210: The Nuclear Brothers Edition (Fission & Fusion), OpenAI GPT-4, Neglected Superpower, Market-Timing, AWS Graviton 3 ARM CPUs, and Zelda's Open World

"Brace yourselves…"

Cowardice causes blindness.

—Oswald Spengler

🤖 GPT-3, the deep learning autoregressive language model with 175 billion (!) parameters, was announced on June 11, 2020. That’s 538 days ago (not that I’m counting…).

I’m starting to feel like GPT-4 must be coming soon, and remembering how the whole internet could talk about nothing else for a couple weeks when #3 came out, I can’t imagine what the sequel will be like if the jump forward is anything like last time.

Brace yourselves…

🎥 🎬 🍿Whenever someone tells you that they liked or disliked a film or TV show, ask them to put a number out of 10 on it (and it can’t be 7), and ask about their expectations before seeing it, and whether they usually like this genre/director/franchise/whatever.

I think it makes the opinion a lot more useful, and helps calibrate your own expectations.

🤔 I wonder why some stuff sounds so much better than other stuff. Some images and words are just powerful, and some combinations are hard to beat, and describing the same thing two different ways will yield different results.

"Sun-dried tomatoes" for example. If oven-dried tomatoes tasted better, it would still sound much worse and be a harder sell.

It reminds me of the Mad Men episode where they’re making an ad for a cigarette company, and they go with the slogan "It's toasted". The guy says they're all like that — it doesn't matter, it sounds good and if others follow you, they just look like they're copying you.

📰 When I read feature pieces by “real” journalists on someone or something I care about, I usually can’t help but think of just how much BS I have to wade through to get to the good parts of the article.

They’ll do an intro about what the journalist was doing before he got the assignment, the weather that day, something about an eccentric person they interviewed during background research, then some pointless stuff about how the person is dressed at the interview, and some editorializing designed to subtly bring down the person, like, yeah, they’re smart and created some amazing company/tech/science breakthrough, but they’re really just a flawed human no different from the rest of us.

So much filler, the core of the piece is probably 3 paragraphs, and if it’s about a person or field you know well, it’ll probably be fairly shallow, and some of it will be wrong.

Painful 😣

I try to take advantage of this newsletter format as much as possible: I can use my own voice, and I don’t feel bound by conventions when they don’t apply.

To be clear: I don’t intend this as part of the chorus of anti-press sentiment so popular these days. It’s a very hard job, because you mostly don’t have time to really research something and become an expert on it, you have to move on to the next thing. And like everything, there’s a wide distribution in skill — there’s a lot of bad programmers, doctors, entrepreneurs, and investors out there too — so sometimes you’re just unlucky, and a very fascinating person/topic gets handled by a hack ¯\_(ツ)_/¯

🧬 Random stuff I saw while visiting family that kind of looks like DNA:

💚 🥃 If you’re getting value from this newsletter, it would mean the world to me if you became a paid supporter.

If you think you’re not making a difference by subbing, that’s incorrect. Only 4.7% of readers are supporters (so far — you can help change that), so each one of you joining this elite group (⚔️) makes a big difference.

I like free stuff as much as the next person, but when I like something, I also want it to continue and be sustainable:

---

Investing & Business

Mostly Borrowed Ideas @borrowed_ideas

It doesn't fit in a tweet, so I'm doing this.

The United States’ Under-Utilized Superpower

Good observation by my friend MBI (💎 🐕).

Here’s the text in the screenshot, to make it easier to read (the embed above is cutting off the beginning):

I have routinely joked with my friends that the rest of the world should be grateful that America seems to be not very enthusiastic to utilize its true super power which is the ability to attract the very best and the brightest from all over the world.

It is bewildering why America still keeps its skilled immigration system SO onerous and chooses not to utilize its true super power. Despite all these, we still get to see how much some of the best and the brightest want to move to the US, defy its complex immigration system, and then run some of the most important companies that will shape the US AND the world.

🎯

If you’re able to attract a lot of the best talent from around the world, this is an increeeeedible advantage — the U.S. seems hell-bent on minimizing it.

Every time I hear about bright and motivated people who uproot themselves and go study in some of the US’ top schools (MIT, Stanford, etc), and then would like to stay and contribute, but can’t and are driven out… such a missed opportunity.

🇨🇦 Come to Canada, I guess ¯\_(ツ)_/¯

While on the topic of immigration, allow me to repost some of my thoughts on pride and immigration from edition #62 (because I figure many of you probably weren’t around back in the Mesoproterozoic era):

Here’s my thoughts on pride:

I don’t feel like I can be proud of things that I have purely because of a roll of the dice. For example, I don’t feel proud of speaking French, because I don’t even have memory of learning it.

But I’m proud of my English skills, because I learned the language relatively late (between 15-18yo, mostly), and it didn’t come easy, as I have no particular aptitude for languages.

To me, learning it was a means to an end — I cared about music, computers, video games, and wanted to watch un-dubbed TV shows. I didn’t learn much from school, it was too boring. It was The Simpsons, Frasier, Futurama, and Third Rock from the Sun with the subtitles on (to see what I was hearing).

In the same way, I don’t feel what I’d describe as pride about being a Canadian or Québecois. I can love many things about where I live, I can want to make the place better, I can appreciate many things about the culture, etc. But to me, it doesn’t fall under “pride”. I was just randomly born here, I could’ve been born an Afghan woman. I had nothing to do with any of it, pure chance. Should I be proud of having brown hair? What’s the point?

I’ve always felt like immigrants have a rational claim to pride about where they live, because they chose it, made the decision to go, worked for it, and it probably required sacrifice (leaving behind friends, family, and culture).

So if you’ve decided to move across the globe to become an American or a Canadian or whatever, and you succeeded, I think you can be proud of that, it’s a real accomplishment. Kudos to you! I know it would be a big deal to me and my family if we had to move to China.

Side note: Another context where the word “pride” is used is gay pride, LGBT pride, etc. I have no problem with that, because I see it as meaning something different than what I’m talking about above.

The way I grok it, it’s a positive word that is mostly used to mean the absence of the negative opposite (ie. “we’re not ashamed”), which was the dominant societal pressure for so long (ie. “you should be ashamed, you should hide” “No we won’t hide, and there’s nothing to be ashamed of.”). So the very words “gay pride” may sound like “white pride” or “heterosexual pride”, but the underlying meaning is different.

Man, people who joined this newsletter for stock tips must be tripping balls with these intros…

Drowning in a river that is 4-foot-deep (on average)

Modest Proposal:

44 companies in the JPM Internet coverage. 7 outperforming the S&P.

Median stock down 13% on the year.

Mega cap - Mid Cap spread is 56%.

What’s Perfect Market-Timing Worth?

Eugene Ng @EugeneNg_VCap

Peter Lynch in 1997 on market timing: The problem is getting the timing right. Few do so consistently. Timing & missing the best months is far worst as returns deteriorate significantly. Stay in, not out. Timing to get the best entry price is not significantly life-changing.

Obviously, these results would be different for single stocks, because they have much higher volatility than indices, but before you think that justifies market-timing, just remember that this cuts both ways…

h/t friend-of-the-show and supporter MBI (💎🐕)

Chaos Fiends Rejoice!

Ryan Petersen @typesfast

If you're better at dealing with chaos than your competitors you not only benefit from chaos, but you benefit from creating the chaos. A recipe for fun in business.

Sam Altman @sama

The world is going to move faster and faster from here on. People and institutions who embrace this will be richly rewarded. People and institutions who do not or cannot are going to be challenged. Adaptability and speed have gone from valuable to critical.

---

Science & Technology

☢️ ‘The Energy of Tomorrow: The Promise, Failure, and Possible Rebirth of Nuclear Power’ ☢️

Jason Crawford, the creator of Roots of Progress (now a 501(c)(3) non-profit), gave a great in-depth talk about nuclear power.

In the 1950s, nuclear power was seen as the energy of the future. Today, it is stagnating on the sidelines, providing only 10% of world electricity, with no fundamental advance in reactor design for several decades. Why did this technology seem so incredibly promising, how did it go so badly wrong, and is there hope for a nuclear renaissance? This talk will demystify nuclear power, explaining how it works and why it deserves development instead of neglect.

I really recommend watching the video above. Lots of interesting stuff you likely won’t just casually learn elsewhere.

The part about what makes nuclear so expensive is particularly good (starting around 30 minutes into the video).

Back in edition #191, I wrote about the ALARA standard, which I think few people know about and understand the massive second-order effects of.

This slide showing the cost curve/learning curve turning around is really telling:

(I know it’s a long video at 1h48, but it’s worth watching, and the presentation is 1h and the rest is Q&A. I watched it at 2x (click the little gear icon on Youtube) and it sounded fine to me, so that was less than 1h for a very high density of good information)

⚡️ Helion Fusion Startup Raises $500m ⚡️

Speaking of nuclear power (the other kind, the vid above was about fission), Sam Altman, best known for running YC for a few years and being the CEO of OpenAI, has made high biggest investment ever ($375m) in Helion, a startup working on fusion.

This round of financing is $500m with $1.7 billion of commitments tied to specific milestones.

They have an interesting approach, not going with the ‘donut’ tokamak model that ITER is using:

Our approach does three major things differently from other fusion approaches:

1) We utilize a pulsed non-ignition fusion system. This helps us overcome the hardest physics challenges, keeps our fusion device smaller than other approaches, and allows us to adjust the power output based on need.

2) Our system is built to directly recover electricity. Just like regenerative braking in an electric car, our system is built to recover all unused and new electromagnetic energy efficiently. Other fusion systems heat water to create steam to turn a turbine which loses a lot of energy in the process.

3) We use deuterium and helium-3 (D-³He) as fuel. Helium-3 is a cleaner, higher octane fuel. This helps keep our system small and efficient.

This isn’t just theoretical, they’ve been running prototypes:

In 2020, we completed our 6th prototype, Trenta. Trenta runs nearly every day doing fusion. It has completed almost 10,000 high-power pulses and operated under vacuum for 16 months. With Trenta, Helion became the first private organization to reach plasma temperatures of 100 million degrees Celsius (9 keV). Trenta is still operating today.

Additionally, we have demonstrated that our magnets run at 95% energy efficiency, exhibited compression fields greater than 10 Tesla, and sustained plasmas with lifetimes greater than 1 ms.

The 7th prototype is already planned:

Helion's 7th fusion prototype, Polaris, will be primarily built to produce helium-3 through deuterium-deuterium fusion. Helium-3 production at lower cost is possible because Polaris directly recovers electricity leftover from the input as well as new fusion reactions, eliminating the high costs usually associated with helium-3 production. Alongside helium-3, Polaris is also expected to generate a small amount of net electricity as a byproduct of its fusion reactions.

When will Helion produce electricity?

We expect that Polaris will be able to demonstrate the production a small amount of net electricity by 2024, but its main focus will be to produce helium-3. We will continue to iterate our device quickly so we can offer commercial fusion power for the grid as soon as possible.

On safety:

Helion’s non-ignition fusion approach means that there is no chain reaction, and the machine can be shut off instantly.

Helion’s fusion generators would produce radiation while they run, but, like an x-ray machine at a hospital, it would be a relatively small amount, and Helion uses shielding to limit the amount of radiation that leaves the machine.

If someone were to stand at our site’s perimeter for every second of an entire year, the extra radiation they experience would be:

-Half as much as one airplane flight between New York and Los Angeles
-About as much as spending two weeks in Denver
-Half as much as the radiation received from eating a banana every day for a year

Great to see a model with relatively fast iteration (vs big mega-projects that take decades to plan and build).

I know it makes people sound smart to be cynical about fusion power.

“It’s the energy of the future, and always will be!”, “forever 30 years away” and such..

But lots of things have seemed like they were very far away until not long before we did them, and then they rapidly become mundane and un-remarkable (heavier-than-air flight, or sequencing human genomes, or having inexpensive Star Trek-style voice and touch-activated super-computers with high-resolution screen and cameras connected to a global network of supercomputers and GPS satellites in our pockets).

I hope fusion will also be one of them, because having access to plentiful clean energy at very very low cost would make the world such a better place — not just by removing harmful pollution, but also because abundant cheap energy makes all kinds of other things possible and can help bring a lot of people out of poverty.

Extra info: If you want to learn more about fusion, what it is, what are the challenges, and the current state of the art, these two podcasts are excellent sources (warning: very nerdy):

• 157 – Fusion at ITER

• 304 – The Past, Present and Future of Fusion Energy

AWS Launches its Graviton 3 ARM CPUs

AWS Graviton3 is the latest in the Graviton family of processors that are custom-designed by AWS to enable the best price performance for workloads in Amazon EC2. They provide up to 25% better compute performance, up to 2x higher floating-point performance, and up to 2x faster cryptographic workload performance compared to AWS Graviton2 processors. Graviton3 processors deliver up to 3x better performance compared to Graviton2 processors for CPU-based machine learning workloads, with support for bfloat16 and fp16 instructions. Graviton3 processors also support pointer authentication for enhanced security in addition to always-on 256-bit memory encryption available in AWS Graviton2. (Source)

AWS says the new chips will use 60% less energy.

---

The Arts & History

How Zelda: Breath of the Wild Redefines the Open-World Game (Warning: Lots of swearing)

You can contrast the above with this other video-essay by yakkocmn about the problem with most open world games.

Mad Men, Christmas Waltz Edition

Yesterday, I watched Mad Men S5E10, titled ‘Christmas Waltz’.

Another great one, lots of good plot lines in the mix. I mean, Paul Kinsey being the Krishnas’ best closer, writing a spec script for Star Trek’s second season?

I particularly liked the bar scene with Don and Joan. Don’s most interesting relationships with women are his platonic ones, like with Peggy and Joan. He clearly recognizes their intelligence, while most others around them don’t.

But I hate seeing Lane dig his grave… Also, apparently spaghetti sauce wasn’t common in 1966 NYC? (though I guess it would’ve made a much bigger mess when Megan threw that plate…)

mad men pics @madmenpics

“Guy Walks Into an Advertising Agency” (Season 3, Episode 6)