Sturgeon’s Law: Are You Not Entertained?
Sturgeon’s Law: Ninety percent of everything is crap
Theodore Sturgeon used this to defend science fiction which was derided as being a particularly low-quality genre, finding instead that most art everywhere was similarly low quality. Daniel Dennett uses this law as one of his intuition pumps, extending it from art to music, medicine, physics, social sciences, everything. I instinctively connected with the idea. I struggle with finding the line between being discerning and over-critical. It isn’t lost on me that my experience is qualitatively better when I am less critical and more open while consuming media, but I can’t categorically conclude from this that being critical is bad, assuming instead that there is an equilibrium between discernment and openness, which for me lies on the openness end of the spectrum. This assumption is not at all self-evident to me, it sounds intuitively correct, but not much else. So when Sturgeon’s Law came around, it made me feel better in 2 ways: a) since 90% of everything is crap, I’m justified erring on the side of discernment because my chances of a false positive are higher than a false negative. b) Now that I know 90% of everything is crap, I needn’t be emotionally invested in the crappy thing I’ve just watched, read, or heard, I can be critical yet not negative because it is in the scorpion’s nature to sting, the bedframe’s nature to fracture your big toe every single day, and the book’s nature to suck 90% of the time. So far so good.
Except now I feel worse. Because the villain of the frog and scorpion story is no longer the self-destructive spite of the scorpion, but the naive idiocy of the frog. Discernment and criticality aren’t on a 1D spectrum that makes for a simplistic equilibrium. It’s better visualized as a 3D set of axes
X: Discernment — This is a longterm variable, best represented by Precision. +ve values have high precision, I have pre-selected the optimal diet that maximizes true positives and minimizes false positives. My criticality and attachment, whatever their values be, have helped me continually improve discernment such that every piece of media has been data to finetune my algorithm for selection. Discernment equilibrium: as high as possible?
Y: Criticality — this is a point variable referring to the specific book/movie and how well I’m able to catch its pros and cons. I’m not sure how independent this variable is, and so where the equilibrium would lie. If discernment is low, I want criticality to be high enough that it generates more valuable data for discernment to increase, assuming my greatest gains in utility come from greater discernment. If discernment is high, then where do I want criticality to be? Does high criticality have to be inversely proportional with utility, or is that in the domain of attachment alone? If long term discernment goes up with high criticality, does it go down with low criticality? Can I achieve a target discernment and then discard all future data points as irrelevant, leaving my algorithm unchanged and being able to drop criticality way down? If this is indeed the case, then there is no real justification for criticality being its own axis, apart from the pointless elegance of Cartesian symmetry. I can just boil it down to discernment, unless these 2 can each move in opposite directions.
Z: Attachment — this has 2 components, the utility generated from the experience itself, which is ultimately the whole point of media consumption, as well as residual affect, which is restricted to its usefulness in guiding discernment more strongly than purely cognitive criticality. In an unrealistically perfect world, I’d be able to be strongly invested in positive emotion and strongly distant from negative emotion. But why should this not be possible? Our brains make real the things we pay attention to. If we focus on the positives, that’s all we see. I might inflate the positivity index of dogshit movies, which throws off discernment, sacrificing long-term value of utility for short term. The assumptions here are, a) Emotion is a stronger motivator for discernment than pure criticality and b) Attachment and Criticality cannot move in opposite directions. ‘Focus on the good’ sounds great but how do I decide what’s good without differentiating it from the bad? Everything I point out as good has intrinsically pointed out everything else as bad. Sure there are thousands of different types of good but here pluralism is a copout, it’s easy enough for me to say find the 1–2 things that this is specifically good at, but then I have other examples which are good at 1000s of things, and when it comes to discernment I’d much rather create a strong distinction between these 2 examples.
Let’s say I simplify the above problem and there are just 2 kinds of movies, 1 good and 0 bad. My task is not to maximize the 1s but maximize my steady state long term utility. It just so happens that this function is poorly understood by me and therefore I just use the proxy of maximizing 1s. This rests on a number of assumptions that Sturgeon’s Law makes a little more complicated.
- My average utility of 1s is greater than 0s: Not entirely obvious because with low criticality it is likely I will enjoy all movies the same.
- 1s have lower diminishing returns than 0s: Slightly more compelling. With high criticality, I can enjoy a good movie even more when I watch it the 2nd, 3rd time, each time unpacking details I’d missed. An important quality of a 0 movie is its lack of uniqueness and layers, both of which bode terribly for the long run. If I sacrifice discernment for enjoyment, I quickly run into decaying returns and then probably need to start my discernment exercise from scratch anyway having lost all that time
- Sturgeon’s Law isn’t blind: There’s a difference between saying 10 out of 100 balls are black, and 1 out of 10 containers contains only black balls. In A, discernment is irrelevant, in B, discernment is everything. I find the container and then can switch everything off except attachment and whatever criticality I need to elevate it further.
- Sturgeon’s Law is fractal: Does Sturgeon’s Law hold within Sturgeon’s Law? Let’s say I find a 10% author, let’s call him Jolkien. Are 90% of his books crap? That’s the strong version. In the weak version, the % of crappy books lies between 0 and 90. This % defines when I can switch off discernment and to what degree. 10 out of 100 balls are black. 10 containers each contain 10 balls. 1 container contains 5 black. 1 contains 3. 1 contains 2. 7 contain 0.
- Your tastes stay consistent over time
- 1s would continue to be 1s without the presence of a few 0s here or there
- Containers and balls retain their original color over time
- New added balls follow the same ratio as the existing containers and pot
- There are infinite balls, practically
I am reminded of Tyler Cowen’s restaurant problem. How to select a restaurant, or items from a menu, or a life partner? This depends on a static number in the overall population. If there are 100 restaurants in total, the ideal solution is to sample 33% of them and make no decisions at all, discarding all of them. Then you pick the next restaurant that’s the best one yet and then never leave. Perhaps we do the same thing with a matrix of preferences, like dialogue, philosophy, realism, art direction, plot complexity, acting quality, music, etc and cycle through the first randomly selected 33% of combinations with high criticality and low attachment/discernment, and then picking the next few combinations that are better on attachment than all the ones before, then sticking with that combination forever? How often to do this for account for all the assumptions above that are wrong, especially 5–8?