294: Bezos Blamed?, Wrecked by Success, Apple M2, Multiples Compression, Ad Market, Carbon Offset Scams, Cancer Good News, and The Last of Us

"do things that would seem godlike to our ancestors"

What makes an air conditioner 'magic' from the perspective of say the thirteenth century, is that even if you correctly show them the design of the air conditioner in advance, they won't be able to understand from seeing that design why the air comes out cold; the design is exploiting regularities of the environment, rules of the world, laws of physics, that they don't know about.

–Eliezer Yudkowsky

🧠 🛠 🤖 I’m fascinated by tools. Especially the invention/improvement of tools.

It’s what has taken our species of naked apes from scratching out subsistence to being able to do things that would seem godlike to our ancestors that roamed the Earth for hundreds of thousands of years with very little progress.

Our tools multiply our natural abilities and capabilities and help us come up with the next batch of tools that will further increase our powers. Rinse and repeat.

So intelligence allows us to make tools, tools allow up to do interesting and useful things, and eventually make even better tools…

I can’t help but think that *longer-term*, almost nothing matters other than developing artificial general intelligence (AGI), because it’s both the ultimate tool and it improves the “intelligence” side of the equation, opening up the tech tree to even better tools (including better artificial intelligence, that’s the whole recursive self-improvement thesis).

Once you have AGI, most problems become AGI’s to solve (need to cure cancer? figure out cheap fusion power? design spaceships or semiconductors?). Of course, in the meantime, these are very much our problems and we should work on them, because we have no idea when/if we’ll get there. But it’s a possibility.

BUT, in such a scenario, it also means that there’s nothing more important than getting AGI right and having our best minds work on the AI alignment problem because, like with nuclear weapons, having access to extreme powers requires extreme caution to avoid ending up accidentally with a future we don’t want.

🛀 🕰 🐢 🐌 🏁 Longevity. The secret weapon to success.

I feel like most people will naturally look at things from a ‘point-in-time’ viewpoint.

They can’t imagine what it’s like to do something day-after-day, month-after-month, year-after-year, before you *really* get anywhere.

Everybody *talks* about iteration, about slowly compounding improvements, about investing in yourself to get a little better every week. Well, longevity in a field is the main ingredient of what it actually *looks* like.

Our imagination just doesn’t stretch easily in that direction, I guess ¯\_(ツ)_/¯

💚 🥃 If you make just one good decision per year because of something you learn here (or avoid one bad decision — don’t forget preventing negatives!), it'll pay for multiple years of subscriptions (or multiple lifetimes).

Thank you for your support!

---

🧩 A Word From Our Sponsor: Heyday 🧠

Do you have 100+ browser tabs open right now? 😬

Give your memory a boost with Heyday, the research tool that automatically saves the content you view, and resurfaces it alongside relevant Google search results. 👩‍💻

It’s like cheat codes for your memory. 😲💡

🧩 Give your memory a boost today 🧠

---

Business & Investing

Feel the compression 🫠

I wish the data went back farther, and the Y-axis was labeled… But it does show that recent times are unprecedented in the past quarter-century.

h/t Matthew Miskin

🚚 📦📦📦📦📦 Bezos blamed for Amazon’s over-expansion?

From a WSJ piece titled: “Amazon CEO Andy Jassy’s First Year on the Job: Undoing Bezos-Led Overexpansion”:

Mr. Bezos had largely relinquished day-to-day decision making prior to the pandemic as he took on more outside interests. But with the crisis, he dove back into the business, helping guide Amazon’s strategy and ensuring that Amazon could meet the surge in demand. [...]

Mr. Bezos and other executives had greenlighted a strategy, guided by a revered internal forecasting tool, that overshot the long-term projections for demand from Amazon. [...]

In 27 months, Amazon added about as many employees as the entire workforces of United Parcel Service Inc. and Costco Wholesale Corp. combined.

“We made a decision to build to the high side to avoid constraining consumers and sellers in any way,” said Mr. Jassy at the company’s shareholder meeting in May.

I wrote in edition #275 about this choice that led to excess capacity, and the trade-offs that would have had to be made if they had decided to be more conservative. It’s not like the alternatives at the time were great either (without having a crystal ball to know how to dose things perfectly right, I mean — it’s always easy in hindsight).

Early in the pandemic, Amazon saw a pivotal moment to meet demand and expand its e-commerce reach. Under its founder, it opened hundreds of new warehouses, sorting centers and other logistics facilities, and doubled its workforce from 2020 through March, to more than 1.6 million people.

That helped the business for a time—Amazon was one of the biggest beneficiaries of the pandemic tech boom. Its revenue grew by a total of two-thirds across 2020 and 2021, and its profit nearly tripled. But demand hasn’t kept pace with that planned capacity, and its setback has been among the most pronounced. [...]

The disconnect was a major drag on earnings this year—Amazon signaled the excess space would contribute to $10 billion in extra costs in the first half of 2022. The company’s stock price has fallen by more than a third during Mr. Jassy’s tenure, erasing more than $600 billion in market value.

Now, Mr. Jassy and his team are working to sublease at least 10 million square feet of excess warehouse space, defer construction of new facilities on land Amazon has bought and find ways to end or renegotiate leases with outside warehouse owners. He has closed down much of the company’s bricks-and-mortar retail operation—68 stores—and is looking to pare back its head count. [...]

Under Mr. Bezos, Mr. Clark’s team would present its plans to leadership just a few times a year, the people said. Once Mr. Jassy took over, he demanded a weekly metrics review of the consumer business, they said.

This is a good example of the Halo Effect. When things were going great, Bezos and Jassy were seen as almost infallible, their decisions during the pandemic were seen as bold and shrewd. AMZN 0.00%↑

Now that things aren’t working out as expected, a lot of blame is put on Jassy and Bezos and suddenly people remember they’re just flawed humans. The cycle continues!

Wrecked by Success? 🤕💰🏆

It’s a common narrative that success has a cost, and those who over-achieve must be deficient in other areas of their lives, or have made heavy sacrifices to do well (and there are plenty of high-profile examples of this to reinforce the point).

But what if we take a more systematic approach and don’t reason be anecdote and confirmation bias? Here’s a pair of studies on this:

We examined the wrecked-by-success hypothesis. Initially formalized by Sigmund Freud, this hypothesis has become pervasive throughout the humanities, popular press, and modern scientific literature. The hypothesis implies that truly outstanding occupational success often exacts a heavy toll on psychological, interpersonal, and physical well-being.

Study 1 tested this hypothesis in three cohorts of 1,826 high-potential, intellectually gifted individuals. Participants with exceptionally successful careers were compared with those of their gender-equivalent intellectual peers with more typical careers on well-known measures of psychological well-being, flourishing, core self-evaluations, and medical maladies. Family relationships, comfort with aging, and life satisfaction were also assessed. Across all three cohorts, those deemed occupationally outstanding individuals were similar to or healthier than their intellectual peers across these metrics.

Study 2 served as a constructive replication of Study 1 but used a different high-potential sample: 496 elite science/technology/engineering/mathematics (STEM) doctoral students identified in 1992 and longitudinally tracked for 25 years.

Study 2 replicated the findings from Study 1 in all important respects. Both studies found that exceptionally successful careers were not associated with medical frailty, psychological maladjustment, and compromised interpersonal and family relationships; if anything, overall, people with exceptionally successful careers were medically and psychologically better off.

🌎 👀 The Matthew Effect in the Ad Market 📈

Five of the world's largest tech companies owned more than half (53%) of all global ad revenues last year, up from 46% last year.

The world's top 25 advertisers now account for 74% of global advertising spend, up from 68% last year and just 43% in 2016.

While Chinese companies continue to grow their portion of the ad market, they are mostly focused on capturing domestic dollars, with TikTok being the notable exception.

Source.

‘How curing aging could help progress’ 👴🏻→🏋️‍♂️

This is a fascinating topic to explore — both on the biomedical side and the futurism/thinking about the implications side — but today I just want to share a limited sub-section of the latter from the excellent Jason Crawford:

here are three reasons why curing aging could help progress:

1. Population. One of the greatest threats to long-term progress may be a slowdown in global population growth. We need more brains to keep pushing science and technology forward. Yet right now, many wealthy nations have fertility rates below replacement levels. Curing aging would help temporarily by lowering the mortality rate. It could help permanently if people decide to have more children, on average. That might happen if longer lifespan means people feel they have time for both children and a career. (Remember that fully curing aging means maintaining reproductive health for all those years.)

2. Burden of knowledge. There is a hypothesis that as knowledge grows, it takes longer to reach the frontier, and so individual researchers have less time to contribute advancements. They are also forced to specialize—but breakthroughs often come from making connections across far-flung disciplines. If individuals had much longer lifespans, it would be no problem for them to spend 30 or 40 years just learning, before making major contributions. And you could spend another 10 or 20 years picking up a couple more specialties in disparate areas.

3. Long-term thinking. How would people’s thinking change if they felt they were going to live 150, 300, even 1,000 years or more? The very long-term becomes much more personal. Posterity is something you’re going to be around for.

I tend to think that #2, taken generally, is incredibly powerful.

In the same way that compound interest is almighty mathematically, compounding knowledge over longer periods inside of brains that don’t suffer an age-related decline in cognitive capacity would be one of these “when you have enough quantitative, it becomes qualitative” changes.

How Carbon Offset Fail to Deliver (+Scams) 🪚🌲🌲🌲🌲

In theory, theory and practice are the same. In Practice, they aren’t.

In theory, offsets make a lot of sense. The Earth is all one system, so all else equal, if you do something here and do the opposite somewhere else, these things can potentially cancel each other out.

For example, if you power your business with electricity that comes from natural gas, but you finance the creation of the equivalent amount of clean electricity somewhere else that displaces the equivalent amount of carbon, the net amount can be zero.

But in practice, it all comes down to whether you’re really displacing or preventing the equivalent elsewhere. Faulty assumptions and bad accounting often mean that the offsets that are being purchased are actually offsetting just a fraction of what is claimed (or sometimes, nothing at all — the modern papal indulgences!).

💊🇺🇸 ‘U.S. drug overdose deaths per year’ 😢

This is a problem.

Source. h/t Friend-of-the-show and supporter (💚 🥃) Frederik Gieschen

---

Science & Technology

🔥 🍎 Apple M2 Overview 🔥

One of my favorite Apple-watchers, René Ritchie (🍏🇨🇦) does a good overview of Apple’s new M2, nerding out a bit on the new goodies.

One of the dynamics that can be lost when you just look at the headline numbers is that the efficiency CPU cores got a much bigger boost than the performance core on this iteration, which means that more workloads can be done entirely on the efficiency cores.

This should further optimize battery life (Apple sells a lot more laptops than desktops, and for most people, an extra hour of battery life is a lot more valuable than a few percent faster CPU, especially since to most people CPU isn’t the bottleneck on the M-series of chips). Life is trade-offs… AAPL 0.00%↑

☢️ ‘Why I changed my mind about California’s Diablo Canyon nuclear plant’ ☢️

You know the wind is changing direction when even anti-nuclear US senators from the state where the anti-nuclear movement has deep roots are writing op-eds arguing for California’s last nuclear power plant to *not* be shut down:

Pacific Gas and Electric Co. should reconsider its decision to close Diablo Canyon by 2025. The utility should get the plant relicensed instead, retiring it once the state can replace its production with clean sources.

When PG&E announced the planned closure in 2016, I supported the decision. I remain concerned about the lack of long-term storage for spent nuclear fuel and am working to develop better solutions. [

It’s funny how it’s impossible to talk about nuclear power without the whole “but what about the waste? We have to find a solution! What a dealbreaker!”

First of all, do people realize that something staying dangerous for X years is still better than a lot of things that stay dangerous forever. Mercury and arsenic and other heavy metals in coal fly ash stay dangerous to humans and other living things *forever*.

Second, we know what to do with waste. You encase it in concrete and steel, and the radiation doesn’t get out. Then you put that relatively small quantity of stuff on-site at nuclear power plants because those locations are over-engineered to be some of the safest places in the world, or you bury it in a geological location where the soil and water take millions of years to move by 1 meter.

This isn’t a Holywood movie, the radioactive stuff isn’t sentient, scheming to magically escape by outsmarting us.

(but maybe let’s not bury it too deep because future breeder nuclear reactors may want to use that “waste” as fuel since something like 97% of the energy remains in “spent” fuel).

Back to the op-ed:

Closing Diablo Canyon would remove 18,000 gigawatt-hours from the grid, nearly 10% of the state’s electricity generation. This is an extraordinary amount of power for a grid facing reliability concerns amid heat waves and wildfires. When the power goes out, lives are endangered.

Moreover, the plant generates 15% of the state’s carbon-free electricity. At least in the short term, that would have to be replaced with fossil fuel generation. [...]

A report last year by Stanford University and the Massachusetts Institute of Technology found that if PG&E extends operations at Diablo Canyon until 2035, California could reduce power sector carbon emissions by over 10% below 2017 levels, saving $2.6 billion and significantly reducing natural gas use.

The report also highlighted the reliability issues the state could face without Diablo. State and grid officials recently issued similar warnings about reliability, warning that peak summer demand may cause significant blackouts. In August 2020, extreme heat led to rolling blackouts and electricity shortages in California. If Diablo had not been running, the shortage could have been three times worse.

Yeah, that sounds like it would be bad. Who could have foreseen this before they voted to close a perfectly good power plant producing large quantities of clean energy?

🥼 ‘A cluster of "un-heard of" results in multiple subtypes of advanced cancer’ 🧫🔬

We can all use more good news, right?

Here’s an overview of a few recent studies that show progress in science’s fight against cancer:

The narrative has been incubating for many years, but in recent days we are witnessing some extraordinary progress in treating and monitoring cancer. The convergence of genomics of the cancer—be it from the person’s DNA or tumor directly or the blood (known as liquid biopsy)—matched with the appropriate therapy is leading to outcomes that are being described as “unheard-of” by expert oncologists. This represents the essence of individualized medicine, whereby understanding the unique biologic basis of a person’s cancer can lead to highly accurate and effective treatment, and also avoid the toxicity of classical chemotherapeutic agents. Here I will review 4 recent studies, all published in the last week, and a new screening test that has become available. [...]

A new report of 18 consecutive patients with MMR, all treated with programmed-death (PD-1) blockade immunotherapy had complete remission of the tumor as represented below by serial, direct endoscopic visualization.

One of the authors, Dr. Diaz of Sloan Kettering, said “I believe this is the first time this has happened in the history of cancer” while an outside expert at UCSF described it as “unheard-of.” Of course this is a finding that needs to be independently replicated with follow-up for potential recurrence. But it has to be seen as good news for individualized therapy of this subtype of cancer, in contrast to the current standard of care that involves chemotherapy, radiation and surgery that only achieves complete remission in 1/4th of patients.

Read the rest.

---

The Arts & History

A close look at the graphics improvements in the remaster of The Last of Us 🎮

It’s a bit weird, but The Last of Us is one of my favorite games of all time, yet I’ve never played it.

I didn’t have a Playstation when it came out (I still don’t), but I wanted to check it out so I started watching someone’s playthrough on Youtube (it was Day9, for those of you who used to follow the Starcraft scene), and I was so impressed, I ended up watching the whole thing from beginning to end.

It felt a bit like I was sitting on the couch next to a friend while he played. Since then, I’ve ranked that game up there with some of my favorite movies as a media experience.

I’m glad to see that it is being updated on the technological front — anything that increases its longevity and allows a new generation of players to discover it can’t be a bad thing.

The video above gets pretty geeky in zooming in on rendering details, but I think you know that this is a nerding-out zone ¯\_(ツ)_/¯