In this issue:
- What Does it Mean to Have a Trillion Dollars?—In one sense, the whole purpose of money is that if you denominate wealth in anything else, it starts to get messy as the numbers get bigger. For Elon Musk, the critics are right that wealth means power, but they need to take seriously the argument that this power is more dangerous than other kinds.
- Anthropic—It's a communications problem.
- Mean and Variance—If your workforce experiences less variance than what's implied by changes in their potential output, you're missing something.
- Fragmentation—As it turns out, global media platforms don't create a monoculture, after all.
- Guidance—Elon Musk is involved in two separate discussions about a trillion dollars.
- Slightly-Heterodox Center-X—How long can you last as the opposition to the opposition?
Talk to this post on Read.Haus.
What Does it Mean to Have a Trillion Dollars?
Years ago, I remember reading various headlines about how Vladimir Putin has a $40bn fortune. And I remember thinking: what is that supposed to mean? Is there any time when Putin needs to reach into his wallet, pull out a credit card, and pay for something? Is there anything in Russia that Putin can't have, if he wants it? And, by the same token: suppose he retires to some dacha outside Moscow in order to enjoy his $40bn fortune. How long (in minutes) does this last before someone is executing a wrench attack to retrieve his bank account credentials and wire the money somewhere else? Furthermore, how do we think about his liabilities? Even absolute monarchs are really the head of a coalition, just one that's fragmented enough that everybody wants to support the winning team. Still, his subordinates need to get something out of the bargain other than access to one of the tiny number of dwellings in Moscow whose windows don't have a tendency to give way when leaned against by dissidents, rival power brokers, etc. They need, and get, some spoils, and those spoils are an economic liability that should be factored into the Putin balance sheet. He can shift consumption around, but he still faces binding constraints, and measuring an autocrat's wealth purely in dollars is like measuring the wealth of a modern member of the middle class in terms of how many years' worth of rice and flour you could buy. It's an accurate measure of what isn't the relevant constraint.
Anyway, as of a few weeks ago Wikipedia could justify the existence of an article about the term "trillionaire," and today that article exists and has an example. (It also has a pedantic-but-necessary clarification that there have been plenty of “trillionaires” in places like Zimbabwe.)
It was inevitable that this would happen. 2% real GDP growth, 3% inflation, a 5% or so equity risk premium, some dispersion among equities, and—critically—being a species with five rather than six or seven fingers on each hand meant that some time this decade or next we'd cross the threshold where a single person is, on a mark-to-market basis, worth a trillion dollars. There's been a lot of discussion about this, most of which treats the threshold as meaningful. And, to be fair, there are plenty of people making arguments that are just as relevant at $990bn or $1.1tr or whatever. On the other hand, it's suspicious that the time to make those arguments is when one person's wealth passed an arbitrary threshold that's exciting to people who don't think about things so rigorously. I'm very suspicious of politicians who want some kind of reckoning about wealth inequality that coincides with a huge drop in how informed the average participant in the debate is. Personally, I'd rather the debate happen mostly among people who know what they're talking about. But I've never been any good at politics.
Money is very intuitive in small amounts and gets less so with larger ones. The simplest interpretation is that it's an input into a Boolean function; you see something you want, and either you can't afford it (upsetting!) or you can afford it and thus must instantly buy it. My younger kids, and many elected officials, follow this model. A slightly more sophisticated model embraces the possibility of saving, but still implies that money is for future consumption—you're saving, maybe for an unusually expensive toy, maybe for the downpayment on a house—with the end goal of converting that saving into consumption. This model is about as far as the average person's abstract understanding of money really needs to go: you don't earn it at exactly the same time that you need it, so you'll sometimes borrow some, sometimes save sum, and, over your lifetime, end up spending roughly as much as you earn.
This model of money works for sums between about $1 and a few million. But even if $1 buys one candy bar, $1 trillion does not buy a trillion candy bars. In Elon Musk's case, a trillion dollars buys him about the same thing ten or a hundred billion did: de facto control over SpaceX and Tesla, and the ability to invest in other projects that strike his fancy (mostly to create things money can’t presently buy). In a money-as-power model, Musk has to get richer roughly in proportion to how powerful his companies get, where "power" implies the ability to sell products other companies can't, or to sell the same products at prices they can't match. This money does, of course, give him an ability to consume, but the super-rich have an extremely low marginal propensity to consume, because the world just doesn't produce enough high-ticket consumer goods. And when they do engage in big one-shot consumption, it's typically bidding for scarce assets (houses, art, sports teams, other rivalrous, excludable goods) that are owned by other rich people. If a very rich person spends $100m on a painting, and they could have used that money to feed the sick, you could complain to them. On the other hand, there is someone on the other side of that transaction who has $100m-minus-the-auction-fee, so they're probably the right person to complain to.[1]
Musk's trillion dollars is, for the most part, not competing with your dollars for the consumption of goods and services. He's competing a tiny bit with me, by buying a couple houses in Austin ($, WSJ), but lots of other people are doing that, too; Musk is maybe three successful dermatologists' worth of suburban Austin housing price appreciation. Musk is shifting some of America's total output from consumption to investment, and within investment he's moving more of it towards space travel, AI datacenters, and electric vehicles. So there is an effect, but the trillion dollars represents the market's estimate of Musk's share of the upside from that investment, i.e. it's a guess about how much more he'll be able to sell in the future. We certainly could tamp that down and shift more of this output towards immediate consumption—a 5% wealth tax, for example, could allow us to increase Federal spending by 0.67% a year, and could accomplish whatever we couldn't quite get done with the first $7.4tr.
Which actually raises an important point, because, as Musk's critics point out, Elon Musk is already partly in the business of executing on the government's priorities. They usually phrase it slightly differently, by saying that he made his money from tax credits and government contracts, and there is plenty of truth to this; we wouldn't have had EVs at the time that we did (and would probably have Chinese EVs) without those early tax credits, and there is not (yet) much demand for private space travel compared to the government-supported kind. Even though both Tesla and SpaceX have since demonstrated that they can build standalone businesses, though it is of course very helpful for Tesla that the US effectively excludes Chinese EVs from the domestic market, and SpaceX benefits from operating under the US regulatory umbrella. Musk would be a lot poorer without these benefits. On the other hand, the point of government subsidies for electric vehicles was to make electric vehicles more common, something Tesla played an enormous role in. And the space program predates SpaceX, of course, though it had a multi-decade period where the cost of getting mass into orbit infamously stagnated at around the time the space shuttle was introduced, and then started to decline again once SpaceX got involved. If you're looking for a trillion dollars of government largesse, add up NASA's inflation-adjusted budget over that period and you get to about $840bn. Obviously that money wasn't all wasted, but equally obviously it could have been spent a bit better. It's paradoxical, to put it politely, for people to simultaneously argue that Elon should pay higher taxes because the government will spend his money better than he does, but also that the only reason he has all that money in the first place is because it's incredibly easy to trick the government into wasting money.
It's important to note that one reason Elon Musk is the first trillionaire is that, relative to all the other deca- and centi-billionaires, he has a much higher tolerance for the possibility of ending up broke, in jail, or on the hook for a trillion and one dollars' worth of child support. Some of this is probably genuine skill at underwriting seemingly-low probabilities, some of it is raw risk tolerance, and as anyone who's followed Musk on Twitter knows, some of it is poor impulse control.[2] Musk is probably not solely committed to maximizing his net worth, or maximizing the odds that he hits some threshold, but it seems to be either part of the motivation or impossible to separate from his actual goals. And part of what's fueled his success is that other people care a lot about their relative standing on the net worth leaderboard, and some of them are probably backing Musk specifically as a hedge—over the last few decades, being underweight Musk has been expensive, though plenty of people have managed to make good money despite it. There's a whole weird constellation at work here.
You can still have a debate over whether a trillion dollars worth of power is too much for one person to have. Unfortunately, the power to expropriate the seventh or tenth most valuable public company in the world probably makes someone more powerful than the trillionaire who runs those companies. Granted, there are checks and balances—but Musk has had his ambitions checked by the market, either for cars or for shares of his companies, in the past. Whereas if it's a national political issue, it's ultimately decided by a small number of low-engagement voters in swing states who, every four years, may or may not get excited by issues that may or may not matter and who then decide who's in charge for a while. No matter how you structure things, shifting power from a market system to the political system means choosing between elites, and it's a question of how informed and aligned these elites are. Meanwhile, as strong as never-bet-against-Elon is as a heuristic, it's really hard to get SpaceX estimates to pencil out to anything that comes close to justifying the current valuation. So maybe in a year or two, worrying about a trillionaire will be passé. Or, if we still are, it will be because Elon has somehow pulled off what the market is pricing in.
There is one way that very rich people engage in huge bursts of the kind of consumption that makes other people worse-off, by bidding for scarce resources and using them in a socially-suboptimal way. It's when they get divorced and their spouse starts making aggressive charitable donations. For whatever reason, this particular kind of charity is just extremely unlikely to look like the output of some process where someone produced a giant spreadsheet and then sorted it by quality-adjusted life years produced per dollar spent. ↩︎
Some might even say his act of purchasing Twitter itself was an impulsive act, given his offer had no diligence contigency(https://corpgov.law.harvard.edu/2022/07/14/twitter-vs-musk-the-complaint/). Others would put this in the genuine skill at underwriting seemingly-low probabilities bucket. ↩︎
You're on the free list for The Diff. Last week, paying subscribers read about why we shouldn't be surprised that the average usage of new apps is declining ($), and how cheaper software development changes the shape of viral growth; thoughts on what indices are for ($) and why owning one is a surprisingly good business; and a profile of Bending Spoons ($). Upgrade today for full access.
Diff Jobs
Companies in the Diff network are actively looking for talent. See a sampling of current open roles below:
- Ex-Palantir, Citadel founders building the meta-harness (the system that knows what hills to climb, and what the right loss functions are) for all the lucrative agents need full-stack engineers that understand that turning AI into economically valuable solutions means a system that includes deterministic infrastructure. If you’re curious about building a system that finds what the efficient frontier between determinism and stochasticity actually is, this is for you. (NYC)
- Series A, ex-Navy defense technology firm building AI-enabled drone defense systems is looking for an electrical engineer with range: schematic design, electrical simulation, printed circuit board (PCB) design, and firmware development in high performance languages (C++, Rust, etc.) If you’re interested in power and control systems for mission critical technology, this is for you. (Austin, TX)
- A startup building a new financial market within a multi-trillion dollar asset class is looking for a data scientist with commercial financial experience. (if you’ve been an investor but are newer to the data side, that’s great too.) (NYC)
- Lightspeed-backed team building the engineering services firm of the future is looking for founding members of technical staff excited about working alongside civil engineers to translate their domain expertise into the operating system that powers the next era of great American infrastructure. If you’re an engineer with strong product intuition, who's energized by access to users, and excited by the prospect of transforming how we design and construct our built world with frontier AI, this is for you. (NYC, SF or Remote)
- AI Transformation firm with an ambition to build an economic world model to run swathes of the private, unstructured economy is looking for FDEs, Platform Engineers, and business generalists who understand how to solve problems.
- Well-funded, frontier AI neolab working on video pretraining and computer action models as the path to general intelligence is looking for researchers who are excited about creating machines that learn from experience, not text. Ideally you have zero-to-one pre-training experience and/or are a high-slope generalist who’s frustrated that the big labs aren't doing this. (SF)
Even if you don't see an exact match for your skills and interests right now, we're happy to talk early so we can let you know if a good opportunity comes up.
If you’re at a company that's looking for talent, we should talk! Diff Jobs works with companies across fintech, hard tech, consumer software, enterprise software, and other areas—any company where finding unusually effective people is a top priority.
And: we're now actively deploying capital into early-stage companies through Anomaly. Our focus is on defense, logistics, robotics, and energy. If you'd like to chat, please reach out.
Elsewhere
Anthropic
Anthropic has, once again, been told that its models and the way it wants them used are not compatible with the US government's desires. In March, the concern was that Anthropic was limiting the government's access to its models; this time, it's that Anthropic is giving people access to Fable, a more constrained version of Mythos. The apparent cause is that when Amazon was asked for feedback on the model, they warned that it was possible to get around some of its restrictions.
It would be very bad if Anthropic had built a model that could reliably put together zero-days, and then unleashed it on the world before everyone had had a chance to patch their software. This would also be extremely out-of-character for Anthropic, which has been cautious to a fault (just a few days ago, Anthropic's own employees were suggesting that people who'd previously asked biology-related questions ought to try asking in incognito mode—it was pretty locked down!). What seems more likely is that there was a game of telephone about what Amazon found, and how generalizable it was, and that Anthropic and the White House already distrust one another. Ironically, this stems from Anthropic being a little too consistently candid about what it sees as the risks of powerful AI, which has put their competitors in a relatively better negotiating position.
Disclosure: long AMZN.
Mean and Variance
Soon after telling employees to use more AI, Meta is telling them to cut back, and building dashboards that will better track their spending ($, The Information). Meanwhile, employees of their AI group are tuning into internal livestreams to deliver expletive-filled rants about how much they hate their job. So: it's all going according to plan. The last year has been the biggest shift in both the potential output of software engineers and the variance in that output; there are people who were bottlenecked by typing, or autocompleting, and who can now produce code much closer to how fast they can think. And there are people who are wasting six figures on unmaintainable messes, putting their API keys in public Github repos, etc. So, any company that's in the business of shipping software and that doesn't feel incredibly chaotic right now is underreacting; the world they're in just got more chaotic, and somebody out there is adapting. In an ideal scenario, Meta would smoothly accommodate the fact that some of their people are worth more than ever, and some of them have skills that have mostly been subsumed by models. But in the real world, if there isn't a fair amount of upset and shifting priorities, someone isn't paying attention.
Disclosure: long META.
Fragmentation
One theory of media platforms is that when they crush the cost of discovering new media, they make the biggest global winners even bigger; popularity begets popularity, and some countries have a big enough head start. We'd get a flatter, more simplistic global culture—more movies centered on explosions than dialog, simpler melodies and repetitive lyrics. But, oddly enough, local artists are increasingly dominating the most streamed list ($, Economist). It's probably still valuable for a streaming service to have global (i.e. American) hits as an option. There are two reasons for these platforms to have globalized their home markets' taste early on before being dominated by locals:
- If there's a fixed price for recording equipment, for music and movies, then the relative cost of experimentation is low in rich countries, until the equipment gets cheap enough that it can happen anywhere.
- On a related note, the cohort that starts paying for media (especially with services that assume the user has a credit card) will tend to be the relatively affluent, English-speaking part of a country. But once they make Spotify or Netflix viable locally, the next incremental user will be closer to a typical media consumer.
It's nice to know that as the production and distribution of media both globalize and get more efficient, the result is that the global monoculture struggles to compete with local culture.
Guidance
As you may have heard, SpaceX went public last week, and is currently valued at $2.25tr. Like a typical IPO, SpaceX's prospectus is backwards-looking in terms of quantities, and to the extent that it included forward guidance, that guidance pertained to the size of the market, not to how much revenue they'd earn or when. There is a good reason for this: the legal protections management teams get for forward-looking statements don't apply as strongly to statements made in connection with an IPO; the measure of how abig deal this is is that SPACs. which made forward-looking statements in connection with what was technically a merger rather than an IPO, were so big a few years ago. Anyway, over the weekend Elon Musk said he thinks SpaceX "might" be able to reach "approximately" $1tr in revenue in 2030, and then added that he "would be surprised if revenue is not greater than $1T in 2031." Now, if you ask a truth-seeking AI such as Grok to interpret "I would be surprised if not-X" as the speaker's implied probability that X is true. Grok said that 90% is a good point estimate. The first tweet is a little bit casual, but the second is Musk stating that in his opinion, the most likely outcome in a specific year is revenue above a specific number. This is a guess, but there is a term for when a member of a public company's management team makes a guess about the value and timing of some performance metric: it's guidance. So:
- No responsible investor who has or plans to have a position in a company controlled by Musk is ignoring his Twitter and patiently waiting for an 8-K. In the S-1, SpaceX said "We may use our website www.spacex.com/ or our X account to make information publicly available for purposes of Regulation FD from time to time." (Bold added, italics in original.) This appears to refer to the SpaceX X account, not to Elon's, but of course informed people know.
- The issue is not really that Elon is disseminating investor updates through social media. At least this time the update was released outside of regular trading hours, and wasn't a fictional going-private transaction—it's progress! The problem is that it gets a lot harder to defend Musk's (very real) accomplishments in engineering and business if he's flouting the rules. He's pulling up the ladder to trillionairedom.
Slightly-Heterodox Center-X
The WSJ has a profile of The Bulwark, a conservative-sympathetic and anti-Trump opinion outlet ($). It's interesting to see the media ecosystem create this kind of niche, which feels more supply- than demand-driven. In regular news media, readers tend to like fairly partisan sources, and that tends to mean they like sources that are loyal to whoever is running the party. But within a given publication, there will always be writers who feel that political exigencies are trumping principles, and those writers will tend to attract a stronger fan-base—it's a selection effect, where if you can't stop praising Trump, your fans are Trump fans first and you-fans second, but if you're criticizing Trump while staying on the right, your fans are yours, and they're more likely to follow if you leave a bigger partisan publication to join or start a more narrow one. It's hard to make this a durable business; The Bulwark is defining itself in opposition to someone who was generally more comfortable being the critical opposition than actually governing. That's a very circumstantial reason to exist.