Incentives

Good Industrial Policy

In economics, firms are approximated as having one goal: maximise profit. Consumers are approximated as having one goal: maximising their utility. This turns out to be a very powerful way of looking at the markets in which they both operate.

Firms only care about themselves, and most only care about the short + medium term. And yet, firms have generated most of the consumable goods and services over human history, and are our engines of prosperity.

The fact that a firm only cares about itself in the short and medium term is however a perfect opportunity for external actors, who have power to set the rules they play be as well as a seperate, long-term agenda that they want to persue, to influence them through policy and incentives. Firms are simple things.

How Asia Works - book review Korea and export discipline China’s state directed investment policy over decades.

Hamilton’s protectionism

Joining a big competitive market (eu accession) which imports discipline: czechia/poland/slovakia/etc. saw large productivity gains after single-market entry via trade, fdi, and tougher competition/standards—basically export discipline without a heavy industrial-policy whip. [THIS COMES LATER AFTER DOMESTIC]

Micromanaging Motivation

Assume that people are a simple agent-like creature, with RL-agent-like behaviours.

In our model, we have reward and effort. a policy is learnt if it leads to more rewards being predicted per effort in. crucially, the need for reward being predicted runs through a couple of filters, due to the nature of predicting accurately: time discounting, discounting due to sparse reinforcement, and a small number of other discounts like this.

They actually also do time-discounting. And aditionally, adults are supposed to have learnt the association between long time horizon tasks and its rewards better, which children are sometimes unable to do.

Adults don’t JUST care about the short-term future, but they also have time discount rates. Also also have a variety of other failure modes that you can hit.

Given we are motivating adults, what are the things that are missing?

what am I missing? missing having learnt the utility of basic studying (no practical outcome difference gradient in the past. I was neurotically / autistically perfectionist, and that has worked up to last year.) missing having learnt useful methods of basic studying (the specific “task breaking down” and repeated drilling system I’m working on now)

paying kids to study (alpha school) -> never learnt to associate studying as reward -> solution: make an external reward system until i figure out it should be rewarding from the successes at drill-time -> never figureded out how to study -> make a system that drills me, semi-automatically, the less the unit cost to work it the better.

utility as log (incentives) reinforcement learning applied to people: immediate, specific rewards.

modelling self as a system 1 reinforcement learning agent.

Goal: I am NEVER FAILING AGAIN.

further: https://slatestarcodex.com/2015/01/31/the-parable-of-the-talents/ https://meltingasphalt.com/personality-the-body-in-society/