The Problem and Opportunity of Scale
This post on Best Of A Great Lot is a part of a series on the subject of designing a new form of governance. Each piece aims to stand alone, but fits together on the Table of Contents.
On a day-to-day basis you can largely ignore the effects of your house's gravity. No satellites circle it to be avoided as you step out the door, and few people worry that their house will ignite into a ball of fusing plasma, or expand rapidly into a red giant and consume the neighbors.
Many aspects of society parallel this in an important way: when they are small, like houses, they behave the way we hope they will. However, when they scale (become extremely popular, become used in societally important ways, etc), like stars, they warp the space around them into something unrecognizable. Here are a few examples.
Scale Fragility
Google's PageRank algorithm, which originally brought them to success, was predicated on an internet where people link to other sites because they find them interesting. Reading these links as indications of interestingness worked really well for the first couple of years, but as Google rocketed to ubiquity, scammers decided it was worth hacking this algorithm. The first hacks were simple linkfarms, pages filled with links to sites that had paid to be more popular. Google fought back, updating their algorithm to check whether the links were coming from a page that had actual content on it. This initiated a cat-and-mouse game: scammers update their pages to match Google's expectations, so Google updates its algorithm to try to exclude the scam pages. The Verge captured some of the flavor of this history with their vignettes of SEO professionals. As of the last two years, the game has changed again with the introduction of LLMs, and Google’s results are a lot worse than they used to be.
In many ways, we can think of this as a fully general extension of the problem of advertising. At small scale advertising is literally telling your neighbors that you've started a bakery and inviting them to try out the bread you're going to sell. At slightly larger scales it's announcing your bakery in the newspaper so people know about it. At enormous scales it's the system we have, where everyone is constantly shouting at everyone else through every channel they can imagine about all the things one could buy.1 A reasonable person puts on the digital equivalent of raingear (adblockers) before exiting their house into the constant thunderstorm of the internet. Are we made better off by it? At small scale, yes, we find out about things that we wouldn't have known about. At large scale, we absorb enormous costs to our attention and cognitive function by either processing or attempting to ignore all the yelling.
A subtly different version of this is Airbnb. Airbnb started from an idea that seems eminently reasonable: many people have a bit of extra space they're not always using, which might be put to use by travelers looking for a more authentic, homelike stay than a hotel. In the beginning this was a delightful alternative to hotels. But the existence of Airbnb made it profitable for people to purchase houses and condos specifically to list there. In 2022, some outlets reported that Airbnb had more listings available in NYC than there were apartments available to rent. The NYC office of the Comptroller claims this has caused significant damage to the rental market in NYC.
People are endlessly inventive and capitalism funnels that inventiveness towards finding money any way that won't guarantee you a jail sentence (as well as some ways that do). Any major success in the market tends to collect people in its wake like a planet's asteroid companions. These asteroids include the full spectrum of capitalism: useful services that make the major success better; tragedies of the commons; and of course, scammers, spammers and fraudsters. Some successes tend toward more of one or the other category. Though this doesn’t happen only to infrastructure, it’s integral to the way that services become infrastructure through mass adoption.
Government-run infrastructure experiences this too. Medicare is famously difficult to navigate; there's a whole ecosystem of people doing everything from trying genuinely to help elderly people to scamming them.
The stock market's history encompasses all of the above. Adventurers going in search of trade and treasure need money to buy a ship and pay a crew, and originally, stock was the paper you used to prove you’d paid for their outfitting. Then came ongoing concerns that became the first corporations, storied ventures such as the Dutch East India Company and the Virginia Company. Once these companies existed longer than a single voyage the wealthy who had invested in them found it natural to trade these investments to each other when in need of cash. Eventually, these companies became enshrined in law and public trading of the ownership of them became normal. Today the stock market is so normalized that we still call it investment even though the public trading of large companies no longer resembles the original meaning of the term - putting money into risky ventures to ensure they happen. Instead, we think of investment as simply hoping for more money later, and have transformed what was once a risky business for the wealthy few into the core of the middle class's retirement plans.
Along with normalization came fraud and scam, of course. By the middle of the 19th century the stock market was notorious for it and we were inventing rules to attempt to protect the naive. There's a delightful lexicon of colorfully named fraudulent schemes that largely rely on the stock market for their existence, like pump and dump, poop and scoop, and corner. Ponzi schemes aren't purely a stock market phenomenon, but they're definitely enabled by them.
As we consider systems, we need to consider both how they will work at small scale, and what will break as they scale up. As I've written before, voting for representative is a process which doesn’t scale well. Opponents of ranked choice voting worry that it only works when people understand it and question whether most people can or will. Similarly, those who oppose a UBI are concerned that small-scale experiments don’t tell us what the ‘universal’ aspect will do. When you give everyone in one village a bunch of money, you're raising their relative wealth compared to the next village over. Many experiments in the western world have given money to a subset of people within a geographical location, raising them up above their peers. Money is frequently positional, such as in tight housing markets, where more money just means higher prices for everyone. Giving everyone a lot of money without increasing housing availability may just be handing that money to landlords, not at all what UBI proponents hope for.
Another dimension that policies must scale across is time. Econ bloggers frequently discuss how child tax credits throughout the western world have done little to increase fertility, without considering that some of these tax credits haven’t been in place for long. A credit that can reasonably be expected to last a long time will change more behavior than one that faces the potential to be revoked at any point. Similarly, UBI experiments face this challenge: people do different things with gift money than with a raise. Few people rent a more expensive apartment in response to an unusual windfall. Raises induce lifestyle creep because we expect them to continue. A temporary UBI experiment where the participants know or suspect that it is temporary is likely to be treated more like a one-time gift than a raise, but a real UBI program is more likely to be treated as a raise. One major concern many have with legislation by executive order, as has become common in the last few decades, is that it doesn’t provide the stability that legislation by Congress does, since executive orders are easily overturned when the other party wins the Presidency.
Scale Antifragility
Some things work better at large scale than small scale. Money and language are some: the more currencies and languages we have and the fewer people who use any one of them, the less useful they are. We lose something beautiful that humans have invented whenever a language disappears, but those who switch to using a common language gain an enormous network to communicate with.
Markets work better at what they’re good at (ignoring externalities, which is perhaps what they’re best at) with larger scale. Economists talk about thick markets (lots of participants) and thin markets (few), and thin markets are much more likely to be “mispriced.”
As we evaluate governance technologies and policies, we should look for those that are better at scale. We should remember that not hundreds of thousands, but millions of people will be using and abusing them if we ever come to rely on them as real parts of our system.
If you find my work interesting, please share and subscribe. It helps tremendously.