Back
Ben Kuhn - CFAR Workshop Review
webbenkuhn.net·benkuhn.net/cfar-workshop/
A 2013 personal review of CFAR's rationality workshop; tangentially relevant to AI safety insofar as CFAR trained many early AI safety researchers, but primarily a personal reflection on applied rationality techniques.
Metadata
Importance: 18/100blog postcommentary
Summary
Ben Kuhn's 2013 review of a Center for Applied Rationality (CFAR) workshop, evaluating whether their rationality-training techniques produce genuine self-improvement or fall into 'derpy self-improvement' patterns. He critically examines multiple hypotheses about why testimonials are positive and reports modest but concrete personal improvements.
Key Points
- •Author was skeptical CFAR would differ from ineffective 'derpy self-improvement' but attended after financial aid made expected value favorable
- •Four hypotheses considered: placebo effect, selection bias among LessWrong community, costly signaling/cognitive dissonance, or genuine effectiveness
- •Reports concrete improvements in email turnaround, to-do backlog, procrastination time, and sleep cycle post-workshop
- •Distinguishes CFAR from typical self-help by noting its grounding in probability, decision theory, and cognitive science research
- •Provides a rare critical yet positive outsized assessment of rationality training, useful for evaluating epistemic/reasoning skill interventions
Cited by 1 page
| Page | Type | Quality |
|---|---|---|
| Center for Applied Rationality | Organization | 62.0 |
Cached Content Preview
HTTP 200Fetched Mar 20, 202613 KB
July 2013
About a month ago I attended a [workshop](http://rationality.org/workshops/) held by the [Center For Applied Rationality](http://rationality.org/) (CFAR), and it was one of the best decisions I’ve made. Here’s a review of the workshop and what I took away from it.
CFAR’s workshops aim
> to give people more understanding and control of their own decision-making. The techniques we teach are inspired by models of reasoning from probability and decision theory, combined with cognitive science research on how human brains actually reason and how we can train ourselves to improve. We at CFAR turn those mathematical and empirical insights into everyday skills (like those described in our [rationality checklist](http://rationality.org/checklist/)) — how to make accurate predictions, how to avoid self-deception, and how to get your motivation where your arithmetic says it should be.
When deciding whether to attend, I wasn’t very confident that these techniques would work or be helpful. This is mostly because they sound suspiciously similar to the a particular archetype, common in various Internet places, which I’ll call _derpy self-improvement_. Like many things, derpy self-improvement is perhaps best characterized by the [relevant XKCD](http://xkcd.com/1027/) (panel 8):
> You look like you’re going to spend your life having one epiphany after another, always thinking you’ve finally figured out what’s holding you back, and how you can finally be productive and creative and turn your life around. But nothing will ever change. That cycle of mediocrity isn’t due to some obstacle. It’s who you _are_. The thing standing in the way of your dreams is that the person having them is _you_.
Derpy self-improvement techniques are mostly useless because derpy self-improvers aren’t very good at discriminating. Self-improvement tips are used as a constant drip of stimulus to make you feel like you’re always getting better, rather than because they’re actually useful. Everything is an epiphany and nothing works.
I was worried that the CFAR workshop would fall into the same pattern. But people’s [testimonials](http://rationality.org/testimonials/) about CFAR retreats didn’t really match this pattern; the written testimonials seemed reasoned, but still impressively glowing, and the _average_ response to “are you glad you came?” was 9.3/10. I was curious about why this was. I entertained four different hypotheses about the testimonials:
1. The workshop is a standard derpy self-improvement technique: really good at making people _feel_ like they’re getting better at things, but has no actual effect.
2. CFAR advertises mainly on Less Wrong and by word of mouth, and the people most likely to bite on the ads are those most tied to the Less Wrong community. Therefore they’re predisposed to like the retreat.
3. The retreat functions as a costly self-signalling gadget that shows you care about self-improvement. Once you spend all that money, cognitive dissonanc
... (truncated, 13 KB total)Resource ID:
6482a8886b52f3fd | Stable ID: MjkxZTA3OW