This week the New Yorker ran a piece from Ronan Farrow and Andrew Marantz that was 16,000 words long and dedicated to exploring whether Sam Altman is trustworthy. I read the entire thing but can’t in good conscience recommend it to anyone who’s vaguely familiar with Altman. The profile was reported across 18 months and several sitdown interviews with Altman, but there’s nothing in the piece that presents as particularly new.
There are, instead, lots of contemporaneous allegations levied by former OpenAI employees who have since left the company, and breathlessly framed anecdotes like this one:
Most of Altman’s employees at Loopt liked him, but some said that they were struck by his tendency to exaggerate, even about trivial things. One recalled Altman bragging widely that he was a champion Ping-Pong player—“like, Missouri high-school Ping-Pong champ”—and then proving to be one of the worst players in the office. (Altman says that he was probably joking.) As Mark Jacobstein, an older Loopt employee who was asked by investors to act as Altman’s “babysitter,” later told Keach Hagey, for “The Optimist,” a biography of Altman, “There’s a blurring between ‘I think I can maybe accomplish this thing’ and ‘I have already accomplished this thing’ that in its most toxic form leads to Theranos,” Elizabeth Holmes’s fraudulent startup.
Can Sam be trusted? It’s certainly a fair question. The story details a notoriously messy exit from Y Combinator, the acrimony with Dario Amodei and a coterie of effective altruist board members, a falling out with Elon Musk that was documented in a series of lawsuit exhibits, and yes, a hollow ping pong brag that is the foundation of a comparison to Elizabeth Holmes two sentences later. The only thing new there was the ping pong anecdote, as well as some exclusive notes from Amodei and Ilya Sutskever that are entirely consistent with previous reporting about their falling out with Altman.
And again, all of that is fair game for an Altman profile. He’s had so many apparent ethical lapses that it’s impossible to catalog them all; remember when Scarlett Johansson refused to be the voice of ChatGPT and Sam tried to use a clone of her voice anyway? Even with 16,000 words, the New Yorker couldn’t fit that one in.
On the other hand, Altman could just as easily be framed as a very effective startup CEO willing to bend rules to succeed and adopt varying messages depending on his audience. The gist of the drama underlying his ouster from OpenAI in 2023 was that he told one safety-focused story to recruit engineering talent who took massive discounts to work at OpenAI, and then, when the company hit it big with ChatGPT, Altman alienated some of that talent (and the board) when he began telling a different story to investors to cover massive compute costs and allegedly obfuscating, internally, his attempts to capitalize on a once in a generation opportunity in consumer tech (which any sane, ambitious executive would do). Between that narrative and the Amodei “Sam is a SCHEMER who BETRAYED us” perspective, I don’t know which read is more valid. The truth is probably somewhere in the middle, though the New Yorker lands much closer to Dario’s side and the ping pong anecdote, allowing readers to draw their own conclusions but leading them to a clear takeaway.
While criticism is fine and warranted, my bigger problem with the piece is its narrow focus. How many times can you ask the same question of sources and get the same vaguely incriminating answers? The New Yorker‘s answer is “a lot!”, but I wish the writers would have used their considerable time, resources and space to go a few other places.
Rather than uncritically recite the complaints of zealous board members and Amodei, what about a brief detour to explore their motivations, the effective altruist movement, and its centrality in the AI space? That’s a story that hasn’t really been done well yet. Or, rather than taking pains to repeatedly and repetitively frame Sam like an AI Svengali who’s destined to inherit the future, maybe spend a few thousand words exploring the tortured economics of OpenAI, Altman’s admitted failures to focus on core businesses, a flip-flop on ads that was two years too late, mounting threats from Google and Amodei’s Anthropic, flat-lining growth, and any number of other very fair questions about a company and CEO that had a once in a generation opportunity and might, in fact, blow it.
During one of their several interviews, Altman told Farrow and Marantz, “The way that I’ve explained this to my friends is: ‘This was the most fun job in the world until the day we launched ChatGPT.’ … And then, since the launch of ChatGPT, the decisions have gotten very difficult.” The article breezes by that quote, and ignores a good bit of evidence that despite his initial startup success, Sam is not necessarily acing any of the difficult decisions as his company matures.
I’ve written magazine profiles before and understand that print timelines can make it difficult to produce timely work, but on the other hand, come on. None of the OpenAI questions are new, and they are all fairly important considering this company has been helping to carry the stock market for the past three years. And here, I’m bummed out not only by the story but also by the lack of alternatives to produce similarly in-depth reporting that’s more interesting. Tech-fluent media has been balkanized and many outlets are now ideologically captured or actively sponsored by the industry, and in any event they are all too under-resourced to perform this kind of investigative journalism. Meanwhile, the outlets that do have resources to do sprawling and ambitious work tend to approach the biggest tech subjects in the world with predetermined conclusions in mind. That blinkered approach often leaves journalists missing better angles, and leaves readers with a story like this one, which is essentially a very comprehensive Wikipedia entry about things people have said about Sam Altman.
It’s too bad. In the rush to frame Altman as some kind of tech sociopath, which he might well be, the New Yorker largely ignored a better and more urgent question: is Sam any good at his job?
Strait Talk
My wife has a rule for problems. When stressing out and spiraling over a stupid mistake, someone who’s not doing what they should be, or a situation that’s partially beyond your control, stop and ask yourself: realistically, will this still be a problem in two days? Will it matter in two weeks? And what about two years?
The exercise tends to be pretty clarifying, and I’ve been reminded of that wisdom reading about the Iran war over the past month. In particular, there’s been a tremendous amount of angst over Iran’s control of the Strait of Hormuz and what that portends for the future of global stability and the global economy. I’m thinking here of a Senator from Connecticut calling Iran’s continued control of the Strait a “history-changing win,” a New York Times op-ed predicting that controlling the Strait would make Iran a global superpower, or a JPMorgan analysis concluding that Iran under the IRGC could become one of the wealthiest countries in the Middle East, netting $90 billion-per-year thanks to its newly-asserted tolling authority at the Strait.
I will take the under on that bet. Leaving aside the aspect of the IRGC’s plan that presumes international partners will allow them to flagrantly violate international law in perpetuity (the Strait of Hormuz is in international waters), the world’s trade routes are not static. Leveraging control of the Hormuz chokepoint has always been a card that Iran has threatened to play, but it’s also a card that can only be played once. While countries and multinational companies have lived with the threat of Iranian disruption for nearly 50 years, now that it’s happened, those same countries and companies will mobilize to make sure this can never happen again.
Consider this op-ed from Badr Jafar, a special envoy from the UAE writing in the Financial Times this week:
Thirty per cent of global seaborne oil flows and a fifth of the world’s liquefied natural gas trade normally transit a waterway that is 21 nautical miles wide. A third of seaborne traded fertiliser and nearly half the world’s seaborne sulphur exports depend on the same passage, with direct implications for global food security. So do significant volumes of aluminium and helium — the latter essential to semiconductor manufacturing and the global AI supply chain. The concentration of so much global commerce through a single contested corridor is an anomaly the world has tolerated for decades. That tolerance has now ended.
Those investing in post-Hormuz resilience are constructing the trade infrastructure of the future. Saudi Arabia’s Red Sea ports and expanded pipeline capacity offer an alternative energy corridor. The UAE’s east coast provides deep-water ports and pipeline routes connecting Gulf producers to the Indian Ocean. Oman’s developments at Duqm and Sohar sit well outside the chokepoint. Goods and energy are already moving along these routes — in some cases through cross-border land bridge arrangements that would have seemed improbable just months ago.
The multinationals operating in the Gulf see this clearly and are positioning for what comes next. … However the current crisis is resolved, no government will return to a posture of strategic dependence on a narrow strait controlled by an unpredictable neighbour. The pipelines will be expanded. The port capacity will be built. The power grids, water systems and trade corridors connecting the region’s economies will be formalised.
Lest you think that’s overly optimistic spin from an interested party in the region, there’s also Bloomberg’s Javier Blas making essentially the same point:
The Islamic Republic has a new sense of its own strategic power, one to potentially match a nuclear arsenal. … And yet, Iran isn’t alone in learning a crucial geopolitical lesson. Its neighbors have, too, and that will gradually weaken Tehran’s strategic advantage. Saudi Arabia and the United Arab Emirates managed to circumvent the chokepoint to a limited degree via their bypass pipelines. Riyadh and Abu Dhabi are almost certain to double down, expanding those emergency conduits further.
Kuwait would doubtless join forces with the Saudis to build its own bypass pipeline. Iraq would struggle with the expense, but it has every incentive to rebuild its old strategic pipeline that let it move oil from the south to the Mediterranean via Turkey.
Iran’s stranglehold over energy supplies will therefore loosen over time. Five years from now, the Persian Gulf will have far better bypass options than it does today. No matter what the US and Iran agree over the future of Hormuz, the strait’s status will change. But the waterway will never be as critical to the global economy as it was when the fighting started six weeks ago.
The war is ongoing, of course, and short-term volatility seems guaranteed. That downside risk is real, and until six weeks ago that risk had been enough to prevent any countries from challenging Iran’s murderous regime. So far, though, the downside has been managed fairly well by a global economy that is often more adaptive than many analysts account for.
In general, I continue to read voraciously about the war and I’m sure I’ll write more about it over the next few weeks, when (hopefully) there will be a bit more clarity. For now I’m comfortable admitting that I’m not entirely sure where this will go, and I’m a little stunned by the strident confidence that everyone seems to have when discussing the implications of ceasefire agreements that are very sparse on details, reports of Chinese pressure on Iran that are vague and unconfirmed by the Chinese side, or Trump and IRGC messaging that changes every hour.
Likewise, for partisan reasons, most of the mainstream media and many politicians seem determined to frame this war as a Trump-led strategic disaster that will do lasting damage to U.S. interests and the world. Maybe that will be its ultimate legacy, but drawing that conclusion now is quite clearly premature. I think the impulse to sensationalize every new development into a Suez moment for the United States is clouding everyone’s judgment. That’s certainly true with respect to the future of the Strait of Hormuz.
Iran is not the only party with the power to control the global flow of oil and critical materials. A dozen other countries are now incentivized to find better, more secure energy routes, and trillion-dollar market participants tend to find ways around intractable problems. Like China playing its rare earths card for tariff relief last year, Iran’s regime shutting down Hormuz is the sort of play that has short term benefits and clear long term costs. Minds are focused now, and markets will mobilize. The war forced Iran’s hand, and leverage the IRGC wielded successfully for the past 50 years is now, at most, a five-year problem.
A Word About E-ZPass and Mount Rushmore
On last week’s Sharp Tech podcast we spent a few minutes putting together a tech company Mount Rushmore, and my list began with Apple, AT&T/Bell Labs, and Google. Three unimpeachable choices, if I do say so myself. Then I got to the fourth, where I chose to zag a bit and went with E-ZPass.
Now, people did not like that nomination. Ben was horrified, and much of the audience was, too. Some e-mailers complained that the E-ZPass idea was normie trolling run amok, others sympathized with Ben for having to deal with me at all, and one guy cited a podcast episode nine months ago where I identified as a Democrat and compared me to Jason Calacanis and Kara Swisher—a very tough beat for me.
I’m not here to fight any of the allegations (except maybe from that last guy). The take was a mess. For one thing, the prompt was to create a Mount Rushmore for tech companies, and E-ZPass is managed by E-ZPass Interagency Group, a collective of dozens of state and local tolling agencies. It’s a multi-government effort, not a private company, and therefore ineligible for inclusion on any tech company Rushmore. Additionally, while E-ZPass adoption has expanded to 17 states since its inception and has saved millions of drivers millions of hours at toll booths up and down the East coast and throughout the Midwest, I accept the objection that the appeal of faster toll lanes may be somewhat limited, and too limited for a national monument celebrating technology. Finally, someone complained that putting E-ZPass on tech company Rushmore was like injecting Eric Snow into a conversation about the greatest NBA players of all time… Which is, again, fair enough!
Hosting podcasts basically every day, I’ve accepted that living with occasionally horrific and embarrassing lapses of judgment, recorded for eternity, is part of the deal. Another thing I’ve learned: Anytime you find yourself repeating a take to someone in the real world and you realize that in order for it to make even a semblance of sense it requires several explanatory caveats, that’s probably a bad sign.
Having said all that, I won’t turn this post into a Dril tweet where I further explain my instincts and my problems with putting Microsoft on a mountain (the correct answer, regardless of its replacement-level products) or in the alternative nominating Meta, Amazon, or Sun Microsystems. Instead, I’m here for history.
Specifically, E-ZPass history. The humble white transponder, the Eric Snow of tech products, has saved millions of people millions of hours since its arrival in the 1990s, and here are two articles that I discovered and loved earlier this week:
- WNYC: From Soviet Russia to the Tappan Zee Bridge, A Short History of the E-ZPass (2016)
- Tedium: Counting all cars (2018)
Rather than excerpt huge portions of both, I’ll recommend each one as follow-up reading for anyone who’s curious about the E-ZPass origins. And here are some bullet points, simply because I find the history interesting (and wasn’t aware of it until after recording last week’s podcast):
- The story begins with Leon Theremin, who invented the theremin, a touchless musical instrument, and then spent time in America working with RCA while actively spying for the Soviets in the 1920s and 30s. He eventually returned to Russia, and, after being falsely accused of spying for the American side, was sentenced to work in a prison camp for Soviet scientists, where he employed an early version of radio-frequency identification (RFID) to invent a bug that could be activated remotely, by radio waves, to transmit audio, and didn’t need a battery (it used the energy of the incoming signal to broadcast back). Theremin’s bug was then hidden in a wooden Presidential seal that had been carved by Soviet schoolchildren and was given to the U.S. Ambassador to Russia. It went undetected in the U.S. Embassy for seven years.
- E-ZPass technology uses similar RFID technology today, and also builds on 1930s radar technology that used transponder technology to send a signature ID to identify friendly aircraft (Identification Friend or Foe, or IFF, systems). What that means when you drive through a toll plaza: the master transponder emits a signal (using RFID) that wakes your passive E-ZPass, which responds with a signature ID (your car, using IFF), and the toll bills the associated account.
- The first person to patent this technology was an engineer named Mario Cardullo, who’d previously worked at Bell Labs and designed rockets for the Navy and NASA. In 1969, Cardullo sat next to an IBM engineer on a flight, and, whipping out his notebook and sketching a solution on the spot, helped the engineer work through an RFID solution to track train cars. From there, Cardullo invented an RFID product that could track automobiles, and secured his patent in 1973. He then pitched his technology to the New York Port Authority and hoped to demonstrate how it could fix tolling on the George Washington Bridge; state authorities demurred at the time, though according to Cardullo, they did shop the idea to other firms like Westinghouse, Phillips, and GE to see whether they could build similar prototypes.
- Cardullo’s patent expired in 1990 and he was never rewarded for his work. Around the same time, though, Larry Yermack was the CFO at the Triborough Bridge and Tunnel Authority and realized that instead of processing 300 vehicles per hour in a cash tollbooth lane, electronic tolling could process 1,000 cars per hour. He also realized that any transponder used in cars would be far more useful and likely to be adopted if it worked across state lines (“What if we had tags that cooperated with the other toll authorities, in fact, what if we had the same tags?”). In March 1994, after a nearly 20-hour meeting during a snowstorm, seven toll authorities from New York, New Jersey and Pennsylvania agreed to use the same device. E-ZPass arrived in 1995.
- Privacy concerns surrounding the use of E-ZPass data, which Ben mentioned on Sharp Tech, are a real thing. E-ZPass transponders have been used to monitor traffic patterns even where there’s no tolls—without consent and without informing drivers—and could theoretically be abused in other ways by government agencies. This practice was exposed in 2013 by a hacker named Puking Monkey, who posted YouTube videos documenting his E-ZPass being accessed in Times Square (miles from any toll plaza).
Surveillance concerns notwithstanding (and not particularly scandalous to me since almost all consumer hardware presents similar concerns), I do think E-ZPass is the platonic ideal of what tech can be: someone recognizes a problem, repurposes existing technology to solve it, others think about how people will use it and realize it will be even more useful if it interoperates across multiple systems (or states, in this case), and it all yields a seamless hardware and software experience that makes life more convenient for millions of people (and all the states that manage traffic). Like most good tech stories, this one spans decades of innovation and iteration, has tragic figures (Theremin), forgotten heroes (Cardullo), and even a hacker named Puking Monkey for good measure. And Cardullo, the inventor who never got paid, seems OK with how it all worked out. “You don’t invent things to make money,” he says. “You do it to solve problems.”
Now, does all of that context justify putting E-ZPass on a Rushmore monument celebrating the most significant companies in the history of technology? Most people would emphatically say NO, and if you’re one of those people, don’t worry, there’s no need to e-mail! On the other hand, if you’re like me and have lots of mixed feelings about tech’s impact on modern life, it’s nice to get the occasional reminder that there are plenty of cases in which well-executed technology is a clear net positive for everyone.
In that case, in lieu of a transponder chiseled into the side of a mountain that road-tripping Americans can enjoy for generations to come—which, again, we are NOT doing—I would at least strongly encourage going out to your car and looking at the dashboard.
Sharp Text is extension of the Stratechery Plus podcasts Sharp Tech, Greatest of All Talk, and Sharp China. We’ll publish once a week, on Fridays. To subscribe and receive weekly posts via email, click here.