401: Stock Indices & The Ship of Theseus, Amazon, Emergent AI, Internet Archive, Build Fast, GPT-4, and Orson Welles

"I feel like I’m David Chang or something!"

The person who says he knows what he thinks but cannot express it usually does not know what he thinks. —Mortimer Adler

🤪 I have been trying to add a little bit of fun to my everyday life.

It all adds up!

For example, we used to keep our cooking oil in the metal bottle on the right. It worked fine, but it wasn’t fun.

I kept seeing experienced chefs use squeezy bottles in random cooking videos and it seemed way more fun, so I got one. Even though for my needs it isn’t that different functionally, it turns out it *is* way more fun!

Every time I’m squirting oil in a pan I feel like I’m David Chang or something! 👨‍🍳

This is just one example, but you can apply this principle almost anywhere. Find things that are fun to use or be around, ways to create small moments of joy and delight, and sprinkle them throughout your life!

A satisfying pepper mill, a good well-sharpened chef’s knife, and a heavy 12-inch cast iron pan all make me want to cook more. And if I cook more, it will over time improve my skills and help me eat better (for taste and health).

We often hear about removing friction so it’s easier to do the things we want to do, and it works, but we don’t focus enough on creating a pull towards these things through fun.

🛀💭 On a good day, most norms are unspoken (implicit), but on a bad they, they become deafeningly loud (explicit).

🗣️💭🇫🇷🇺🇸🇵🇱 I was having a conversation with friend-of-the-show and Extra-Deluxe supporter (💚💚💚💚💚 🥃 ) Daniel Harriman (👋 to our youngest supporter, David Mark!) about some of the benefits of being multi-lingual, and how it can give someone the ability to “step out” of a language to better understand it, as well as to indirectly do the same with our thinking process.

This led us to discuss the effect where people feel that they have slightly different personalities when speaking or thinking in a different language, and how much of our thoughts are influenced by the embedded culture in language.

It’s an interesting topic, so I decided to ask GPT-4 about it:

Q: Is there a name for the effect where multi-lingual people feel that they have slightly different personalities when speaking different language?

GPT-4: Yes, there is a term for this phenomenon: it is called the **code-switching effect**. Code-switching refers to the process of alternating between different languages or language varieties during speech or conversation. When multilingual speakers code-switch, they often feel that they adopt different personas or express different aspects of their personalities while speaking in each language.

Linguistic and cultural factors play a significant role in shaping one's identity, and language serves as a powerful medium to communicate emotions, thoughts, and cultural nuances. As a result, speaking different languages can evoke varying emotional associations and experiences, leading to the perception of having slightly different personalities when speaking different languages.

I was aware of behavioral code-switching. For instance, I act differently when I’m with my high-school friends vs with strangers. But thinking of it in the context of switching between languages (French and English for me) makes a lot of sense as well.

I asked GPT-4 the same question again and got another interesting angle:

the phenomenon where multilingual people feel as though they have slightly different personalities when speaking different languages is often referred to as the **language-personality link** or **linguistic relativity effect**. It suggests that one's personality, emotions, and thought processes may change depending on the language being spoken.

The language-personality link is related to the Sapir-Whorf Hypothesis, which proposes that the structure of a language affects its speakers' world view and cognition. When multilingual speakers switch to a different language, they may experience different cultural contexts, expressions, and thought patterns, subsequently influencing their behavior and perception of themselves.

Sapir-Whorf just reminds me of Arrival 👽🐙…

💚 🥃 🐇 This is the free edition of Liberty’s Highlights with 16,000+ subscribers.

You can get 2 extra editions/week full of juicy stuff + access to the private Discord 🗣🗣🗣 community by becoming a paid supporter (it’s quick & easy).

Paid post since last week 🐇:

• 🔒 399: GPT-4 vs Humans, Meta's Cubicles, Waymo, Microsoft, Stability & OpenAI, Stripe, Diversification, Credit Suisse, and Railroads

• 🔒 400: Microsoft Office's GPT-4 Copilot, TikTok Prisoner's Dilemma, Contagious Banks, Prestige Inflation, Midjourney V5, and The Offer

If you click the links above, you can see the intros for free and there’s a link to get a 7-day free trial of the paid version. What have you got to lose?

---

🏦 💰 Liberty Capital 💳 💴

Stock Indices & The Ship of Theseus: When is a Thing Not the Same Thing Anymore?

People often think they’re comparing something to an older version of itself when in fact they’re comparing two different things. This mistake is often made because the things have the same name.

A common example is stock indices. When comparing the gyrations of “the Dow/S&P500/etc” during a certain period of time and comparing it to other periods of time, it may seem like you’re trying to learn about the behavior of a singular thing to better predict its future behavior.

But the thing itself is similar to the Ship of Theseus, of which every part was replaced over time, or the proverbial grandfather’s axe, of which both the head and the handle had been replaced multiple times.

Is it the same thing?

Well, it depends on what you’re asking, and how you have replaced the parts.

Functionally it may serve the same purpose and may even have the same name. If you replaced every part with a nearly identical part, both the ship and axe may have the same functional performance as before.

But if you take a basic old axe and replace the handle with a high-performance composite handle and the head with a fancy modern steel alloy, is it the same axe?

This section of Andrew Walker’s post (💚 🥃 🎩) made me think of this (note that since he wrote these words, Big Tech has been going through a rough patch, but even if they need to cut costs and refocus for a while, they’re still excellent businesses):

I can't help but wonder if worrying about CAPE ratios or market cap to GDP is missing the forest for the trees. The largest companies in the indices (Apple, Google, Microsoft, etc.) are better businesses than anything the marker has ever seen. Their returns on incremental capital are off the charts, they are much less cyclical than the largest businesses of yesteryear, their growth opportunities are much more interesting, and honestly it's just harder to think of something that could kill them than the giants of 20-40 years ago. [...]

In 1965, of the 10 largest companies in America, you had the three major car companies (GM, Ford, Chrysler), U.S. Steel, and then four oil majors. All of those are somewhere between somewhat cyclical to super cyclical companies. Growth for those companies is also extremely, extremely expensive as those are very capital intensive businesses. Growth requires buying a new steel plant / car factory / oil well.

Are the Dow and S&P500 still the same things that they were 60 years ago?

Once again, it depends on what you’re asking… the most important thing is to make sure you understand the question you’re asking and to keep these nuances in mind.

This is an updated version of something I wrote in the Eoarchean era of Edition #20. I’m making the bet that you weren’t around back then and haven’t seen it before, but if you have, kudos to you, you’re OG! 🎩

🛒 🚚📦📦📦📦 Amazon keeps cutting ✂️

Jassy wrote in a memo:

As we’ve just concluded the second phase of our operating plan (“OP2”) this past week, I’m writing to share that we intend to eliminate about 9,000 more positions in the next few weeks—mostly in AWS, PXT, Advertising, and Twitch.

Why didn’t he announce all the cuts at the same time, along with the 18,000 cuts from January?

Some may ask why we didn’t announce these role reductions with the ones we announced a couple months ago. The short answer is that not all of the teams were done with their analyses in the late fall; and rather than rush through these assessments without the appropriate diligence, we chose to share these decisions as we’ve made them so people had the information as soon as possible.

The same is true for this note as the impacted teams are not yet finished making final decisions on precisely which roles will be impacted. Once those decisions have been made (our goal is to have this complete by mid to late April)

Sounds like more are coming…

📚👩🏻‍⚖️ Book Publishers suing the Internet Archive for ‘not being a real library’ 🤦‍♀️ 😡

I 💚 the Internet Archive.

You may know it better by its URL of Archive.org, or the amazingly useful Wayback Machine that archives snapshots of millions of websites over time.

I use a browser extension to archive versions of the webpages that I read when I notice that they aren’t on the IA yet — years down the line, if URLs in my notes are dead, I know I’ll be able to still find what I saw thanks to the Wayback Machine.

It’s an extremely important institution in the digital world.

So I’m a little angry to see this:

Despite its incredible library collections, which serve the needs of millions of people, Hachette Book Group, HarperCollins Publishers, John Wiley & Sons Inc., and Penguin Random House assert that the Internet Archive is not a real library.

In their lawsuit against the Internet Archive, which could extract millions of dollars from the nonprofit organization, the publishers claim that the Internet Archive “badly misleads the public and boldly misappropriates the goodwill that libraries enjoy and have legitimately earned.”

In their view, the archive’s “efforts to brand itself as a library” are part of a scheme to “fraudulently mislead” people, circumvent copyright law and limit how much profit publishers can extract from the ebook market. They describe the Internet Archive as a “pirate site” and its business model as “parasitic and illegal” and characterize controlled digital lending as “an invented paradigm that is well outside copyright law.”

More details here.

Do you know what I would like to see?

A tech billionaire needs to endow the Internet Archive with a bunch of money for operations AND a bulletproof legal defense fund. 💰💰💰

If Wikipedia is our library of Alexandria, the Internet Archive is the Internet’s oldest and largest museum, with a priceless collection of digital artifacts that would otherwise be lost to 404 errors, shut down servers, and bit rot. 🖼️

📈 Tracking AI’s ‘emergent utility’ as large language models scale in complexity 🤖🦾🦿🪛🪚

I’m going to put this one in the business section because I think we’re likely to see the impact hit the economy quite a bit:

Recent investigations like the one Dyer worked on have revealed that LLMs can produce hundreds of “emergent” abilities — tasks that big models can complete that smaller models can’t, many of which seem to have little to do with analyzing text.

They range from multiplication to generating executable computer code to, apparently, decoding movies based on emojis. New analyses suggest that for some tasks and some models, there’s a threshold of complexity beyond which the functionality of the model skyrockets.

“That language models can do these sort of things was never discussed in any literature that I’m aware of,” said Rishi Bommasani, a computer scientist at Stanford University. Last year, he helped compile a list of dozens of emergent behaviors, including several identified in Dyer’s project. That list continues to grow. (Source)

The list now tracks 137 emergent abilities, including some pretty impressive stuff like “multi-step reasoning by generating a chain of thought before giving the final answer” and “multi-stage hierarchical reasoning for complex reasoning”.

Some of the emergent skills are quite unexpected. For example, I asked BingGPT to summarize the plot of various films and TV shows using emojis, and it did quite well:

This may seem like a silly example, but if I take a moment to think about it, there’s a fair amount of nuance and abstraction involved in going from plot summary to generic pictures that have to stand for something — BingGPT makes a connection between something very specific and something fairly general (the emojis graphics), and the reader needs to be able to reverse that connection (go from the general emojis to the specific plot of the film that they’re familiar with). There’s a kind of theory of mind in there, like, you have to decide what will stand out from all the plot details and that the representation will make sense...

You can’t be too literal about it! Using the alien & octopus emojis for Arrival is quite clever.

Or the black heart for the ending of The Last of Us 🥲

🚧 🏙️ Why build fast? + ‘When did New York start building slowly?’ 🌆 🏗️

Very interesting post by Brian Potter over at Construction Physics about skyscraper construction over time, particularly in New York City and Chicago.

But first, I love this introductory paragraph.

It should be plastered all over to remind people that speed isn’t just about being impatient or bragging rights or whatever:

It’s good to be able to build things quickly. 

The faster you build something, the quicker the benefits from it accrue, making it more valuable. Additionally, you can reallocate your resources to another project, letting you build more things overall. Building things faster also generally means building them cheaper. If you’re building slowly, you’re probably building inefficiently, spending more time and effort than is really required. It’s also hard to avoid accumulating costs as a project drags on: you end up with more financing costs, more project status meetings, more time spent getting people up to speed as the team working on the project changes. And the longer a project takes, the more chances there are for something to go horribly wrong.

Back to NYC skyscrapers:

To sum up, construction speed of New York skyscrapers has declined significantly over time, particularly after 1970. Depending on the metric you use, modern New York skyscrapers are built at 40-70% the speed of those built in the 1940s-60s, and around a third of the speed of those built in the 1930s (though the unusually fast Empire State Building distorts the data here.)

Chicago, the only other US city that has built significant numbers of skyscrapers for many decades, doesn’t show this same pattern of decline (though it also doesn’t build as fast as it once did), and New York skyscrapers are built at around 50-70% the speed of Chicago skyscrapers.

This seems partially due to New York constructing many more skinny skyscrapers with very small floor plates than Chicago does, which take proportionately longer to build. 

---

🧪🔬 Liberty Labs 🧬 🔭

‘GPT-4 is able to infer authorship from a passage of text based on style and content alone.’ 📄🔍🤖

I wrote about statistical analysis of text to identify the author, but I never thought of trying it with GPT-4.

Mike Conover did the test. He gave the LLM a few paragraphs of text from a recent Stratechery post and asked the AI if it could identify the author.

Here’s what it replied:

To be clear, the text couldn’t have been part of the training data since it’s so recent, and this isn’t BingGPT with internet access, so it really seems to have identified Ben just from statistical analysis of his writing with a relatively short excerpt (the more text you have, the better the ‘fingerprint’ you can develop — if you have pages and pages of text by someone, that makes them much easier to ID).

I ran my own experiment with text from Matt Levine’s last 2 columns, and each time it replied:

The author of this text is Matt Levine, a writer for Bloomberg Opinion.

🎯

🤖🏋️‍♂️🧠 GPT-4’s Advice on how to become smarter

GPT-4's advice on how to become smarter is actually pretty, pretty good 🤔

Humans can become smarter by adopting various strategies and practices to develop their cognitive abilities, knowledge, and skills. Here are some suggestions for enhancing intelligence:

1. Continuous learning: Lifelong learning contributes to cognitive improvement. Keep exploring new subjects, read books or articles, take courses, seek opportunities to expand your understanding, and stay updated on current developments.

2. Critical thinking: Analyze information critically and question assumptions. Practice logical reasoning and adopt objective evaluation methods to assess problems and make sound decisions.

3. Problem-solving: Engage in activities that challenge your problem-solving skills. Work on puzzles, games, and real-world problems to develop adaptive and creative solutions.

4. Memory enhancement exercises: Train your brain with memory-improving techniques, such as mnemonic devices or practicing active recall.

5. Mental stimulation: Participate in intellectually stimulating activities, like chess, sudoku, or crosswords, to keep your brain sharp and exercised.

6. Embrace creativity: Experiment with different creative outlets, be it writing, drawing, music, or any art form, to enhance your out-of-the-box thinking and generate novel ideas.

7. Cultivate curiosity: Be curious and ask questions. Developing a natural curiosity helps to widen perspectives and encourages exploration.

8. Develop a growth mindset: Believing in self-improvement and seeing challenges as learning opportunities is key to becoming smarter.

9. Physical exercise: Engaging in regular physical activity is known to improve cognitive function, memory, and overall brain health.

10. Healthy diet: Consume a well-balanced diet rich in essential nutrients, including omega-3 fatty acids, antioxidants, vitamins, and minerals, to support brain function.

11. Reduce stress: Chronic stress can impair cognitive performance, so engage in stress-reducing activities such as meditation, exercise, or hobbies.

12. Sleep: Consistent, high-quality sleep is critical for learning, memory consolidation, and overall brain function.

13. Socialize: Engaging in social interactions and building strong relationships facilitate the development of new skills, spark creativity, and improve mental well-being.

It’s nothing unexpected, but most people could benefit from being reminded of these things periodically and making a mental inventory of their lives to figure out which areas could benefit the most from improvement.

It’s once again a case of “the most important things are fairly simple, yet hard to keep in mind and put into practice”.

---

🎨 🎭 Liberty Studio 👩‍🎨 🎥

Great 1960 BBC interview with Orson Welles about how he made his first film, Citizen Kane 🎞️📽️ 🎬 🛷

He was 25 years old!

All he cared about was *control*, not money, which ultimately allowed him to make one of the all-time classics as his first film.

He didn’t know what “couldn’t be done” with the camera, so he invented innovative techniques of visual storytelling that influenced much of what came afterward.

It reminds me of this quote by the great Kevin Kelly:

Experience is overrated. Most astounding accomplishments were done by people doing them for the first time. Therefore when hiring, hire for aptitude and attitude and then train for skills.

Sounds like exactly what happened with Welles.

h/t Brian Roemmele