# Introduction
Several months ago, Minhaaj Rehman asked me questions on his podcast about a broad range of topics: data science, supply chain, economics (& wealth inequality), as well as the efficiency of the US education system. I have experience with & passion for each, but I am by no means an expert. Imposter syndrome is real. But it doesn’t mean that I can’t participate in these domains, it just means that I need to be structurally sound in the mental frameworks I use to opine on them. If I can reduce the inconsistencies from one domain to the next, I should be able to avoid making glaring mistakes and contribute meaningful ideas to the discussion.
Ideas, for me, typically stem from good questions (questions that keep you thinking). And Minhaaj has developed a keen sense for what makes a good question. Following the podcast, I thought a lot about the commonalities between those topics and I started to construct a mental framework to evaluate my prior beliefs. At the center of the framework stands the idea of preferential attachment which in the simplest terms means success generates more success. When I combine that perspective with competition, incentives, and probabilistic thinking, I believe it gives me (and you) solid ground for formulating opinions on data science, supply chain, economics, and the US education system.
# Building Blocks
Any single mental model is helpful. But the real power and applicability to complex real-world problems comes from combining multiple mental models. As noted above, here we want to combine the models of competition, preferential attachment, incentives, and probabilistic thinking. It could be more complex – I could layer in more models. But, the more complex my framework gets the less useful it becomes as an applied thinking tool. I want to upgrade from simple heuristics and anecdotal evidence, but I also want to declutter the noise that accumulates from too much information. Models are always meant to be an abstraction of reality. Imagine you had a map of the United States with a scale of 1 inch = 1 inch. It’s no longer useful. That’s why we keep these frameworks only as complex as they need to be in order to be useful and no more.
1. Competition
Competition allocates scarce resources better than any other economic system. Competition is the bedrock of the capitalist economic system. Despite the ambiguity of the quote’s origin, I still think there is wisdom when people say “Capitalism is the worst economic system, except for all the others.” There are unlimited sources out there that describe the advantages of competition. “Competition pushes individuals, firms and markets to make the best use of their resources, and to think outside the box to develop new ways of doing business and winning customers. This not only drives productivity up, it also improves our own standard of living.” There is a general recognition that competitive environments have massive benefits, but also shortcomings.
In 2018, Tom Piketty wrote a lengthy book detailing a historical perspective of wealth inequality in several countries; the United States being one of them. One of the recurrent themes is that in a capitalist economic system, capital clusters together (capital meaning land, business, financial assets). It accumulates into fewer organizations and fewer hands over time. This is another, shorter article from Pew Research looking at wealth consolidation.
If this is true – that capital clusters – why might it be true? Is capital accumulation and consolidation what I would expect from a competitive, capitalistic economy?
2. Preferential Attachment
Preferential attachment is a helpful mental model used to understand why capital clusters together. This simple simulation demonstrates the idea (visualized below and starter code here). Here’s how it works. In the first instance, an option is selected randomly. If an option is selected in this instance, the probability of being selected in the next instance increases. Over time, the options that were selected early – even randomly – get selected at a much higher rate than those that weren’t.
Preferential attachment feels right. There is something that intuitively as well as mathematically makes sense. Examples stand out all over the place.
1) In his book Freedom, Sebastian Junger writes that “boxers that win a fight have increased testosterone levels… giving them an advantage to win again.” A simpler way to say this: winners tend to win.
2) When container shipping began evolving and growing in the 1970’s, ports that grew the fastest won the most business… and because of their success were able to grow even bigger. Port towns that didn’t grow, died. The Port of Felixstowe became Britain’s largest container terminal (1967), almost overnight according to the description in the book The Box. The winners won repeatedly, the losers lost repeatedly. Network effects make these nodes important. Everyone else is a distraction. In other words, winners tend to win.
3) When you post on social media platforms, you receive some level of engagement; views, likes, comments, etc. The first hour is referred to as the Golden Hour by some because, at least according to this article, the engagement your content receives in that hour determines how “valuable” the content is and who else the platform’s algorithm decides to show it to. Content that does well in the first hour will have broader reach thereafter relative to content that doesn’t. Content that wins in the first hour, becomes content that is more likely to win in the ensuing hours.
At this point, the framework tells us that in unchecked competitive environments – especially environments that exist over time – resources are allocated efficiently, but power consolidates unevenly between the environment’s participants.
3. Incentive Structure
Recognition of economies of scale is a modern miracle. It describes the benefits resulting from winners winning. Costco & Walmart. Google & Facebook. Low prices, free tools. Hell yeah.
When an organization captures market share, they can produce at lower cost per unit. At lower cost per unit they can reduce their retail prices (sometimes to free!) which is good for the consumer. The organization gains more market share and the cycle continues. The incentives – nearly everyone’s incentive – is aligned towards large scale institutions.
But scale creates its own problems. It creates imbalance. Walmart has low prices for consumers. But Walmart also has significant control over the retail labor force. It has massive control over suppliers and partners. It has control over the US economy.
The saying usually takes a form similar to “If you owe the bank $100, that’s your problem. If you owe the bank $100 million, that’s the bank’s problem.” The favorability tips when there is imbalance.
Short run imbalance is ok, and probably even good if it promotes continuous improvement. Long run imbalance is bad. I can’t think of any model that shows that people that thrive with perpetual imbalance. Rapa Nui was imbalanced for years building wooden statues to preserve the energy of the natives after their death … and then they ate each other (note: I recognize that the archeological evidence for this is changing, but I still find it to be a good, albeit extreme, example of failing to find balance).
It seems reasonable to strive for long run balance and create systems that favor it over imbalance.
I want economies of scale to an extent. I want more variety in consumer goods. I want lower prices. I want our people to be more efficient with how we use what we have. However, when you put all your eggs in one basket, are you setting yourself up for disaster?
Do we want one major port where the bulk of your imports come into the country? It makes sense to reduce per unit costs for ocean shipping. But, what happens when it shuts down for an extended period because of a pandemic or labor strike?
Do we want one unified education system federally funded? Standardization of curriculum helps us ensure that all individuals are obtaining a baseline of knowledge. But, what happens if the curriculum isn’t diverse enough for future problems?
Do we want one tiny group of business leaders with a majority of a nation’s wealth? This is the reward for taking risks and working hard and innovating for the betterment of everyone in that nation. But, what happens if those individuals hoard it (or flood the economy with cash) creating financial instability either way? Or what if they chose to shut down great innovations (like the electric car back in the 80’s?). Again from Junger: freedom is a measure of how much control one group has over another group (or individual over individual) and freedom is a balance rather than an absolute.
If the incentives of a system lead to the destruction of part of the system, it will inevitably collapse. What I mean is that competition will eventually reduce competition. Alas, it is not clear if this is net good or net bad.
4. A Spectrum
Absolutes are an illusion. Matt Taddy gave me this idea: if you are boiling it down to a binary option (yes or no), you are thinking too simplistically. Everything lives on a spectrum. Good to bad. Poor to rich. Right to wrong. This is probabilistic thinking – we live in a world where the answer to the overwhelming majority of problems is “maybe.” Where something lands on the spectrum depends on at least two components. 1) empirical, quantified trade-offs (that are falsifiable) and 2) individual values (that are not falsifiable).
So, we can’t say “competition is good” or “competition is bad” and be done with it. Competition is good in context and up until a certain point. Competition works better than any other economic system for allocating scarce resources. But, the degree to which it is good will depend on values. One individual may believe that in an economic society, the single individual with the most wealth should own no more than 2x the average individual. Others may believe unlimited disparity is appropriate so long as the average level is rising. They might say it depends on “how hard you work.” Ah! That is the value component. Hard work means drastically different things for different people – dependent on their value system.
The values component distorts the perceived success of an outcome. Every decision has trade-offs. Poor understanding of those trade-offs makes it much harder to arrive at actionable insights. And ignoring the values component is short sighted. I think of it sort of like a prior belief (and perhaps a very strong prior belief). In Bayes statistics – it’s a starting point that needs to be proven or disproven with data. Hence, where we land on the spectrum will depend on the mixture of both falsifiable and non-falsifiable beliefs.
# Applying the Framework
We have the framework (competition -> preferential attachment -> incentives -> spectrum). The framework is useful and I feel comfortable using it as a starting point to think about the broad range of topics described in the introduction.
In data science, we should build teams of competitive A-players with these behaviors: grit, extreme ownership & insatiable curiosity. However, you should also build the reward structure on collaborative outcomes. Don’t set up the team so they are competing against each other for recognition and annual bonuses – have them compete against what “people” thought was possible. Remember, individuals have different values so hone in on what is important for them – some value time flexibility, some value public recognition, some value deep work, some value happy hours, etc. Use this to push the proverbial envelope.
In supply chain, you should treat the flow of your product like hedge funds and venture capital companies. Incur smaller costs over time to operate peripheral, perhaps redundant flow paths that hedge disruption risk. Resist – if possible – the short term incentive of undiversified scale. Continuously monitor and compare performance of similar flow paths – as partners rather than opponents. Over time, these smaller costs will net out by avoiding major disruptions and/or gaining market share during turbulent times. Remember, we can’t predict when or what the black swan event will be, but we do know that one will occur.
In economics (particularly thinking about wealth inequality), you should start with these tenets: consolidation of power has never been good and it sucks to have a significant fraction of your community with relatively nothing (food, clothing, shelter, etc.). Attempt to accept that success is 60% effort and 40% luck. Finally, recognize that competition yields efficiency, innovation, and overall prosperity.
With those tenets, there is no reason to tear down US institutions circa 2021. But there are reasons to rethink parts of it. In their book Good Economics for Bad Times, Banerjee and Duflo suggest several adjustments based on years of research; cap executive salaries, reinstitute a death tax, and establish a universal basic income. Some may say that these policies impede the winner’s freedom to keep what they’ve earned. To me, these policies are a step toward balancing freedom across the country.
In US education, competition for youth should only have one goal. Help them figure out what they are relatively good at and identify a community of people that will support you in the endeavor (especially through the failures). I continuously appreciate Scott Galoway’s perspective that you shouldn’t pursue your passion – you should pursue what you are good at which will lead to A) income, B) respect at work, and C) contributing to your community. It’s not for everyone, but it feels like a better default than “follow your passion.”
Again, from Good Economics for Bad Times, the authors discover that individuals with more experiences have a higher probability of monetary success. Our 21st century education system should be geared toward exploration. I think Project Based Learning schools set a good precedent and we should use competition to promote this type of diversity in our public education systems. Lastly, I think the ideas of incentivizing your young people with cash to travel or start a business are also worth experimenting with. Our youth hold values too, so we should do our best to create options for them to embrace those values, by doing so we take another step toward a more free & prosperous world.
# Summary
This entire framework really revolves around preferential attachment; the observation that winners tend to win. However, it’s not totally useful standing alone (it is only interesting). But if you connect it both upstream and downstream to partner models, you find actionable insights. Preferential attachment stems from competition and randomness upstream and its results can be reduced downstream with clever incentive mechanisms as well as keen awareness of trade-offs. The best way to increase the probability that you can create good solutions to new problems is to expand the thinking frameworks you have at your disposal. At the very least, it will help you avoid the most costly mistakes – the ones that will take you out of the game.