518: Carl Icahn, SpaceX, Google TPUs, Cancer Vaccine, Social Media, Fossil Fuels, Ford Hits Brake on EVs, Longevity 101, and Die Hard

"Progress is not making things more complex"

Humans don’t mind hardship, in fact they thrive on it; what they mind is not feeling necessary.

–Sebastian Junger

Halifax, Nova Scotia 🇨🇦

🏰💥🤔 When I visited Halifax, I spent an afternoon at the Citadel (aka Fort George), a fortification built on a hill that overlooks the city and water. The first structures were erected in 1749 (!), and the current citadel was built in 1828.

I’ve seen plenty of star-shaped forts over the years, and has some ideas about why they are built that way — I assumed there were some defensive advantages compared to what we think of when we think of more medieval castles — but after digging a bit, I found interesting stuff.

It turns out that once again, technological progress in one area (artillery) has forced innovation one another area (fortifications).

The earliest known star fort is Yedikule in Istanbul, built in 1458.

The key elements of a star fort include:

• Low, thick walls: Unlike medieval castles with tall stone walls, star forts had lower, thicker walls that were better able to absorb cannon fire.

• Bastions: These diamond-shaped outcroppings eliminated blind spots and allowed defenders to provide covering fire along the walls.

• Glacis: A sloped earthwork in front of the walls designed to deflect incoming artillery fire.

• Ditches: Wide, deep trenches surrounding the fort, often filled with water to create a moat.

• Ravelins and other outworks: Additional defensive structures placed in front of the main walls to further impede attackers.

This has many benefits over earlier fortification designs:

1. Improved defense against artillery: The angled walls and bastions were more effective at deflecting cannonballs, reducing the damage from bombardment.

2. Elimination of dead zones: The protruding bastions allowed defenders to cover all approaches, eliminating the sheltered areas that existed near round towers.

3. Interlocking fields of fire: The layout enabled defenders to provide mutual support, with each bastion able to fire upon attackers threatening adjacent sections of the fort.

4. Enhanced obstacle to assault: The complex geometry of star forts made it more difficult and time-consuming for attackers to approach the walls, exposing them to defensive fire for longer periods.

5. Flexibility in design: Star forts could be adapted to various terrains and scaled to different sizes, from small outposts to large city fortifications.

6. Improved resistance to mining: The lower profile and thicker walls made it more challenging for attackers to undermine the fortifications.

7. Better use of defensive artillery: The bastions provided ideal platforms for positioning defensive cannons, allowing for more effective counter-battery fire.

Star forts dominated Europe for nearly three centuries, from around 1500 to 1800. They were also exported to other parts of the world.

Here’s a cool bit of trivia for the US: 🇺🇸🎶

Fort McHenry in Baltimore, USA inspired the writing of "The Star-Spangled Banner", which became the national anthem 

🤖👋👋👋👋👋👋 Here’s an interesting example of how the “personality” of various LLMs can be very different. Zack Witten decided to just repeat “Hi” to various AIs to see what happened. Some of them respond in a very robotic way while others… Well, see for yourself.

Unsurprisingly, Claude 3.5 Sonnet, my favorite model right now, has some of the most interesting and human-like replies while ChatGPT and Gemini are the most boring.

---

🏦 💰 Business & Investing 💳 💴

💰💸 Carl Icahn’s Giant Hidden Margin Loans 😧😬

Last week, the SEC charged Carl Icahn and Icahn Enterprises with civil securities fraud for failing to disclose billions of dollars worth of personal loans pledged against his company’s stock.

Icahn and the company will pay $500,000 and $1.5 million in fines, respectively, to settle the charges without admitting to or denying the findings, the SEC said.

I can’t write this one better than Matt Levine — nobody can — so I’ll just give you a few highlights and encourage you to check out the whole thing.

At the end of 2022:

1. Icahn Enterprises LP (usually called IEP) had about 353.6 million shares of stock outstanding.

2. Carl Icahn, the chairman of IEP, owned about 85% of them, about 300 million shares.

3. The stock closed the year at $50.65 per share, giving IEP a market capitalization of about $18 billion, and making Icahn’s stake worth about $15 billion.

4. Icahn had pledged about 181.4 million of his shares — roughly 60% of his stake, and roughly 51% of the total market capitalization — as collateral to secure margin loans with seven different banks.

5. Those pledged shares were worth about $9.2 billion at then-current market prices.

6. Against that $9.2 billion of collateral, the banks loaned Icahn $5 billion of cash.

Ok. What’s the problem?

IEP is, loosely speaking, an investment partnership, owning large stakes in publicly traded companies as well as some private companies. It reports an “indicative net asset value,” that is, its own valuation of the stuff that it actually owns. In December 2022, that reported value was about $5.6 billion, or about $15.96 per share. That is, IEP traded at more than a 200% premium to its net asset value.

Closer to 300% of NAV at certain points…

If you thought of IEP not as “a publicly traded company with a market cap of about $18 billion,” but rather as “an investment fund with about $5.6 billion of assets,” then it is really really weird that the banks would lend Icahn $5 billion against half of that fund.

🤯

Since then, the stock tanked:

IEP closed at $50.42 per share the day before the Hindenburg report, but was at $21.81 a month later; it closed at $15.89 last Friday. One simple model here is that (1) IEP traded at a huge premium to NAV, (2) Hindenburg just pointed that out, (3) everyone was like “huh turns out this stock is worth like $16, not $50” and (4) the price quickly adjusted.

What were the bankers thinking? Doesn’t anyone do diligence? (call me naïve…)

You’d think that when billions are involved and all the information needed is in public filings, they’d at least ask some intern to do the math ¯\_(ツ)_/¯

Meanwhile, the dividend yield on IEP stock (last I checked) was a tad over *25%*. And because the money to pay this isn’t coming out of operating businesses cashflow, Icahn needs to keep issuing more stock or cut it (on August 26, he issued $400m with Jefferies).

I couldn’t find anything about an actual margin call, but looking at how much the stock has dropped, Icahn likely had to do something. Either pay back some of those loans or find new collateral.

How long can this craziness continue?

The evergreen lesson here is: However rich you are, live within your means and don’t get into a huge amount of personal debt. The risk/reward isn’t worth it, especially if you already have more money than you could ever spend.

IEP traded way above NAV for an extended period. But nothing says it can’t also trade way *below* NAV for an extended period too — what if all this had happened in 2022 when general market sentiment was much more negative than now? Icahn could’ve had a *huge* margin call and lost most of his business in a fire sale — potentially creating a vicious cycle since unloading a lot of stock to meet a margin call can feed on itself, driving prices lower, forcing the sale of more stock. Rinse and repeat.

🐦 Every Social Platform Should have ‘Community Notes’ 💡

Community Notes is a crowdsourced fact-checking system that allows users to add context to potentially misleading posts. Contributors can write notes on any tweet, which are then rated by other participants. Notes deemed helpful by a diverse group of users are publicly displayed on the post.

I think it’s one of the rare foundational innovations in social media of the past few years. It’s not that it's perfect, but it's better than not having it, and I wish something equivalent existed on Tiktok, Instagram, Facebook, YouTube, etc.

A few weeks ago Twitter offered me to become a Community Note contributor. I was curious to see how the system worked, so I accepted.

There’s a lot of crap going on in the back, but most of it doesn’t become public, so the system seems to be pretty well designed.

It’s not impossible to game in theory, but by focusing on accounts that have existed for a while and have good engagement, by giving more weight to accounts with a good track record of voting for notes that end up being accepted by the community and voting against notes that end up being rejected, and by making sure every note needs votes from contributors who are in different “twitter circles” (ie. within different communities), you end up with something fairly neutral and useful (some of that DNA no doubt comes from Wikipedia’s NPOV guidelines).

🔥🛢️ Fossil Fuel Consumption 🪨🏭

Some highlights:

• China and the U.S. are responsible for almost half (47%) of global fossil fuel consumption.

• Fossil fuels accounted for over 80% of the global energy mix.

• In 2023, China consumed 140 exajoules of fossil fuels, equivalent to approximately 5.8 billion tonnes of hard coal. The U.S. followed with 76 exajoules, while India was third with 35 exajoules (quite the power law!)

I don’t think most people understand just how difficult it’ll be to replace fossil fuels. The scales involved are not intuitive. Numbers that big become meaningless to most. All this stored energy packed very densely allows our civilization to exist.

If we are to truly make a dent, we need to become a lot more serious about it, and yes, that includes harnessing what is *by far* the densest source of energy, nuclear power.

Coal is the dirtiest source of energy and priority #1 should be coal in China and India. This varies by specific sources, but a ballpark figure is that:

• Coal-fired power plants emit about 2.2 pounds of CO2 per kilowatt-hour (kWh) on average

• Natural gas-fired power plants emit about 0.97 pounds of CO2 per kWh, so even within fossil fuels there’s a huge variance.

Source.

🛑🔌🚘🔋 Ford Hits the Brakes on EV Rollout + China EV Worries 🇨🇳

Speaking of replacing fossil fuels…

Ford Motor is canceling plans for a large electric sport-utility vehicle and expects to take $1.9 billion in related special charges and write-downs [...]

Ford instead will offer a hybrid gas-electric version of a future large, three-row SUV [with] Explorer and Expedition nameplates.

The company also pushed back the launch of a new electric pickup truck by one year, until 2027. In addition, Ford said it would trim its capital spending on fully electric vehicles to about 30% of its budget, from 40%.

If we do the math and look at things purely from an optimization point of view, plug-in hybrids can make a lot of sense.

For a given amount of batteries manufactured, you can often reduce oil consumption more through PHEVs than pure EVs.

In the past, I’ve argued that this view — which has been championed by Toyota among others — is a little shortsighted. The way to get good at EVs is to make a lot of EVs to go up the learning curve and scale up to lower costs.

I guess we’ll see how both approaches stack up against each other because this is what China is doing. As he’s slowing down EV plans, Ford’s CEO is worried about the competition:

Ford Chief Executive Jim Farley has said that China’s EV companies have the advantage of a lower-cost supply chain and that Ford needs to find ways to lower its own costs to compete.

“We believe that the fitness of the Chinese in EVs will eventually wash over our entire industry in all regions,” Farley told analysts last month.

This clash may not happen because of trade barriers, but this could leave Western automakers far behind the Chinese, where the harsh competitive environment will breed all kinds of advances and innovations in EVs (and is already doing so).

---

🧪🔬 Science & Technology 🧬 🔭

🚀 SpaceX Raptor Engine Progression from Gen1 to Gen3

This image has *got* to become a slide in every engineering classroom!

Progress is not making things more complex, but rather as simply as they can be.

What an amazing work by the SpaceX team!

It’s this simplification that has allowed them to do this:

🇺🇸 SpaceX launches by year:

2006: 1
2007: 1
2008: 2
2009: 1
2010: 2
2011: 0
2012: 2
2013: 3
2014: 6
2015: 7
2016: 9
2017: 18
2018: 21
2019: 13
2020: 27
2021: 33
2022: 61
2023: 98
2024: 82 as of August 26

They’re on track to do a launch about every three days! 🤯

🧬👩‍🔬 67-year-old receives world-first lung cancer vaccine 🤘

Yes!

Janusz Racz, a 67-year-old lung cancer patient, is the first to receive this groundbreaking vaccine. He is part of a clinical trial that is taking place across multiple countries.

BioNTech, a German biotechnology firm, has developed this mRNA-based vaccine dubbed BNT116. The vaccine works by activating the immune system, which then recognizes and combats cancer cells. [...]

This experimental cancer immunotherapy is designed for the treatment of non-small cell lung cancer (NSCLC). NSCLC is the most prevalent type of lung cancer.

“Lung cancer remains the leading cause of cancer deaths worldwide, with an estimated 1.8 million deaths in 2020.”

Immunotherapy is extremely promising.

The experimental vaccine is specifically designed to boost immune responses against targets primarily found on cancer cells, thereby minimizing the risk of harm to healthy, non-cancerous cells. This differs from chemotherapy, which often damages both malignant and healthy cells.

Hopefully, someday (soon!) we look back on radiation and chemo as the blunt instruments that they are and can target all cancers directly, leaving other cells unharmed.

I just wish we could accelerate the process. It can take 10-20 years or more for some of these therapies to be fully approved, and that feels too long. There *must* be ways to speed things up while maintaining a good balance of safety and meticulousness.

I suspect that we too often give too much weight to mistakes of commission vs. omission. People who suffer and die while they wait because of a process that is longer than it needs to be are just as harmed and should be considered when making trade-offs.

🗣️ Podcast: Longevity 101 🏋️‍♀️ 🕰️

In the past, I’ve recommended Peter Attia’s book ‘Outlive’.

But if that’s too much of a commitment for you, you can start with this pod to get a decent overview of many of the concepts:

• 🎧 #311 ‒ Longevity 101: a foundational guide to Peter’s frameworks for longevity, and understanding CVD, cancer, neurodegenerative disease, nutrition, exercise, sleep, and more

I still recommend the book. There’s a lot more in there, and it’s packaged in a way that is likely to have more impact and stick in your brain better.

Even if you pick up just one thing from it — change one habit, add one thing to your routine, etc — it can have huge leverage in your life. Health is the foundation of everything else we value in our lives. We forget it at our own peril.

🏎️🤖 Google’s TPU History — A Decade of AI Acceleration

Google published a neat little retrospective overview of their TPU efforts. Some highlights:

Just over a decade ago, a group of Googlers discovered that the company’s AI compute demand was going to outpace our infrastructure at the time. [...]

“We thought we'd maybe build under 10,000 of [TPU v1],” said Andy Swing, principal engineer on our machine learning hardware systems. “We ended up building over 100,000 to support all kinds of great stuff including Ads, Search, speech projects, AlphaGo, and even some self-driving car stuff.”

Today:

TPUs also underpin Google DeepMind’s cutting-edge foundation models, including the newly unveiled Gemini 1.5 Flash, Imagen 3, and Gemma 2, propelling advancements in AI.

Trillium TPUs deliver more than 4.7x improvement in compute performance per chip (compared to the previous generation, TPU v5e)

---

🎨 🎭 The Arts & History 👩‍🎨 🎥

🎄 ‘Die Hard’ Analysis and History with Kevin Smith 🍿

It’s easy to forget this film's impact because it has influenced so many other films since. If you haven’t watched it in a long time (or ever!), consider making it your next pick on popcorn night.

I really enjoyed this podcast about it. It’s all over the place, but Kevin Smith’s encyclopedic knowledge of film history adds a lot of flavor:

• 🎧 Die Hard with Kevin Smith

It’s kind of strange how hard to replicate this film has been. So many have used it as a template, but how many are anywhere near as good? Once in a while, things gel together in a way that is very hard to control or plan for.