Epistemic status: speculative

Choice is bad. I want to focus on one aspect of this badness: regret. I’m going to argue that increased choice predictably increases the amount of regret that an agent feels, even if they are actually better off, and that this is bad for humans in particular.

I read Kristen Roupenian’s excellent short story ‘Cat Person’, and had a really strong reaction to it - so here are some thoughts.

First, if you haven’t already, go read the story. Done? Good.

Introduction

This post owes a great deal to prior work and thought by Spencer Greenberg, Eric Gastfriend, and Peter Hartree.

This post is a summary of the object-level thought on what makes a startup high impact which we developed while working on the Good Technology Project.

A lot of this material is more-or-less obvious applications of EA thought to startup theory. Nonetheless, it managed to be surprising and useful to people, so perhaps it is less obvious than it seems. I’ve condensed the presentation given the intended audience of this post - there is a lot more to say on many of these points. This material might have eventually developed into a “guide” to effective entrepreneurship.

In addition, some of the material relates to how to manage a startup in later stages. We never really got a chance to try that out, so it is especially speculative.

What makes a startup high impact?

We’re interested in startups because we think that they might be a mechanism by which we can have a large positive impact on the world. But what are the qualities that we should look for in a startup?

Introduction

This document aims to be two things: a summary of the things that we learned from the Good Technology Project (GTP), and a post-mortem of the project itself.

I’m going to simply state my beliefs in this post, but I should clarify beforehand that I am not very certain about these, they are my current best guesses.

What was GTP?

GTP started in late 2015 when Richard Batty and I met up for coffee in Oxford. We ended up talking about entrepreneurship: both Richard and I were working in software, and we believed that entrepreneurship could provide a route to leverage our skills. But work on the concrete problem of how to actually do that was frustratingly sparse.

A lot of ink has been spilled about how to get various decision algorithms to cooperate with each other. However, most approaches require the algorithm to provide some kind of information about itself to a potential cooperator.

Consider FDT, the hot new kid on the block. In order for FDT to cooperate with you, it needs to reason about how your actions are related to the algorithm that the FDT agent is running. The naive way of providing this information is simple: give the other agent a copy of your “source code”. Assuming that said code is analyzable without running into halting problem issues, this should allow FDT to work out whether it wants to cooperate with you.

Security needs of decision algorithms

However, just handing out your source code to anyone isn’t a great idea. This is because agents often want to pretend to have a different decision algorithm to the one that they really do.