Snowflakes vs Lemmings

First principles or fast-follow?

My professional training and experience give me tremendous respect for the “wisdom of crowds”. In a prior post, Dinosaur Markets, I defend using the “outside view” as a surrogate for your own.

Obviously you don’t want to follow this logic right off a pixelated cliff. The tension is in knowing when the consensus is wrong.

You can fail in 2 ways.

Snowflake Error: contrarian when you shouldn’t be
Lemming Error: consensus when you shouldn’t be

If the consensus is correct most of the time, you’ll make Snowflake errors more often than Lemming errors. This is counterbalanced by the fact that Lemming errors are more costly.

When I hear “first principles” thinking my mind sees snowfall. Lots of snowflakes. Since I’m a default efficient market mindset, my first instinct is Chesterton’s Fence:

Don’t ever take a fence down until you know the reason why it was put up.

If you think something that exists is wrong, you need to be able to explain why the conditions for its existence are no longer valid. This idea is often invoked when people describe the difference between rigid conservatives and eager reformers. Well, ok. I suppose there will always be some people who put ketchup on steak.

The idea is more powerful than that. It’s more broadly about epistemic humility. When someone thinks they are “thinking from first principles”, my instinct is to ask “are you the first person to think from first principles on this problem?” It’s firsts all the way down. Chesterton’s fence is more generally about who has the burden of proof. A current example: Does the urbanist need to prove that cities will continue to be humans’ preferred means to self-organize or is it the dissenter’s duty to show otherwise?

In sum, nobody’s life rule is “copy what other people do”. But the opposite binary, “always think from first principles”, which is somehow more acceptable to say, is just as ridiculous. It’s almost like the phrase “reinvent the wheel” hired a PR firm.


The subject of when you should actually think from first principles vs listen to markets is obviously complicated. If you are interested in when to diverge from consensus, then check out the free book Inadequate Equilibria: Where and How Civilizations Get Stuck by Eliezer Yudkowsky. I have plugged it before and plan to re-read it soon enough. (Link)