Smania and Hidden Levers
What keeps solutions invisible?
Smania
Italian culture was a huge part of my upbringing.
Food,
History,
Traditions,
and Food, again.
One Italian word I was taught sticks out: Smania.
Smania means longing or craving, but it’s more specific than that:
It’s when you long for something, but you aren’t quite sure what that thing is.
The normal use of this was in the context of food.
Smania is useful, but paralyzing. It is all the want without any of the why.
So many feelings offer much more transparency: hunger tells me to eat and pain tells me to take my hand off the hot stove.
Smania tells me very little. It is opaque. So are fulfillment, boredom, and happiness.
Opacity does not necessarily correlate with solvability, but rather with the ability of an agent to discern the “levers” of a problem.
An opaque problem is not always difficult to solve, and conversely, a transparent problem is not always easy to solve.
Understanding the levers of a problem helps make the outcome more consistent. The more levers you understand, the more transparent a problem becomes.
As problems become more transparent, you have more control over the result, and outcome variance can decrease.
As problems become more opaque, you have less control over the result, and outcome variance can increase.
Transparent problems are easier to know how to solve, whether or not you have the means to solve them.
Hunger, as a signal, has a very clear lever that I could pull to solve the problem. I may or may not have food available, but my outcome variance is narrow because I have the necessary know-how to solve it.
Smania is a trickier signal. Not only is it unclear which levers are the right ones to pull, but you may not even be aware of the complete set of levers that are “pull-able”.
Maybe all I’m craving is pasta and tomato sauce, but the foggy desire of Smania sends me digging through the cabinets for hours.
The solution to the problem may be simple, but finding that solution through all the noise is not easy.
The opacity of problems is entirely subjective, or “agent-dependent”.
Learn to manage your Smania.
Legibility
When presented with opaque problems, grasping at random levers is often unsatisfying.
Your lack of understanding dampens your instrumentality.
One (flawed) approach can be to repeat attempts at lever-grabbing and hope to produce an explanation through induction.
This does not work.
The scientific approach starts with a hypothesis. What you observe is shaped by your views.
Your goal should be to make a problem less opaque, and more transparent.
You accomplish this by learning about the problem, studying its features, and acquiring more knowledge.
The rate you can acquire knowledge on a problem can be considered its “legibility”.
Legibility is both a subjective and objective property.
When software developers “borrow” lots of code from StackOverflow, they use objective legibility — the amount of coding resources available, but also subjective legibility — the developers have to understand the resources to use them.
Someone who is not as familiar or skilled with coding would find the vast coding resources incomprehensible.
To help yourself solve more problems:
Improve the way you tackle complex problems (framework)
Improve the rate you pick up new material (learning)
Improve the way you store and process information (note-taking)
And if you want to help others by improving objective legibility
Create material that helps others learn about and solve problems
Further distribute material that was helpful to you
Even with this, some problems are currently impossible to solve. Intractable.
Some contradictions in general relativity and quantum mechanics are impossible to resolve within our current understanding of each, despite both providing useful explanations on their own.
Due to this, it is well understood in physics that we need a new “theory of everything”.
While general relativity has helped us explain a lot, it breaks down when used on some of the most complex problems.
When used to understand black holes, general relativity comes in direct contradiction with quantum mechanics.
Despite this flaw, our current science has helped us accomplish amazing things, and there are still many things it can accomplish that don’t need a new “theory of everything”.
There are many opaque problems that are solvable using our current theories.
And just as well there are many opaque problems that are unsolvable using our current theories.
Opacity obfuscates this difference.
The inability to see the levers is the inability to understand whether the levers are out of reach or within reach.
Expanding humanity’s knowledge is an imperative:
It helps us solve problems
It helps us identify “unsolvable” problems
Currently unsolvable problems motivate us to increase our knowledge
Progress has brought humanity from caves to skyscrapers, from scraps to meals, from less to more.
Progress is created by people doing real work to create it, it does not happen accidentally.
Progress clarifies the levers that affect our universe, and we use those levers to build beautiful things.
Illegibility
Bad decisions can impact legibility.
Psychedelic and hallucinogenic drugs seem to house some exciting prospects for treating or assisting in the treatment of mental health problems.
Policy-wise, these substances are almost entirely written off and banned from study.
Our current level of knowledge on these drugs remains stagnant (we fear them) and no further exploration will allow us to push this frontier further.
If these banned drugs can create positive effects for mental health patients, then it seems as though our strict policies are directly contributing to the problem.
The current level of understanding is also not permanent, and while harder now, losing understanding and knowledge has occurred all through human history.
Many times over, libraries were burned, empires fell, and information was destroyed. These events move society backward, eliminating troves of knowledge once under humanity’s belt. Regressions like this can still happen today — the future is not guaranteed. All societies may have thought of themselves at the end of history, but they each fell anyway. It takes vigilant work by people today to avert cataclysms of Alexandrian proportion and maintain progress.
These events are shocks that damage the frontiers of our understanding.
They keep problems opaque.
Bad governance, and the bad philosophy that underpins it, hurts progress massively.
The more carelessly we act, the more illegible problems can become.
Meaning
When solving problems, we need to set the scope correctly.
We value being usefully capable, or instrumental.
Tackling big problems (such as societal advancement, climate change, poverty, and world hunger) can be a recipe for failure.
The ratio of your instrumentality to the problem needs to be sufficiently commensurate.
Youth nihilism is the acceptance that the impact you can have on these existential problems is inconsequential. It is maladaptive logic when you fail to properly adjust or scope your problems.
Instead of lamenting our inadequate skills, we can choose more appropriately adequate problems.
Good problem selection allows us to be instrumental. We don’t need to solve this scary problem, but if we can solve this sub-problem, we become useful.
So, find smaller problems within the confines of something meaningful, and continue to improve its legibility until it is solvable.
The ability to choose meaningful problems is part of agency.
I spend a lot of time thinking about how creators (entrepreneurs, artists, or others) choose problems to solve.
Most advice pushes for bottom-up approaches: user interviews, running into a problem yourself, and finding what people hate.
Why? You want to build something people want.
Yet, this can sometimes be a shifting sand on which to lay a foundation.
When GPT-3 was released, lots of people built off of it; then, when GPT-4 was released, their startup was completely steamrolled. Bottom-up approaches risk building features rather than companies.
This exists everywhere. If you fixate on a problem by bottom-up induction you will often find yourself displaced, enveloped by an existing platform, or stuck in a niche market.
Choosing problems for solving should be driven by bottom-up discovery and curiosity, but that can lead to fad-chasing and misalignment.
Introduce a timeless top-down lens. The AI startups that had a real mission and supported it with a competitive advantage in their field (data moat, distribution, etc) were more likely to be resilient.
Construction tech is one of the most interesting spaces to build in because of this. We need to be able to build big (and small) things quicker, cheaper, and easier. Sometimes we need to build in water, and eventually in space.
Far in the future, as a more advanced civilization, building things quickly, cheaply, and easily will still matter.
Building shallow plug-ins off of OpenAI products is not timeless; you are making a bet reliant on a level of progress not happening.
Finding a timeless space and then pursuing bottom-up targeting is more optimal.
Finding good problems is a problem. When faced with problems we should increase legibility by:
Finding areas with well-documented problems (objectively legible)
Getting better at identifying problems (subjectively legible).
Getting better at sufficiently scoping problems (subjectively legible)
Capacity
The level at which you can be instrumental in solving complex problems is your capacity.
Regardless of your capacity, opaque problems are still opaque, and therefore your control over the desired outcome is lessened.
The problems that we will one day face are unknown.
Therefore, the specific skills that we will need are impossible to discern in the present.
People who advise to pursue specific careers will therefore be wrong often. Choosing a career can be equivalent to choosing a category of problems to solve.
Luddites believed that technology would remove workers. People who hold millenarian views (the belief that the world will soon be destroyed by a powerful force) have similar feelings toward AI.
Just as the Industrial Revolution did not remove workers but instead allowed us to produce more things, AI will not displace all work, though it may automate annoying tasks.
The problem space always exists outside of our solution space. Whatever problems AI solves, are now solved. Whatever problems AI does not solve will become the new set of problems.
Since there are always problems, and therefore worthwhile work, at this boundary your goal should not be to become a specific profession based on current knowledge.
The meaningfulness of capacity is intimately related to the problems that we are facing. To give advice on what capacity is worth building is to make claims on what problems will be faced. This is flawed.
This applies to life as well, deciding on meaning is a problem for you to figure out.
In the process of living, we receive opaque signals: fulfillment, boredom, happiness, and Smania.
In the process of living, we face problems. Here’s what we have to do:
Make these problems more transparent; understand their levers.
Improve legibility; your ability to learn about their levers.
Minimize illegibility; that which prevents you from learning.
Decide what is meaningful to solve.
Increase your capacity in that meaningful direction.
Humans are good at finding problems, to live is to choose what problems to work on and then relentlessly pursue them.
So, the next time you feel Smania, that mysterious longing for something indiscernible, remember that you might just have the capacity to fulfill it.