What We’re Reading (Week Ending 17 October 2021) - 17 Oct 2021
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 17 October 2021):
1. Nature Shows How This All Works – Morgan Housel
California has been devastated by wildfires for a decade. Back to back, year after year. Long-term droughts turned forests into dry tinder.
So everyone was elated when 2017 brought one of the wettest winters California had seen in recent memory. It was epic. Parts of Lake Tahoe received – I’m not making this up – more than 65 feet of snow in a few months. The six-year drought was declared over.
But the fires just got worse. The wettest year in memory was followed by “the deadliest and most destructive wildfire season on record.” And those two things were actually related.
Record rain in 2017 meant a surge of vegetation growth. It was called a super bloom, and it caused even desert towns to be covered in green.
That seemed great, but it had a hidden risk: A dry 2018 summer turned that record vegetation into a record amount of dry kindling to fuel new fires.
So record rain led to record fire.
2. The DMZ Partners Owners’ Manual – Soumil S. Zaveri
We will invest in a business only if we are willing to potentially own it for a decade. This is important to us for four reasons: 1) It ensures we focus on quality businesses whose fundamentals are likely to persist over time. As per Nassim Taleb’s advice, we will think long and hard about resilience in alternative future outcomes (say, in times of regulatory, economic or competitive stress). 2) We are not too excited by the prospect of getting rewarded on the basis of how an asset ought to be valued by catalysts in the medium term – nor do we want to deceive ourselves into believing that we have any expertise in being able to do so. Even if successful, such an investing style may deviate us from the prospect of compounding capital over decades by remaining patiently invested in exemplary companies. We would be deceiving ourselves in assuming that we can be better capital allocators than the people that run among the most outstanding companies we can find. 3) We are blessed with the privilege of patience – we intend to monetize it by identifying exceptional management teams, building franchises with immense scalability prospects over decades. This importantly allows us to partake in the “optionality value” that emanates from the bounty of unforeseeable surprises that accompany the actions of exceptional people. Experience has shown that it would be a folly to discount this phenomenon. Finally, 4) Taking an unusually long-view also gives us the advantage of an uncrowded spot as institutional imperatives often force professionals to check their relative performance scorecard every quarter, half-year or year. Our approach, if deployed well, is designed to help us deliver superior outcomes over a decade. Having an exemplary outcome over a decade is not the same as having ten exceptional one-year outcomes, much like a five-year-plan is not five good one-year plans.
We will never choose to own an asset solely based on valuation. No point bringing home junk for free – it still occupies valuable and limited space. Opportunity costs are very real, which we will remain acutely aware of. Gregory Mankiw wastes no time reminding microeconomics students that “the cost of something is what you give up to get it.” We take his advice to heart. To put it in practical terms, if you own a poorly governed, mediocre business solely because it is seemingly a mouth-watering bargain – time is effectively your enemy. The longer you must wait for your value to be realized, the greater the chances that the mediocre business faces setbacks or that inept management commits grave errors – in effect, permanently impairing your investment. We want time to be firmly on our side. In owning wonderful businesses run by exemplary people, time is an exceptionally potent tailwind!
3. DocuSign CEO Dan Springer offers surprising lessons learned from four years as a stay-at-home dad sandwiched between two IPOs – Byron Deeter and Dan Springer
Dan considers his role as a father to be his greatest career accomplishment. He believes that to be a great leader, you need a firm sense of what’s important in life—beyond just the confines of your office walls. What’s more, Dan says that the skills you develop as a caregiver will serve you greatly as a leader and mentor.
And he also points out that, while he’s gotten praise and media attention for his decision to pause his career for fatherhood, making a similar decision is considered wholly unremarkable for his female leader counterparts. “My feminist friends say if I were a woman, nobody would be asking me why I paused my career at its height. They say, ‘Well, you should have been staying home with your kids anyway, at least a little bit.’ As a dude sitting here, we should be aware we have these biases.”
Now Dan can’t unsee the double standards that plague career-oriented women. “Feedback is truly a gift,” he says. “So thank you to all the women and men in my life who are great feminists and have helped educate me in areas I was missing.”
Both his personal experience and burgeoning awareness of bias prompted him to do things differently as the leader of DocuSign. He created a parental leave program that guarantees all employees who become biological or adoptive parents six months of paid leave. “I hope everyone copies it. If you’re a founder out there, take away my competitive advantage. Offer this to your employees,” says Dan. “They will love you, and you’d be amazed how quick it’ll transform your culture.”
4. Slackers of the World, Unite! – Ellen Cushing
Eight years, more than 10 million users, and an acquisition bigger than the GDP of El Salvador later, Slack has managed to mostly hold on to the cachet of its early days. “All of the other messaging apps that we tested just felt sort of corporatey,” says Melanie Pinola, who wrote the Wirecutter review that declared Slack “by far” the best team-messaging app. “And the ones that were fun were really just imitations of Slack.” The user-experience researcher Michele Ronsen, who has done work for Slack and other global brands, told me that she’s seen no other product evoke such uniformly positive reactions among consumers. “When I recruit and conduct studies, over half of the people volunteer their love for the product and the platform and the benefits, completely unsolicited,” she said. “That does not happen very often.”
This is great for Slack, and also a little ridiculous: Enterprise software is meant to blend in, silently and only semi-effectively wringing more productivity out of us before we can call it a day. It is not supposed to create zealous brand loyalists. But Slack so thoroughly permeates companies’ culture that it changes them. It changes the language of the office and the texture of the workday. It enables a sui generis kind of communication, one that’s chatty, fast, stream-of-consciousness, and always on; one that often feels less like an email than a group text. It is work software that insinuated itself into our lives precisely by feeling unlike work software—and, in turn, it has made work feel less like work…
…On Slack, everyone has the same size megaphone, regardless of hierarchy or chain of command. And between the jokes and the special channels and the spontaneity and the freewheeling way of talking to your colleagues—who are also kind of your friends—it encourages a type of personal expression that is new to the American workplace.
A decade or two ago, identity formation, friendship, meaning-making, and political agitation were much more likely to be the things we did on nights and weekends. Now they’re central to work. If you’re an entry-level grunt, this might be thrilling. If you’re a boss, it can be scary. In August, Apple blocked employees from starting a Slack channel devoted to discussing pay equity, citing a policy that Slack activity “must advance the work, deliverables, or mission of Apple departments and teams.” (Channels about dad jokes, pets, and gaming were left alone.) In April, Basecamp, which makes software with a function similar to Slack’s, banned “societal and political” discussions on its own Basecamp account. And in 2018, employees at the luggage company Away were fired after creating an unsanctioned private Slack channel where employees—particularly those identifying as LGBTQ and people of color—talked freely about what they felt was an inhospitable work environment.
Slack’s inherent flatness means that anyone can emerge as a leader. In fact, the most influential person on Slack is almost never the boss, in part because in many organizations the more powerful you are, the less you use Slack. Being good at Slack is a skill, and it’s a different one from being well liked, or effective in meetings, or even good at your job. It’s more like being a social-media influencer. “People can amass power in the organization by being good at this tool,” Dash said. “They are not elevated by an institution; they just happen to have mastered a technology. And that is a thing that people can find threatening or find upsetting or that can be misused.”
5. Embracing Complexity – Tim Sullivan and Michael Mauboussin
A complex adaptive system has three characteristics. The first is that the system consists of a number of heterogeneous agents, and each of those agents makes decisions about how to behave. The most important dimension here is that those decisions will evolve over time. The second characteristic is that the agents interact with one another. That interaction leads to the third—something that scientists call emergence: In a very real way, the whole becomes greater than the sum of the parts. The key issue is that you can’t really understand the whole system by simply looking at its individual parts.
Can you give us a concrete example?
A canonical example of a complex adaptive system is an ant colony. Each individual ant has a decision role: Am I foraging? Am I doing midden work? Each one also interacts with the other ants. A lot of that is local interaction. What emerges from their behavior is an ant colony.
If you examine the colony on the colony level, forgetting about the individual ants, it appears to have the characteristics of an organism. It’s robust. It’s adaptive. It has a life cycle. But the individual ant is working with local information and local interaction. It has no sense of the global system. And you can’t understand the system by looking at the behavior of individual ants. That’s the essence of a complex adaptive system—and the thing that’s so vexing. Emergence disguises cause and effect. We don’t really know what’s going on.
Why is an ant colony the first example you think of?
Complex adaptive systems are one of nature’s big solutions, so biology is full of great examples. Ant colonies are solving very complicated, very challenging problems with no leadership, no strategic plan, no Congress.
Once you’re aware of how the structure works, though, you’ll see these systems everywhere—the city of Boston, the neurons in your brain, the cells in your immune system, the stock market. The basic features—heterogeneous agents, interaction, and an emergent global system—are consistent across domains.
Why should businesspeople pay attention?
So what could a biologist or an ant specialist or a honeybee specialist possibly tell us about running businesses? The answer is, a whole lot more than you might guess, if you are willing to make some connections. This to me is an essential way to think—especially in the 21st century.
Consider capital markets. Rather than looking at them through the rational-expectations model, or even using the no-arbitrage assumption—the idea that you won’t find any $100 bills on the sidewalk because somebody has already picked them up—you can look at them through a complex adaptive systems model, which empirically fits how the markets work. But complexity doesn’t lend itself to tidy mathematics in the way that some traditional, linear financial models do.
6. Kyle Samani – Solana: Faster, Cheaper, More Scalable – Patrick O’Shaughnessy and Kyle Samani
Patrick: [00:02:49] So Kyle, while this is going to be a breakdown on Solana specifically on which I think you’re one of the great experts and someone who can explain it in ways that I think everyone listening will understand. I think it’s probably necessary to take a step back from Solana and first frame, how you view the opportunity or the landscape in blockchain technology, generally speaking. And maybe the first question I’ll ask is what do you think the killer app of decentralized ledgers or blockchains is given that so many people think it’s Bitcoin, it’s this kind of new non-sovereign money. I think you have a different take. So maybe just frame the entire conversation by what the huge opportunity is here, and then we’ll get more specifically into Solana.
Kyle: [00:03:30] The history of the crypto-ecosystem, like all things is kind of pat dependent and there’s different cultural movements kind of that have bubbled up to the top of it at different points in time. Bitcoin was found in 2009. Satoshi, I don’t think had a strong view of what Bitcoin should be or what it could do, but he made something that was a breakthrough in a number of ways. Ethereum took a lot of those same ideas and just said, “Hey, just make it a little bit more programmable.” But there was no real plan for how to make it large scale. Certainly, again, even if you go look at Vitalik’s first introduction of Ethereum, which was in January 2014 at the Miami keynote, through that 17-minute video, and you can tell, he has no idea what this thing is useful for. He vaguely alludes to a couple of DeFi concepts in the video, but can’t coherently articulate what DeFi is or why it should matter.
And the important thing I would think we as an industry have learned really since the last probably five years or so is that the killer app for blockchains is DeFi. And I think you should probably interpret DeFi as broadly as possible. That means recreating existing financial contracts for trading spot assets, derivatives, options, interest rates, whatever, certainly in this kind of new paradigm where you get auditability and composability and its instant settlement. All those things are obvious. But I think the other implication of DeFi is then can you take financial concepts and inject them into new places that haven’t really traditionally had financial concepts in them? You’re just now starting to see this in a little bit with NFTs and people starting to play with fractionalizing NFTs. You look at this loot game thing that just came out a few weeks ago and you can see we’re kind of at the tip of the iceberg of lot of that stuff happening.
I think if you add social tokens kind of onto that and then combine NFTs and social tokens, this is a very ripe design space to do a lot of interesting new forms of capital formation, community engagement, create monetization, all those things. But again, all of these are still kind of finance centric concepts. So the conclusion I came to internally probably about a year ago was that what if you reframe the point of blockchains, not as non-sovereign money that happens to be programmable, which is what Ethereum launched at as, but what if you just reframe blockchains as the best conceivable DeFi system that happens to have non-sovereign monetary properties to it?
If you reframe the question that way, then the right design fundamentally probably is not something that looks like Bitcoin, but it’s something that is written from the ground up to really be finance native. And that probably means a few things. One, it means you need to have as low of latency as possible, because anything in finance that has derivatives means you have leverage. If you have leverage, you have risk of blowouts. And if you have risk of blowouts, you need to have low latency. You need to have high throughputs so that you can manage liquidations and risk in the system. The other thing that probably means is you want to have super high performance program languages and look at where all HFT is written at the bleeding edge of high performance realtime systems, and you want to be writing in those languages to just have optimized performance in every way. There’s some other implications as well, but those are broadly speaking, the two obvious ones.
Patrick: [00:06:37] Maybe just one click deeper on the notion of DeFi as the north star for crypto systems rather than not in sovereign money. Right? Like very, very, very big change from, I think what most people just starting to get familiar with this system would describe crypto as, they’d probably go straight to Bitcoin. But maybe this is the right time to compare sequentially. Actually, what is happening here? This is just a database and it’s just a record of who owns what, whether it’s Bitcoin or Solana or Eth or whatever. And there are really clever mechanisms for the world to all agree without any centralized authority on who owns what inside the ledger?
And transactions per second maybe is one interesting data point to talk about from Bitcoin to Eth, to something like Solana, given the frame that you just gave us. If all we’re trying to do is change the state of that underlying database or ledger, and maybe just tell the transactions per second story, starting with Bitcoin all the way through where we are today and why you think that’s interesting.
Kyle: [00:07:37] Bitcoin launched in 2009. Satoshi, I believe it was 2010. Some people were like spamming the Bitcoin network or something. And in order to prevent the system getting over flooded with too many messages bouncing between the computers, Satoshi just put in a very, very rudimentary fix, which is he just like added a few lines of code and said, “Blocks cannot be bigger than one megabyte.” Super arbitrary determination. He definitely didn’t consult with anyone publicly about it. My guess is he didn’t spend more than 10 seconds thinking about it and just put something in there with an expectation that he would change it later. Unfortunately, by putting that one megabyte cap in there that set a hard cap on the ceiling of Bitcoin at about seven or so transactions per second, maybe 10, somewhere in that range, I guess, is if he thought that that was going to persist in perpetuity, he probably wouldn’t have done that, but he did.
And then kind of as the culture of Bitcoin evolved over the next five to seven years, this really becomes apparent in the block size wars, which was 15 to 17 kind of timeframe. And ultimately the side that one was basically the side that said you can’t introduce the hard fork that breaks the rules of the system. And a hard fork would’ve meant changing that one limit to something else. And that camp kind of won in whatever Bitcoin is today. The only ways really to scale Bitcoin that have emerged are ways to compress data. So to fit more data into the same amount of space, which the SegWit thing did in 2017. And then the only other way is really like off chain transactions, meaning like lightning, which has not grown very effectively. People have been trying to operate within those constraints for last five, six years. And I’m extremely disappointed I think with the aggregate results of that.
Not to say there hasn’t been no gains, but it’s like a 3X gain in six years is by software standard pretty bad. Ethereum launched with the same basic proof of work model as Bitcoin for consensus. And then the programming model is pretty different actually. One of the big things, Ethereum people did not think too hard about that’s really creating a source of a lot of problems today is parallelism. In Ethereum, you have this basic problem of right. You’ve got all these people all over the world sending transactions to update the state of the system, right? Move money from point A to point B, do this trade, whatever it is. The vast majority of transactions that probably happen within a block, whether a block is milliseconds or even whether it’s 15 seconds, even probably whether it’s a minute, probably don’t have dependencies on one another.
So like a simple example would be if your account balance is zero and you want to send money to Bob, but I need to send money to you first, then there’s obviously some dependencies there for that to happen. So chronology does matter. But if you think about most things that happen in the world, at least within the context of 10 seconds, even a minute, you probably don’t have very many dependencies. You can just make the payment between people. So the unfortunate thing for Ethereum is the way that the Ethereum virtual machine is designed. They never really tried to deal with transaction parallelism. The challenge here, just in the kind of basic computer science problem terms is you have two people sending a transaction in the system. There’s a pretty high probability of those transactions don’t write to the same piece of the global state at the same time, but you have no 100% guarantee that they won’t.
So you need to make sure they don’t overlap with each other, because if they do, then you need to figure out which one to execute first. And this has been a basic problem in computer science for 30 or 40 years. Basically, the only solution is to know which parts of the state it’s going to touch before you execute it. And then if you see overlaps in what you’re going to touch, then you run them serially. Otherwise, you can run them in parallel. Pretty intuitive. Not too hard to reason about that in abstract terms. Implementing that in operating systems and such as just mechanically, a lot of work and Ethereums didn’t do it. And the EVM, which is the Ethereum Virtual Machine is written that way. And then all of the tooling around the EVM and all of the actual transactions today are all written with that assumption that there is no parallelism in the system.
So the EVM just runs everything serially. So your laptop today probably has four cores, maybe eight cores in it, and your graphics card is probably a thousand cores in it, maybe 4,000 cores. And you’re only taking advantage of one core because you’re just running everything serially. So Ethereum, when they launch, I think it was like call it 10 transactions per second or thereabouts, they’ve increased the gas limit a few times, which is kind of a very simple way of increasing the throughput. They’ve got it to call it 30 or so transactions per second by doing that, but there’s no been real major breakthroughs in the core system. Solana, if you look at all of the NextGen chains, people have tried to solve this problem. The only one that’s really attempting to do intra-shard parallelism is Solana. And this is, if you look at why it’s like, look at Anatoly’s background,. You did chip design at Qualcomm for a long time at high performance systems, at Dropbox and some other OS places.
His whole life, he spent saying, “How do I take an existing piece hardware and make it go as fast as possible?” That’s what he’s done for 20 years. And he looked at a modern computer and said, “Okay. How do I make network of computers all over the world that don’t trust each other to just go as fast as humanly possible.” The key to that unlock is parallelism. So Solana runs transactions natively on graphics cards, modern NVIDIA cards that I’d call have 4,000 cores. I think the next ones coming up have 8,000. You can obviously then run 8,000 jobs in parallel. The key obviously to be able to do that is each transaction header needs to specify what part of the global state it is going to touch. And so long as the header states that, then the system can line everything up and say, “Okay, these things aren’t going to interfere with each other. So run them all in parallel.”
And anything that has dependencies, you run serially. There’s some other approaches to thinking about parallelism that other teams have taken the most notable, which is sharding and Ethereum, Polkadot, Avalanche, NEAR, and perhaps others are all doing various… and Cosmos are all doing various forms of sharding. What sharding gives you is you get parallelism where you get one thread per shard. So you get parallelism in the sense that each shard gives you a new lane to move forward. If you need to communicate between the shards, there’s like a lot of latency, a lot of additional gas costs in doing so. So kind of the key questions I think about scaling these systems is can you scale a shard. If you can’t scale a single shard, how few shards can you get away with on a global scale to minimize all of the additional latency costs and gas costs that come from cross shard stuff. That’s kind of the basic framework of the thing. And today’s Solana runs at, I’d call that 50,000 transactions per second.
7. In depth: behind HNA’s fall, a web of nepotism from N.Y. to Hainan – Ji Tianqin, Yu Ning, Han Wei, and Denise Jia
Details of the alleged crimes committed by the two executives were not disclosed by the police, but Caixin’s yearlong investigation, including a review of the company’s filings and previous interviews with multiple former and current executives, found that HNA Chairman Chen Feng, now deceased co-founder Wang Jian and multiple senior executives owned companies controlled or invested in by family members that conducted business with HNA. These businesses, many registered in New York as well as Hainan, where the company headquarter was located, obtained funds and contracts from HNA ranging from aviation materials to real estate development, advertising and insurance. Some of those relatives even became frequent guests in New York’s philanthropy circle and leaders of Chinese businesses associations in the U.S.
None of the related-party transactions, some of which were related to the conglomerate’s overseas acquisitions, were fully disclosed in HNA’s regulatory filings.
Chen’s and Wang’s brothers were both involved in aviation material businesses that have supply contracts with HNA. HNA might have paid 30% to 50% more than competitors for aviation materials and 10% more for aircraft, a former HNA executive said.
“The more expensive, the more commission they could get,” the former executive said. “This is impossible at state-owned enterprises. Isn’t this embezzlement?”
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 DocuSign. Holdings are subject to change at any time.
