Updates #3
New wow ideas, notes, wiki entries, and AI apps
Hiiiiii,
While I haven’t written in a bit, pawel.world has been growing—not in leaps, but in slow, patient rings. I’m making some changes, so here’s a heap of excerpts, highlights, and glimpses from the last year or so (did you notice my new url?). Among other things, here’s a selection of new wow ideas, notes, wiki entries, and AI apps.
New in 💁 Wow ideas
Enlightenment feels like correctness, like every passing second feels correct.
– Aella via interview with Spencer Greenberg on Clearer Thinking podcast (28:29)
Why would I trust opinion bc its my own?
– Julia Galef via Rationally Speaking #143
You’re like the Catholics who are trying to get into heaven by being good enough
– Anon
New in ✍️ Notes
Overfitting or underfitting – as a useful self-calibration question?
Do you tend to overfit or underfit? It’s a valuable self-calibration question. Asking it myself I realized I have a habit of overfitting—I tend to apply a theory to a wide range of observations, sometimes too generously. I let the idea frame stretch too far, allowing it to absorb more than it probably should. If I have a theory, I’ll just keep shoving different inputs into it, without being critical enough about how specific or distinct their properties actually are.
Asking this question can also help calibrate thinking on any specific topic. For example if you try to teach me a concept of emergence
Overfitting is assuming all biological processes must involve emergence - like claiming cell division is emergent when it's actually a well-defined mechanical process. Treating any collective behavior as emergence (misses that emergence requires new properties not present in individual components)
Underfitting is assuming emergence only occurs in specific phenomena like consciousness emerging from neural connections. This view fails to recognize that emergence is a broader concept that can occur across many different types of systems
December 30, 2024
Canalization of psychopathological traits
I really like the metaphor of canalization—both as a potent concept and as a hypothesis for why all mental illnesses are correlated. That is, people with depression also tend to be correlated with higher levels of anxiety, psychosis, and attention problems. I pulled some excerpts from The Canal Papers | Astral Codex Ten to illustrate this.
First, let's understand gradient descent—an algorithm commonly used in machine learning. Picture an uneven terrain where a ball will naturally roll toward the nearest low point. As the ball moves downhill, it can change direction at any point. If you were guiding this ball, you'd consider: What's the optimal path downward? Which route provides the most efficient descent? And could small uphill movements lead to finding an even better low point? The same mechanism was used in this research
you can model the brain as an energy landscape with various peaks and valleys in some multidimensional space. Situations and stimuli plant “you” at some point on the landscape, and then you “roll down” towards some local minimum. If you’re the sort of person who repeats “I hate myself, I hate myself” in a lot of different situations, then you can think of the action of saying “I hate myself” as an attractor - a particularly steep, deep valley which it’s easy to fall into and hard to get out of. Many situations are close to the slopes of the “I hate myself” valley, so it’s easy to roll down and get caught there … What are examples of valleys other than saying “I hate myself”? The authors suggest habits.
or another examples
Walking normally is a valley; there’s a certain correct sequence of muscle movements, and you don’t want to start rotating your ligaments in some weird direction halfway through. … Given that you’ve started moving your leg to walk, you have a high prior (or an “extremely precise belief”) that you should bend your knee a certain way … a steep valley can also represent a very persistent belief - either very obvious beliefs like that the sky is blue, or very deeply-held beliefs like one’s religion. When a zealot person refuses to reconsider their religious beliefs, we can think of them being at the bottom of a very steep valley which is hard to move up. … What I call a trapped prior - a belief with such a strong gravity well that no evidence can shift you out of it - the authors call canalization, based on the metaphor of a canal having very steep walls and railroading you to a specific destination.
But how this connects to psychopathology?
You usually hear about general factors in the context of IQ. All intellectual tasks are correlated; people who are skilled at math also tend to be skilled at reading, or chess, or solving analogies … Recent research has suggested a similar “general factor of psychopathology”. All mental illnesses are correlated; people with depression also tend to have more anxiety, psychosis, attention problems, etc. As with intelligence, the statistical structure doesn’t look like a bunch of pairwise correlations, it looks like a single underlying cause. The authors suggest that deeper thing is canalization. If psychiatric conditions are learning mishaps that stick you in dysfunctional patterns, then maybe the tendency to canalize contributes to all of these problems… This doesn’t mean canalization is necessarily bad. Having habits/priors/tendencies is useful; without them you could never learn to edge-detect or walk or do anything at all. But go too far and you get . . . well, the authors suggest you get an increased tendency towards every psychiatric disease. … The paper does some good work suggesting a biological basis; canalization seems correlated with less synaptic growth and fewer dendritic spines. You can sort of see how this might make sense; if a “journey” through the mental “landscape” involves “traveling” from neuron to neuron, forcing potentials down a few big well-established connections is more canalized than having infinite different branches for any impulse to travel.
These two examples makes me think that with complex problems, it’s worth focusing more on how to elevate System’s potential gauge 🎨 — prioritizing a holistic approach rather than getting too caught up in the details.
December 28, 2024
Put your mind in a right wavelength as the key to good learning
The key to learning might not be about stacking as much new information as possible but about arriving at the right state of mind, a right “wavelength”?. That is, following the flow and learning at a pleasant rate, knowing there will be patches of boredom, not rushing, being calm and composed, feeling humble about the huge scope of knowledge I will not ever have.
August 7, 2024
Unconditional love to self via Anita
In this conversation Anita describes an interesting perspective on unconditional love that stayed with me. Listen yourself via this timestamped link
Here’s my reconstruction. Within myself, I feel the potential to be both the repulsive criminal and the most desirable saint. At any given moment, I perceive parts of myself as being somewhere on the spectrum between these extremes. Therefore I may experience unconditional love for myself by a) questioning why I would love myself differently based on the uncontrollable circumstances b) by feeling into other potential versions of me. In the video Anita describes this and a shift she experienced, feeling strong unconditional love for herself, love in an e even if she became a saint, there wouldn't be more of it.
May 25, 2024 – August 1, 2024
Review: Examples of barbell strategies
1 min review: Instead of doing things in the middle of distribution do things from both extremes: very bumpy, risky, extreme and then very flat, safe, mild. E.g. Instead of working full-time and doing mediocre side-project. Take 3 jobs for a half year and work full time on your project another half year.
April 23, 2024
Understanding addiction: the person had a gaping hole in their lives that (addictive activity) filled
I'm not sure if this is phrased better, or if I've just been exposed to the idea I've been pondering for a while, but when I read this, yeah, it's clear that's how addiction works:
„I’d like to think really long and hard about how popular gaming is with young boys all over the world, and how quick mainstream society has been to dismiss that as frivolous-at-best and soul-destroying at worst. I won’t pretend that it can’t be bad, I’ve personally seen friends go deep into the abyss of gaming-induced hell. But if you ask me, it’s never really the games themselves that are the root of the problem. The deeper issue is almost always that the person had a gaping hole in their lives that gaming filled. That’s an important difference. I think many addictions are like this. (Consider the rat park experiments, where rats in healthy social environments were found to consume less morphine than rats in isolated cages.)” via Are you having fun, son by Visa
March 14, 2024
New in 🌱 Wiki
Most of the text below are excerpts—click on links to expand. Ideas, definitions, or names authored by me are marked with 🎨.
Big changes are often beneficial
People may be excessively cautious when facing life-changing choices. … For important decisions (e.g. quitting a job or ending a relationship), Individuals who are told by the coin toss to make a change are much more likely to make a change and are happier six months later than those who were told by the coin to maintain the status quo
Focus on inputs, not outputs 🎨
Don't focus on outputs. Focus on inputs. And then score will take care of itself.
– Bill Walsh
Coordination and epistemic tools
That’s a complex document, click on the link :)
Reference class forecasting
Reference class forecasting (also called comparison class forecasting) is a method of predicting future outcomes by examining similar situations from the past.
Daniel Kahneman and Tversky found that human judgment is generally overoptimistic (See: Planning overconfidence). People often focus on the "inside view" – a perspective that concentrates on the specific details, plans, and circumstances of their situation. In contrast, the “outside view” (or reference class forecasting) involves comparing the current situation to a broader class of similar cases and using historical data to make more accurate predictions. This approach helps counter our tendency to ignore statistical and historical evidence.
Belief stuckness 🎨
A bias stemming from people being fixed in their point of view. Most conversations and exposure to new information rarely change people's minds. Philip Tetlock calls this cognitive conservatism—the tendency to maintain existing beliefs even when confronted with evidence that proves them wrong.
This concept relates to Belief as taste 🎨 and shares similar dynamics and causes with My-side bias cognitive bias family
Projection fallacies 🎨
A group of cognitive biases where one assumes that others experience the world in the same way as oneself:
Typical mind fallacy – assuming others experience, react to, and relate to the world in the same way as you do – e.g., assuming someone must be upset by public criticism because you would be upset in that situation.
Epistemological projection fallacy – assuming that things appear the same way to others as they do to you. Due to differences in minds, eyes, and other sensory organs, two people looking at the same thing may experience it differently – for example, a rose might appear completely different to someone with colorblindness
Semantic projection fallacy – assuming that others interpret words the same way you do. Example: you say "tree" thinking of a leafy oak while the other person pictures a palm tree
Four simulacra levels of discussion
Level 4 - Word Salad: The discussion is incoherent or merely plays with ideas, showing no concern for affiliation, influence or truth.
Level 3 - Signaling: The focus is on showing group allegiance and loyalty.
Level 2 - Influence: The goal is to affect your beliefs or actions, or make you do something.
Level 1 - Truth: The focus is on factual accuracy and objective reality.
Beliefs as tastes 🎨
A shortcut to understanding belief formation: We adopt beliefs from the people we love and trust. Like tastes, beliefs are something people possess that aren't governed by logic. They persist even without social pressure and remain unchanged when their underlying reasons are disproven.
Guess to understand 🎨
A learning technique that involves making guesses about how something works to deepen understanding. This approach ensures that concepts don't remain in an uncertain state but instead collapses into a best guess. It’s like a quantum particle that collapses from superposition into a specific state when observed.
Singulatrons 🎨
Singulatrons refer to a dynamic present in all known comprehending systems. Minds and processors start from a single perspective or point, then gradually add more data, points, and perspectives as they attempt to grasp larger and more complex concepts. However they are ultimately limited by their understanding capacity or finite computing power.
Admittedly, we're not exactly singular or starting at "zero"—scientists can team up to tackle complex problems, computers can be connected together, and a human brain has more synapses than an ant. However, this points to the fundamental dynamic that even with a team of 10 or 10,000, we're still start on a limited side, closer to 1 than infinity, with limited understanding capacity.
Families of cognitive biases
That’s a bit more complex page, click on the link :)
System’s potential gauge 🎨
The potential gauge points to a key relationship: when our overall human system has higher potential, on average we perform better at all tasks—even seemingly unrelated ones. For example: after 30 minutes of cardio exercise, your thinking becomes clearer.
Normalcy bias
People tend to assume things will remain unchanged, often underestimating the possibility of major changes or crises. This mindset creates an "illusion of permanence"—a belief that the present situation will continue indefinitely. It resembles cognitive conservatism, where people maintain their existing beliefs despite contradicting evidence.
Most commonly, normalcy bias is used in the context of catastrophes, but I use it more broadly as a tendency to believe that what's here and normal will persist.
When catastrophes occur, people often refuse to believe or accept what is happening. 70% 9/11 survivors first talk to other people before escaping, when Vesuvio blew up a lot of peeps from Pompeii were just watching.
Prediction horizon 🎨
The longest period over which predictions about a situation remain reliable, after which outcomes become random. Example of different prediction horizons:
finest meteorologists may not be able to predict what is gonna happen in a couple of hours
We no exactly where Pluto will be in relation to other planets in solar system very far into the future
deck of cards loses any information after 7 shuffles
Do things with love 🎨
The strongest people are ones who love lifting weights. The most stretched people are the ones who love stretching.
New in 💿 AI apps
🔮 Epistemic check
Evaluate any idea, claim, or fact with weighted pros, cons, and a total confidence score.
🧮 Parallel facts
Write a hypothetical scenario and I'll predict its potential outcomes with probabilities.