Sanity Matrix

Dimension 1: sociability. Straightforward, permeable. Ask questions, remember people’s names. Politick. I get it.> Dimension 2: agreeableness. Get along? not really, but better to make friends than enemies. Not about the value as much as the implementation. Manage self-respect.> Dimension 3: neuroticism. Well.> Irresponsible to assign values to a metric that doesn’t apply. NULL != 0. But social subspaces don’t merge well. Easier to handle if you assign arbitrary values. “I’m very neurotic” okay, sure, I get you. Have to be consistent, though. Static value. If you let it slide they catch on, “you’re faking it, you’re bipolar, you’re weird” yes no hmm. Impossible to tell without peering into other people’s subjective experiences. Impossible to establish best practices. Fake it or suffer an incomplete JOIN?

If you’re reading this I feel sorry for you. I don’t think this is very penetrable writing, and I’m not sure it contains any worthwhile insight. I can’t really go back and clean it up — that would destroy what little meaning it might contain. But I can at least comment my code: add little sprinkles of syntax to keep you from feeling completely lost. Maybe the weaving in and out will make things easier, maybe it will make things worse. Feel free to rearrange the pieces as you see fit.

Low-dimensional conformity, high-dimensional hyperconformity. Need one or the other, not both, not neither. XOR. Both: passed over, unremarkable. Neither: sticking out, grating. “I’m not like other people” what are your operating assumptions? You break the law and follow the metalaw? You either eat fast food or you eat 2000 calories a day, not both, not neither.

Certain patterns aren’t visible on the individual level. Not explicitly talked about and no one person makes them happen, but step back, aggregate and they appear. “I would never do X, and neither would my friends, but together we do.” I know this because I watch. How did I redesign my mental process to analyze data clusters? If only I could remember. Most aggregators are extrinsic to the observer, they shouldn’t be the same process. Unexpected behavior may result.

Irresponsible to say “process” “subroutine” “function” — mental state is 4th dimension immutable. Many more than 4, truthfully, but those are the easiest to grasp. Metaphorical dimensions can still be measured, calculated, operated on. What separates them, mathematically, from physical? Nothing. No difference — you are high-dimensionality system.

So you are a hyperdimensional mind, but you see world in not even 4 but 3, sometimes only 2 dimensions. The natural result is poor predictions. Build a regressor on 3 metrics: it’s worthless. No wonder you never know what’s going on. Only using 10% of your brain? You don’t even use an infinitely thin sliver of it.

Dimensionality missing or unobservable ?— not your job to be omniscient. Work with what you have, you can do better than you are now.

If you see a friend, and that friend is sick, alone and in poverty, how do you approach the situation, mentally? Do you consider their immediate ailments (L0), or do you think about the things that happened in their life to bring them to that level (L1), or do you think about the root social problems that have contributed to inflicting this suffering on them (L2)? Multiple levels of analysis, multiple levels of thought. But you can meta-analyze as many times as you like, you’ll never add a new dimension. If you instead think “how can I best help them now,” (J0) that’s a line of reasoning totally orthogonal to what you were thinking about before. You’ve opened up a new dimension. You can’t make that jump by repeating the same sub-dimensional pattern on the L dimension, you have to think in a new direction. A solely L-dimensional mind is unable to have J-dimensional thoughts.

You recurse like this because you don’t feel your full dimensionality, and feel limited in your options. But you’re not limited, only atrophied. It’s a tiny amount, granted, but present. The current regressor that is your mind has very small weights on certain metrics. But it doesn’t have to stay that way. Guided, supervised learning is possible — retrain yourself, redefine. There’s no reason to stumble through life blind, like an actual machine (p-zombie). Physician, heal thyself!

Every mind a regressor, then. Some bad, some good, most fairly simple: we usually only think about 3 dimensions, after all. We are local maxima in a very turbulent hyperdimensional plane of Truth. Unsurprising fact: minds converge at these local maxima. Shared models, shared weights, shared predicted outcomes. In the same way that if you roll a million marbles down a hill, they’ll all congregate in the same two or three divots on the slope.

Social norms converge at local maximum. Everyone agrees more or less, and trains up their children in the way they should go. Suddenly it is not just one regressor— it is a family, a cluster, an established shared truth. Now try introducing an alien model, trained to a different maximum somewhere else. Its predictions contradict established social truth. Where the two disagree, wisdom of crowds prevails, every time.

This is insanity in a postmodern world. To say that an insane person is “wrong” is a specious claim, because there is no global “right” to compare them against. People have spoken before about “what if you were the only sane person in a world full of insane people” which is, if we were to take it at full face value, equally unfounded. But that idea really points at this: sanity is commonly shared behaviors, beliefs, reactions and predictions. To be sane is to respond to stimuli in a way that is socially appropriate. I don’t want to imply that there’s no such thing as mental dysfunction — that exists too, but it shares an umbrella with antisociality. Thus accusations of insanity as a political or legal weapon. Thus the malleability of the DSM. Thus the malleability of psychology in general.

Insanity is not merely inappropriate, but intolerable. Shrinks are engineers, trained to bring your model to back within acceptable tolerance limits. “Psychoengineer” is not just in fiction.

Ergo drugs. Not important: dysfunction, subjective experience, the actual existence of the local maximum you claim to have found. Important: brute-forcing your weight matrix to resemble the established consensus. Resocialization.

Why this priority? Consensus is comfort. It’s important to know how other people are going to behave. There’s no time to learn their nuances not in a world with billions of people, so the only practical solution is to assume they will behave just like you, and whenever someone doesn’t you call the police.

But model consensus is no more robust than a single model. Random forest: each regressor has to consider different parameters for their efforts to combine usefully. If every model looks the same then all but one of them are redundant. The wandering madman — юродивый — is a gift from God. A new perspective, a skew perspective, different subspace, different weights. Add to the consensus instead of rejecting. A diverse random forest will better approximate the truth than any single regressor alone.

Better approximations imply closer to truth than any one person can get alone. Does truth (Truth) exist independent of consensus? To restate the problem: can a bunch of people disagreeing with each other approach Truth better than a bunch of people agreeing with each other, or vice versa? If so, we do ourselves a tremendous disservice by discounting the insane and the anormal. Enforcing consistent behavior, putting as much effort as we do into socializing everyone, starting with 12 years of state-guided schooling which drills into children at their most formative age the mental model they ought to conform to — this is hurting our ability to reach for that objective truth.

Necessary sacrifice. Impractical and unsafe to live around a bunch of unsocialized minds. Too much effort would be wasted on cautious engagements with strangers. More safe assumptions about others = more free time to be spent on other pursuits. Socialization is efficiency.

I know. Just something worth considering.