Prentiss Riddle: Time

aprendiz de todo, maestro de nada

Prentiss Riddle

home art austin books
causes chuckles garden
kids language movies
music time toys travel
Search this site

Archive by date
Archive by title

Sturgeon's law and Xeno

Over at Reenhead we were discussing Sturgeon's Law, that 90% of everything is crap.

What I've always wanted to know is, does Sturgeon's Law work recursively? Take a million jelly beans and remove the 900,000 that are crap. Are you left with 100,000 good jelly beans or are 90,000 of those crap as well? And so on.

time 2004.05.10 link


If 90% of all crap/non-crap evaluations are crap (which a rigorous application of the law would require), then it's the latter. If you remove the 900,000 you think are crap, you're left with 90,000 that are still crap, and 10,000 that are good. I think. Not long before you're likely to have thrown out all the good ones in the quest for just the good ones.

There's probably a really deep lesson to be learned here.

Adam Rice [adamrice cxe crossroads punkto net] • 2004.05.10
If you apply Sturgeon's to the non-crap 10% of some initial sample, you're just establishing a new upper crap-bound. Conversely, casting a wider net for your initial sample lowers the threshold of what's considered crap.

Put another way: applying Sturgeon's Law twice allows you to compute certain useful subsets, like the crème-de-la-crème, or the crap-de-la-crap. N applications will yield the crème-(de-la-crème)^(N-1) and so on.

[I just re-read what I wrote there. What a bunch of crap!]

Dan Sandler [dsandler cxe dsandler punkto org] • 2004.05.10
Actually, I think there are some slightly more serious lessons to be learned from this exercise. They have to do with "the paradox of choice", which is written about too well here for me to restate it.

Or put another way, they have to do with expectation management. If you're trying to please one person (yourself or one boss) with one jellybean, it's best to expand your jellybean pool by 9 inferior jellybeans. If you're trying to please two people in series (yourself and your boss, or your boss and your boss's boss), then you'll need to expand your jellybean pool by 99 inferior jellybeans. And so on.

Actually different people have different expected crap ratios (C) under different circumstances; C = 90% is just an estimate. Another expectation management technique is to manipulate C. Saying "I don't think I'll be able to find more than 1 good jellybean out of 100" and then delivering 10 good ones is gold.

I tried to find a way to turn this into E = MC^2 but I didn't try hard enough.

Prentiss Riddle [riddle cxe io punkto com] • 2004.05.10
More time >