At the Futurati Podcast we recently recorded an interview with Irene Ng on market design, modularity and legibility in systems of behavior, innovation, game theory, privacy, the blockchain, and a few other subjects.
Though it's far removed from the AI alignment and Bayesian epistemology that LWers normally discuss, I know that there's enough interest in economics and related disciplines for me to think it'd be worth posting the episode and the show notes from it.
- Irene began her journey as an entrepreneur in the 1990's at age of 26, where she successfully turned around a cruise ship company before selling it at 33.
- After this she enrolled in a PhD program to study economics, industrial design, and market design.
- 'Market design', which occupied much of our conversation, centers around designing systems of transactions which are safe and efficient.
- These can often be used by public sector entities as a way of handling problems when public funds don't do the trick.
- We spent some time discussing the connection to repugnant markets. An example might be a market for kidneys.
- On the one hand, we want to do whatever we can to shorten the time required to get someone a kidney. On the other, we don't want to incentivize people to steal each other's kidneys.
- Dr. Ng cites the work of other economists in successfully using market design to create markets which solve this problem without repugnant side effects.
- I take a step back and ask Dr. Ng what it even means to design a market. Most economists treat markets as spontaneous orders which emerge when humans engage in certain kinds of activity, so how can one be designed.
- Dr. Ng's answer centers around the idea of transactions, which can involve exchanges of money or favors or almost anything else.
- Transactions can be thick or thin, and in a well-designed market there will be thin transaction points where new resources can enter. This depends on the modularity of the system.
- An example might be two new parents. In the first month of baby's life every transaction is thick, direct, hands on. But if they break their routine into modules a think point may emerge where a nanny can enter the picture and help.
- I tie this back to the idea of legibility, with thick transactions being exactly the sort which can't be read by an outsider.
- Modularization makes systems more legible, and therefore easier to manage and optimize.
- I ask for clarity on the distinction between designing a market and centrally planning one. For all sorts of reason we know that command and control economic systems don't work, and Dr. Ng says that market design is about structuring the incentives and transactions within the system, not dictating how it works.
- We then turn to Dr. Ng's new venture, Dataswift, which focuses on privacy and data ownership.
- The basic idea behind Dataswift is setting up small, easily-operated private servers upon which an individual's data can live.
- This data can then be accessed or monetized in various ways at the individual's discretion.
- She likens the core technology to a digital passport which makes navigating different digital 'nations' easier.
- We then spend a great deal of time discussing the information diets of modern internet users, and we all agree that there should be a great deal more discretion exercised with respect to what we allow in.
- Given what Dataswift is trying to accomplish I asked why she didn't use blockchains, as this would seem to be an ideal use case for them.
- Her company actually does work with a lot of blockchain companies, and is one way of solving the problem of data storage and management.
- Thomas throws a classic Futurati Podcast curveball by asking Dr. Ng whether she thinks banks will exist in 20 years. She says to never doubt the resilience of markets, and that banks will likely evolve considerably.
- In the final leg of the conversation we talk about how people engage with the first iterations of new technologies by using the metaphors with which they're familiar. Citing Chris Dixon, I note that soon enough this gives way to entirely new modalities for using them, and someday soon NFTs, cryptoassets, and similar innovations will be as different from ledgers and trading cards as Wikipedia is from a static web page.