What We’re Reading (Week Ending 03 July 2022) - 03 Jul 2022
Reading helps us learn about the world and it is a really important aspect of investing. The legendary Charlie Munger even goes so far as to say that “I don’t think you can get to be a really good investor over a broad range without doing a massive amount of reading.” We (the co-founders of Compounder Fund) read widely across a range of topics, including investing, business, technology, and the world in general. We want to regularly share the best articles we’ve come across recently. Here they are (for the week ending 03 July 2022):
1. Why Foundational Models will Revolutionize the Future of Work and Play – Daniel Jeffries
It’s 2033 and you’re coming home from a dinner and realize your sister’s birthday is tomorrow and you forgot.
You ask your phone what’s the best gift for her and where you can get it at this late hour?
Your phone has dedicated processors for running Machine Learning (ML) models locally but it’s not powerful enough to answer that question with its small memory and slower chip speed. But it is strong enough to ask a more powerful model in the sky.
The local model also learned a lot about what you and your family likes over time, so it packages up some key things it knows, anonymizes them, and fires off a query to a Foundational Model (FM) in the cloud via API.
In a fraction of a second, the answer comes back.
Your sister’s latest social media pics show she recently got on a serious health kick, lost weight, stopped drinking and got really into vegetarian cooking so it recommends an AirBnB cooking experience near her, with a local vegetarian chef. It gives you two alternative experiences that are good but a bit further away and not as highly rated. You don’t even need to go to the store and it’s the perfect present that makes you look like a hero…
…Across the world people are using cascaded FM’s networked together to do amazing work. FMs on their own are amazing but working together they’re capable of astonishing feats and when they work with you they’re centaur units, a combination of man and machine working together to create something neither could do on their own.
Centaurs are named after Gary Kasparov’s early experiments with chess tournaments, where an AI and human teams bested pure AI and humans on their own. The tournament’s name came from the mythical beast of Greek legend that’s half horse and half man, symbolizing how man and machine can work together better…
…Biotech companies search through massive databases of proteins and chemical interactions and quickly use a fine tuned FM to design twenty potential drug candidates to fight a rare motor neuron disease that recently cropped up in South Africa.
A musician jams out a new tune and then asks the models to iterate on the chorus. The 17th one is awesome and the musician plays it and then modifies it with a few tweaks to make it even more catchy based on a song he couldn’t get out of his head a week ago that he overheard on a radio at the local park. It goes on to be a huge hit on Soundcloud.
Materials scientists are designing new materials that make everything stronger and lighter, from skyscrapers that flex more easily to resist earthquakes, to electric bikes that are light enough to carry on your shoulder and fold up neatly to carry on the train.
Elite coders are simply telling the coding model what they want it to do and its spitting out near perfect Python code but it also recommends Go for several libraries because it will be faster and more secure. It automatically does the translation between languages and tests it. It’s paired with an evolution through language model (ELM) coupled with a Large Language Model (LLM) and those models helps the coder create brand new, never before thought of code too, in a domain the model was never trained for by iterating on concepts quickly.
All of it is happening because of a vast global network of ambient AI models. AI is everywhere now. Every device is waking up and getting smarter. We’ve industrialized intelligence and sparked a revolution in how we work, design, and play.
Welcome to the age of ambient AI…
…What are Foundation Models and why do they matter?
In essence, FMs are large models that exhibit remarkable capabilities, such as the ability to understand language, reason, create working computer code, do translations and arithmetic, understand chains of logic, generate totally new art from text prompts, and much much more.
The basic concept of FMs comes to us from Stanford University where they primarily refer to Large Language Models (LLM), like GPT-3, that are typically transformers. But the implications of FM’s go way beyond today’s architectures. They’re a groundbreaking type of software, that’s not limited to transformers or language.
We can think of FM’s are any large and sophisticated model. We can also think of them as a chain of cascading models that work together to do a complex task such as generate music or images or video, create mathematical proofs, design new materials or discover new drugs and more.
Many of them are already here.
GPT-3, from OpenAI, powers GitHub co-pilot that quickly writes code for developers, especially boring, repetitive code so they can focus on more creative tasks. It’s one of the first fantastic examples of a centaur. Originally, GitHub’s team wasn’t sure who would use it. Would it be beginning or advanced coders? Since its wider release to all developers, the answer is clear: advanced coders love it and use it most often. Advanced coders are in the best position to understand when the model makes a mistake and it dramatically speeds up their day to day coding…
…In another article, called The Coming Age of Generalized AI, I highlighted researchers who were working on even more groundbreaking approaches by combining mega-models with several other key techniques. One of techniques, called progress and compress that comes to us from DeepMind, combines three techniques, progressive neural networks, elastic weight consolidation and knowledge distillation.
The idea is simple. Create two networks, a fast learning network and a base model. That roughly mirrors the functioning of our brain yet again. Think of it as the hippocampus and neocortex. As Hannah Peterson writes in her article on catastrophic forgetting, “In our brains, the hippocampus is responsible for “rapid learning and acquiring new experiences” and the neocortex is tasked with “capturing common knowledge of all observed tasks.” That dual network approach is called a progressive neural network.
The fast neural network is smaller and more agile. It learns new tasks then transfers the finalized weights to the base model. So you end up with a lot of stored neural networks good at a bunch of tasks.
But there’s a problem with basic progressive neural nets. They don’t share information bi-directionally. You train the fast network on one task and freeze those weights and transfer them to the bigger network for storage but if you train the network first on recognizing dogs, it can’t help the new network training on cats. The cat training starts from scratch.
Progress and Compress fixes that problem by using a technique called knowledge distillation, developed by deep learning godfather Geoffrey Hinton. Basically, it involves averaging all the weights of different neural nets together to create a single neural network. Now you can combine your dog trained model and cat trained model and each model shares knowledge bi-directionally. The new network is sometimes slightly worse or slightly better at recognizing either animal but it can do both.
It opens the door to cat-like intelligence.
A cat is a remarkable creature. It can run fast, sleep in tiny boxes, find food and water, eat, sleep, purr, defend itself, climb trees, land on its feet from great heights and a hundreds of other subtasks. A cat won’t learn language or suddenly start composing poetry. That’s perfectly fine because a cat is really well suited to its set of tasks; it doesn’t need to build skyscrapers too.
Having a cat level intelligence is incredibly compelling. If you have a cleaning robot that can wash dishes, pick up clothes, fold them, carry them from place to place and iron shirts, that’s an incredible machine that people would clamor to buy. It doesn’t also need to write music, craft building blueprints, talk to you about your relationship problems, and fly a plane too…
…AI is a universal, general purpose technology.
The greatest breakthroughs in history are always universal technologies that affect a broad range of sectors as they branch into countless other domains and inspire unexpected breakthroughs.
Think of the printing press and the way it leveled up human knowledge across the board because now we could scale, save and replicate knowledge much faster.
Think steam engines that changed the very nature of work from human and animal powered muscle work to work done by machines.
Think of the microprocessors and computers that changed how we do art, communicate, design skyscrapers and houses, fight wars, find love, do science, make music and movies and more.
A general purpose technology like AI has direct and secondary effects on the world at large, both good and bad and everything in between.
We can think of ideas and technology as they grow and change and affect both their own domains and unexpected domains as a growing tree. The roots are precursor ideas that eventually inspire the primary idea. The trunk is the central breakthrough idea, which leads to a branching series of closely related ideas and some unexpected inventions in parallel domains.
2. Reducing Inflation Will Come at a Great Cost: Stagflation – Ray Dalio
More specifically, I now hear it commonly said that inflation is the big problem so the Fed needs to tighten to fight inflation, which will make things good again once it gets inflation under control. I believe this is both naïve and inconsistent with how the economic machine works. That’s because that view only focuses on inflation as the problem and it sees Fed tightening as a low-cost action that will make things better when inflation goes away, but it’s not like that. The facts are that: 1) prices rise when the amount of spending increases by more than the quantities of goods and services sold increase and 2) the way central banks fight inflation is by taking money and credit away from people and companies to reduce their spending. They also take buying power away by raising interest rates, which increases the amount of money that has to go toward paying interest and decreases the amount of money that goes toward spending. Raising interest rates also lowers spending because it lowers the value of investment assets because of the “present value effect” (which I won’t get into because it would be too much of a digression), which further lowers buying power. My main point is that while tightening reduces inflation because it results in people spending less, it doesn’t make things better because it takes buying power away. It just shifts some of the squeezing of people via inflation to squeezing them via giving them less buying power.
The only way to raise living standards over the long term is to raise productivity and central banks don’t do that…
…In summary my main points are that 1) there isn’t anything that the Fed can do to fight inflation without creating economic weakness, 2) with debt assets and liabilities as high as they are and projected to increase due to the government deficit, and the Fed also selling government debt, it is likely that private credit growth will have to contract, weakening the economy, and 3) over the long run the Fed will most likely chart a middle course that will take the form of stagflation.
3. The Beer Game – Peter Dizikes
Thursday, August 29, 1:00 p.m.
It is a miserably muggy afternoon in Cambridge as the incoming class of the MIT Sloan School of Management—roughly 400 students from 41 countries—files into a second-floor ballroom at the Kendall Square Marriott. They are here to play the Beer Game, a Sloan orientation tradition. Unfortunately given the weather, the Beer Game does not involve drinking cool beverages…
…Rather, the Beer Game is a table game, developed in the late 1950s by digital computing pioneer and Sloan professor Jay Forrester, SM ’45. Played with pen, paper, printed plastic tablecloths, and poker chips, it simulates the supply chain of the beer industry. In so doing, it illuminates aspects of system dynamics, a signature mode of MIT thought: it illustrates the nonlinear complexities of supply chains and the way individuals are circumscribed by the systems in which they act…
…1:30 p.m.
Each Beer Game team is divided into four units of two players each, who play the roles of retailer, wholesaler, distributor, and brewer. The goal is to keep team operating costs as low as possible. We learn that teams will be penalized for having too much inventory (50 cents per case of beer per week) or unfilled back orders ($1 per case per week). Each link in the supply chain keeps track of its own costs, but a team’s score is the sum of these tallies. The lower the score, the better.
As we begin the first of 50 rounds (which represent weeks), each retailer unit draws a card indicating consumer demand for cases of beer; at the same time, all the units send slips of paper with orders up the supply chain. In response, cases of beer—represented by poker chips—move in the opposite direction, from brewer to retailer. A small number of chips are already at every station when we start.
2:15 p.m.
After 20 rounds, my team is on a hot streak.
I’m sitting at the retailer station with finance student Adah Jung, who’s been submitting orders at a level closely mimicking consumer demand. Our score at the retail station is low, and there are few chips elsewhere on the table, meaning our team’s costs are minimal. It’s hard to see how things could go wrong: with seven smart teammates and a stable supply chain, why can’t we win this thing? I can almost hear Sterman asking us to stand for a round of applause.
2:35 p.m.
Seemingly out of nowhere, our team’s distributorship has an inventory of 178 surplus cases of beer, which lasts seven weeks, adding $623 to our costs in a game where the average score after 50 weeks is $2,000 per team. How did that happen? Can’t someone tell our two teammates at the brewery just to stop making so much beer?
Well, no. “I can’t tell them anything,” observes teammate Juan Trujillo. Indeed, to simulate the incomplete information we deal with in real life, players cannot communicate across stations, apart from relaying orders. And somehow, someone on our team ordered way too much beer…
…3:30 p.m.
Sterman’s assistants tape charts to the ballroom walls detailing every team’s performance. Today’s winning score was $460 (the best possible score is about $200), while the worst-performing team racked up $6,618 in costs.
Sterman initiates a discussion, pointing out how inventories and backlogs spike and plummet erratically. The distributor on today’s last-place team went from a backlog of 70 cases to an inventory of 191 in three weeks.
One thing to learn from the Beer Game, then, is why many businesses experience boom-and-bust cycles—oil and gas exploration and housing among them. Complex systems produce nonlinear phenomena.
4:15 p.m.
Sterman pounds home a bigger lesson: our psychological habits and limited perspectives often keep us from properly understanding complex systems. To prove it, he asks distributors, wholesalers, and brewers to estimate their consumer demand; their responses are wildly inaccurate.
All too often, Sterman adds, this means we attribute problems to other people rather than to flawed systems. For instance: “I found that some people were kind of slow to take corrective action,” offers one student—who had just played for the winning team, a fact Sterman emphasizes to much hilarity.
It doesn’t make sense for us retailers to blame our teammates—who had imperfect information—for our disappointing scores. “It just cannot be true that, by chance, all the smart people ended up as retailers and all of the people running the factories were dumb,” Sterman says. The Beer Game’s structure makes it hard for certain players to perform well. It’s not the people; it’s the system.
Thus, firing people tends to be a futile management action. “Your role as a leader is to create a system in which everybody can thrive,” he says…
4. Why does the Stock Market go up? – Eugene Ng
A Google Search of “Why does the Stock Market go up?”, and Investopedia gives you up a broad range of factors.
The factors range from the supply and demand of buyers and sellers, to economic indicators, consumer confidence, wars/politics, concerns over inflation / deflation, government fiscal / monetary policy, technological changes, natural disasters or weather events, corporate or government performance data, regulation/deregulation, and the level of trust in the financial sector and legal system, amongst so many others.
But this doesn’t really answer the question, doesn’t it? It only leaves you, more confused, and begging for a better answer…
…The factors listed above are not wrong. Yet, they do not help you figure out why stock prices rise.
In the short-term, stocks will move up and down for a variety of random reasons — all of which does nothing to increase your chances of a positive return.
Thus a better question would be:
“Since the short-term does not really matter as much, why then does the stock market go up over the long-term?”
To get closer to the truth, you need to understand the components which drive the returns on your stock investment.
The Total Shareholder Return (TSR) from holding common publicly-traded stocks can be broken down into three key components: (1) growth in Earnings per Share (EPS), (2) change in the Price-to-Earning (PE) valuation multiples, and (3) earnings from dividends…
…With S&P Global providing us with historical data on the S&P 500’s closing levels, Sales per Share (SPS), Earnings per Share (EPS) and Dividend per Share (DPS), they provide clues on what the growth has been thus far…
…Take 2021 to 2003, the longest period spanning over 18 years (first row, last 5 columns from the right). During this time, the S&P 500 Index more than quadrupled from 1,112 to 4,766, with TSR* growing by ~4.3X (8.2% CAGR).
The contribution of the Earnings per Share (EPS) growth is telling. Earnings per Share (EPS) grew by ~4.1X (8.1% CAGR) from 48.7 to 197.9. Further breaking down that EPS growth, Sales per Share (SPS) grew by ~2.2X (4.5% CAGR) and Net Income Margin Growth (NIM) grew by ~1.8X (3.5% CAGR).
Thus the growth in earnings (EPS) accounted for the majority (~95%) of the TSR* growth, with growth in sales/revenues (SPS) and improvement in net income profit margins (NIM) accounting for ~52% and ~43% of TSR* growth respectively…
…Given what we have laid out so far, you, you should not be surprised to learn that over the long-term, it is earnings growth, supported by revenue and profit growth, that drives the stock market higher, and to a much lesser extent, valuation multiples.
5. Pioneer Helped Turn Her Family Store Into Japan’s Biggest Retailer – Chieko Tsuneoka
First her father died young, then her mother, then her older sister. At 23, Chizuko Okada inherited the job of running her family’s clothing store in Mie prefecture, Japan.
It was 1939, and war with America was just around the corner. Few could have foreseen that the little business would develop into Japan’s largest retailer by sales—or that a woman would be its driving force.
By the time Chizuko Kojima—her married name—died on May 20 of old age at 106, the company now known as Aeon Co. had thousands of stores around Japan and the rest of Asia and annual revenue equivalent to $64 billion…
…Ms. Kojima was born on March 3, 1916, as the second daughter of the Okada family, which had run a fabric and kimono store since 1758 in Mie prefecture, just west of Nagoya in central Japan.
Chizuko’s father, Soichiro Okada, modernized the business but died of heart disease in 1927 at age 43. Then Japan was hit by the Great Depression, which caused bankruptcies and joblessness.
In a 2003 book, Chizuko wrote that she believed it was necessary to be ready for such cataclysms by studying history. The hard times deprived her of a chance to pursue higher education in Tokyo.
After taking over the family business, Chizuko managed to keep it going during World War II until a U.S. bombing raid destroyed much of their home city of Yokkaichi in June 1945, including the Okada store’s stock.
At the time, customers held coupons similar to gift certificates entitling them to store goods. The store no longer had anything to offer, but Chizuko posted notices throughout the city saying her shop would give cash in exchange for the coupons, recalled her younger brother, Takuya, in a 2005 autobiography. It was a way of maintaining customers’ loyalty that would pay ample dividends in years to come.
Chizuko loved studying and during the war, she read a book about Germany’s inflation after it lost World War I. When Japan surrendered in World War II in August 1945, she predicted the same would happen. She gathered her cash and bank loans and bought as much merchandise as possible, reopening the shop in March 1946, ahead of an inflationary surge that hurt other businesses.
“All the merchandise flew off the shelves,” Takuya recalled.
Chizuko wrote of the episode, “Through my own experience, I learned the importance of studying and reading records of the past.”…
…In 1959, when the Okada family business still had just two stores, she came back to take charge of personnel and other behind-the-scenes management issues.
That year, Chizuko and Takuya made their first visit to the U.S. and toured the famous Sears, Roebuck and Co. store in Chicago. Takuya wrote that he was impressed by the giant scale of the business. Paging through the thick Sears catalog full of pictures of refrigerators, washing machines, clothing and a myriad of other goods, he imagined the day that Japan, too, would enjoy that kind of affluent life.
Chizuko was impressed by the Sears pension system, thinking it would create a loyal workforce. She introduced one a decade later, as her brother rapidly expanded the retailer through mergers. She also introduced an in-house training organization, today known as the Aeon Business School…
…Chizuko was one of the first managers in Japan who aggressively hired female full-time employees and homemakers as part-timers. She saw that many women worked in the U.S. and believed Japan should follow suit.
By having women at the company, “we were able to bring on board the viewpoint of the customer—how much to sell and at what price,” she said in a television interview when she was 90.
6. Make Haste Slowly – Chris Mayer
I had been reading The Art of Worldly Wisdom: A Pocket Oracle, a book written in 1647 by Baltasar Gracian, who was a witty Jesuit from Spain. His book of 300 aphorisms, with his commentary on them, has been translated into many languages and has earned the praise of many philosophers ever since.
Arthur Schopenhauer loved it so much that he prepared a German translation himself. Schopenhauer said it was particularly good for young people, as it would give them experience it would otherwise take years to obtain. “To read it through once,” he wrote, “is obviously not enough; it is a book made for constant use.”…
…Anyway, there is a passage where Gracian talks about the motto “festina lente.” This Latin phrase is usually translated as “make haste slowly.” One must be very patient and yet ready to act swiftly. And the fastest way to achieve your goals is sometimes by doing nothing.
The motto was a favorite of the Roman Emperor Augustus. Engravers captured the idea with an emblem of a dolphin wrapped around an anchor, which they stamped on coins. Another emblem captured the same idea with a crab and a butterfly; again marrying this idea of fast and slow.
Festine lente recurs throughout history and has been captured in a variety of images, such as a rabbit coming out of a snail shell. The Medicis chose it as their motto and illustrated it with a sail-backed tortoise.
I thought the idea beautifully captured an important idea in investing that is often counterintuitive: to get where you want to go the fastest often means acting very slowly if at all…
…It does seem incredibly counterintuitive to say, “No, you shouldn’t try to sell before a recession.” Or: “No, you shouldn’t ‘reposition’ your portfolio based on recent events.” Don’t these seem like logical things to do?
Not if you want to enjoy the wonderful effects of compounding capital over long periods of time. The main problem with trying to do the above is they are too hard to do well enough. You have to think about trying to do these things repeatedly over a lifetime of investing. The odds against you are very great. Sure, you may be right sometimes. But you will most certainly sit out stretches of time where you could have earned great returns because you’re afraid of a recession. Odds are you won’t get those “repositionings” right repeatedly either.
7. How Parents’ Trauma Leaves Biological Traces in Children – Rachel Yehuda
After the twin towers of the World Trade Center collapsed on September 11, 2001, in a haze of horror and smoke, clinicians at the Icahn School of Medicine at Mount Sinai in Manhattan offered to check anyone who’d been in the area for exposure to toxins. Among those who came in for evaluation were 187 pregnant women. Many were in shock, and a colleague asked if I could help diagnose and monitor them. They were at risk of developing post-traumatic stress disorder, or PTSD—experiencing flashbacks, nightmares, emotional numbness or other psychiatric symptoms for years afterward. And were the fetuses at risk?
My trauma research team quickly trained health professionals to evaluate and, if needed, treat the women. We monitored them through their pregnancies and beyond. When the babies were born, they were smaller than usual—the first sign that the trauma of the World Trade Center attack had reached the womb. Nine months later we examined 38 women and their infants when they came in for a wellness visit. Psychological evaluations revealed that many of the mothers had developed PTSD. And those with PTSD had unusually low levels of the stress-related hormone cortisol, a feature that researchers were coming to associate with the disorder.
Surprisingly and disturbingly, the saliva of the nine-month-old babies of the women with PTSD also showed low cortisol. The effect was most prominent in babies whose mothers had been in their third trimester on that fateful day. Just a year earlier a team I led had reported low cortisol levels in adult children of Holocaust survivors, but we’d assumed that it had something to do with being raised by parents who were suffering from the long-term emotional consequences of severe trauma. Now it looked like trauma leaves a trace in offspring even before they are born.
In the decades since, research by my group and others has confirmed that adverse experiences may influence the next generation through multiple pathways. The most apparent route runs through parental behavior, but influences during gestation and even changes in eggs and sperm may also play a role. And all these channels seem to involve epigenetics: alterations in the way that genes function. Epigenetics potentially explains why effects of trauma may endure long after the immediate threat is gone, and it is also implicated in the diverse pathways by which trauma is transmitted to future generations.
The implications of these findings may seem dire, suggesting that parental trauma predisposes offspring to be vulnerable to mental health conditions. But there is some evidence that the epigenetic response may serve as an adaptation that might help the children of traumatized parents cope with similar adversities. Or could both possible outcomes be true?..
…It is tempting to interpret epigenetic inheritance as a story of how trauma results in permanent damage. Epigenetic influences might nonetheless represent the body’s attempts to prepare offspring for challenges similar to those encountered by their parents. As circumstances change, however, the benefits conferred by such alterations may wane or even result in the emergence of novel vulnerabilities. Thus, the survival advantage of this form of intergenerational transmission depends in large part on the environment encountered by the offspring themselves.
Moreover, some of these stress-related and intergenerational changes may be reversible. Several years ago we discovered that combat veterans with PTSD who benefited from cognitive-behavioral psychotherapy showed treatment-induced changes in FKBP5 methylation. The finding confirmed that healing is also reflected in epigenetic change. And Dias and Ressler reconditioned their mice to lose their fear of cherry blossoms; the offspring conceived after this “treatment” did not have the cherry blossom epigenetic alteration, nor did they fear the scent. Preliminary as they are, such findings represent an important frontier in psychiatry and may suggest new avenues for treatment.
The hope is that as we learn more about the ways catastrophic experiences have shaped both those who lived through those horrors and their descendants, we will become better equipped to deal with dangers now and in the future, facing them with resolution and resilience.
Disclaimer: None of the information or analysis presented is intended to form the basis for any offer or recommendation. Of all the companies mentioned, we currently have a vested interest in Alphabet (parent of Google) and Wix. Holdings are subject to change at any time.
