’s more like railroads except we never got to overbuild. There is still shortage of compute. So it’s somewhere between that and the modern railroad to nowhere that I just tweeted about yesterday, where they cost too much and aren’t finished. The reality is it is fine to build a railroad to nowhere on FB and Google ad revenue but bond guys don’t want it.
If we are going to fund further infra buildout, we are going to need a money printer. NVDA was doing that for us off-balance sheet but just judging by Collette’s cold corporate tone it seems like that might now be over or at least reduced.
BTW you have to imagine the lion’s share of productivity growth for the next 5 years should not be building more compute but using the existing massive capacity at heavily subsidized rates to really unleash the power of these tools. Because of the decentralized nature of that work it will be hard to find ways to incorporate that into SPX but I have much faith in Wall St to cook it up somehow. In standard and poor we trust.
Bitcoin in my mind is 2 things. A liquidity mop and an abstraction for energy storage. Both are in extreme deficit at the moment and for the foreseeable future so it makes sense to die but the phoenix will eventually rise. Every crypto bear is the same. A year of limp carcass. No need to rush to buy and catch the falling knife but also don’t forget to buy...
Bitcoin in my mind is 2 things. A liquidity mop and an abstraction for energy storage. Both are in extreme deficit at the moment and for the foreseeable future so it makes sense to die but the phoenix will eventually rise. Every crypto bear is the same. A year of limp carcass. No need to rush to buy and catch the falling knife but also don’t forget to buy...
Replying to @jbooger91
I still think there’s a lot of risk in markets until November.
Political risk is high- October surprise potential, war ramp up after (or even before it seems), communists winning too much, etc.
Anthropic vortex around the corner and this neocloud stuff feels terrible. There’s a clear inflection in the bond market which may be telling us something about growth. That being said breadth is horrible and most names have been in vicious bear territory for a while already and the generals have been supporting everything. So McDonald’s up, NVDA down?
I’m not quite outright bearish but don’t think the risk-reward skew is great and would be seller for choice.
2xdog retweeted
Replying to @panabee
if compute never gets overbuilt it will be the first commodity in the history of the world to not have done so.
























