Saturday, May 30, 2015

Against Being For Or Against Tell Culture

Ever since I posted about Tell Culture a year ago, people have been debating whether direct or indirect communication is better.

(One day I will learn to frame my important points in a way that is controversial enough to popularize them.)

I find this frustrating.

The concept of "communication cultures" is a kind of cognitive first aid. It's better to have tourniquettes than to not have tourniquettes, because otherwise people bleed to death. But there's a lot more to medicine than first aid, and a tourniquette will never reattach a severed arm.

What kind of cognitive first aid is "communication cultures"? What does it prevent people from dying of before they make it to the hospital?

Harmful misunderstandings can happen when people from one communication culture interact with people from another communication culture without recognizing that the other group employs different assumptions, and relies on a different skillset. That's the thing knowing about "communication cultures" saves you from.

But that's first aid, no more or less. If "Tell Culture vs. Guess Culture" is all you know about communication and you want to communicate effectively, you're alive, but you're a long damn way from "healthy".

There are skills, techniques, and insights you have to gain before you can communicate well, in a way that satisfies the values of everyone involved.

To master communication, you can't just be like, "I prefer Tell Culture, which is better than Guess Culture, so my disabilities in Guess Culture are therefore justified." Justified shmustified, you're still missing an arm.

To reattach a limb, you need lots of medical knowledge and advanced surgical skills. To master communication, you have to actually learn things that empower you to communicate.

My advice to you - my request of you, even - if you find yourself fueling these debates, is to (for the love of god) move on. If you've already applied cognitive first aid, you've created an affordance for further advancement. Using even more tourniquettes doesn't help.

A better use of your resources would be identifying what you most want to do with communication, and what factors are most important for accomplishing that. What is the next thing you need to learn in order to get what you want out of communication? What is the most important problem in the art of communication, and what can you do to solve it?

If you're comfortable with direct communication, it may be that what you need most right now is one of the central Guess Techniques. Basic empathy, maybe. Go talk to someone who thrives in Guess Culture, and instead of picking a fight, try to learn something.

Saturday, May 16, 2015

Tortoise Report 3: Empathy

What's a "Tortoise Report"? See the Tortoise Skills page.

Habit: Empathy

Duration: 2 Months

Success: 9/10

Trigger: Curiosity directed at another person

Action: Pushing the curiosity through their external presentation to penetrate their internal experience

Result: I enjoy socializing with people I like. (!!!) I'm giving this 9/10 because of how surprising and useful the success is, not because I feel like I've almost mastered the skill.

ETA 6/30/15: I went to a party last night where I saw someone who hadn't interacted with me in person for a year. He said I seemed a lot different socially, and in particular he described me as "warm". That's definitely a first.

ETA 2/18/2016: I retained the ability to empathize deliberately, but as my motivation to practice a new skill simply because it was new skill faded, I exercised that ability less and less frequently. By November I was back to avoiding socialization completely. In the past month, I've gained a supporting skill (or property or something) that makes me feel motivated to empathize pretty much all the time: I've gained human connection as a terminal value. I feel motivated to understand people merely because I want to understand them. It feels like a direct urge, not a deliberately calculated tactic. As a result, it's pretty easy to initiate and maintain empathy. I'll write more about this later, but it seemed important to note that a second piece was needed.

A dialogue, in which I teach this skill to myself from a year ago.

In addition to being a Tortoise Report, this is an experiment in gaining pedagogical content knowledge by imagining the most efficient way to bestow a skill I now possess on my pre-skilled self.

Potentially Interesting Person: No, I'm in the Bay Area now. I'm working through Code Academy. Been volunteering for CFAR and sleeping on Tilia's couch.

Me: thinks: Ugh, why am I at this party. I hate this hate this hate this. How do other people enjoy socializing???

Me: So you're planning to Earn To Give?

PIP: Yeah, that's the plan.

Me: Cool.

PIP: What about you? I follow you on Facebook but like I don't know what you actually do with most of your time.

Me: Suppresses a sigh. Oh, this and that. I do some stuff for CFAR myself, and I just got back from giving a series of mnemonics workshops in the Midwest.

PIP: Right, I remember you posting about that! How'd those go?

PIP freezes in the middle of a hand gesture. The party around us goes quiet. Standing beside PIP is another version of me, slightly plumper and with longer hair.

Future Me: Hi! I'm you from a year in the future, and I'm going to teach you about empathy.

Me: "Empathy?" That sounds boring.

Future Me: It isn't.

Everything plays backward briefly.

PIP: Right, I remember you posting about that! How'd those go?

Me: The first one went exceedingly well. I was shocked, actually. The second one was so-so, though people seemed pretty happy with it.

PIP: Cool! What did you teach?


Future Me: How are you feeling right now?

Me: Trapped, bored, tired, like I'm wasting my time. I'd rather be at home reading. Why do I even go to these parties?

Future Me: If I recall correctly, you think that since you no longer have social anxiety, you should be participating in socialization like ordinary humans.

Me: That does sound sort of silly doesn't it. Just because I'm not terrified doesn't mean I'm actually benefiting from this. Maybe I should just stop going to parties.

Future Me: That might be wise. But what do you think would happen if you declined literally every invitation to any kind of socialization, even coffee, that you expected would make you feel bored and tired?

Me: ...Well then I guess I just wouldn't interact in person with anybody but Eliezer.

Future Me: Indeed. Why do you think you feel this way in social interactions so frequently?

Me: looking a bit sad and helpless I guess I probably just... don't like people. I mean, I know I like People, as an abstract category, at least sometimes. My whole life is about making sure People continue to exist for a very long time. But whatever it is that makes other people enjoy in-person socialization, I just don't have it.

Future Me: You think that you never care about individual people in physical proximity, that you're not a compassionate person, right? Like you have some long-range compassion, or at least long-range aesthetic appreciation for humanity, but no short-range compassion, no empathy.

Me: That sounds about right, yeah.

Future Me: So first of all, I want to point out a problem with your conception of self.

Me: I don't really have much of one of those anymore. I mean I know I used to, I used to have a solid story about who I was and I didn't think the central features of it could change much. Now I think my properties are fluid and dynamic.

Future Me: Yes, that's much better than before. But you say that your properties are "fluid and dynamic". I know you think you're bad at visualization, but you don't realize yet how much skill you've gained in that area recently. Use the same mental motion you'd use to solidify an association, and tell me how you visualize your properties being fluid and dynamic.

Me: Hm. I'm seeing this picture of the inside of my head, and it's full of cloud puddles of different colors all swirling around making whooshing sounds When I learn something, a new one flows in through my skull, and sometimes one of them leaves.

Future Me: So when mnemonics gave you the ability to visualize like that, a new skill flowed into your head and you became "someone who can visualize things".

Me: Sure.

Future Me: Doesn't that sound awfully essentialist to you?

Me: ...No?

Future Me: No?

Me: Well I'm just a collection of abilities and aversions and goals and a bunch of other things, and things in that mix can change at any time. If I were essentialist, I'd think I had a single solid soul-like thing that was Who I Really Am, and I think that's bullshit.

Future Me: I see. What I meant is that you're an essentialist about personality characteristics, not about personalities.

Me: Huh. Ok, I'm listening.

Future Me: Imagine yourself as an algorithm in a neural network. (Have you read those parts of the Sequences yet? I think you probably have.) Can you picture that?

Me: Sure. There's a series of orange marble-like nodes with lots of wispy connections to other nodes. I use this image a lot. I am a brain, after all.

Future Me: Oh right, you're all about "association networks" for mnemonics. I never realized that's where this understanding came from. Ok, so tell me about association networks. What happens when you think of a "horse"?

Me: A group of nodes representing "horse" fires, and the things that are highly connected to the "horse" cluster are likely to fire. Strongly connected ones, such as "hung like a horse", are more likely to fire than weakly connected ones, such as "horseradish".

Future Me: Suppose your ability to do the Shim Sham dance steps is a result of having strengthened the right series of connections in your association network.

Me: startled pause of realization I've been thinking of "associations" as limited to "experiences".

Future Me: But experiences affect behavior, don't they. And most of the things that happen in your brain don't make it to the conscious level. Your association network is firing all the time -

Me: And when I do the Shim Sham, I don't have to think about it anymore because I've strengthened the connections through repetition so thoroughly that the right series of nodes fires in the right order without my attention ever focusing on that process. The only thing that makes it to my attention is a feeling of effortlessness and satisfaction as I move.

Future Me: I'll give you a minute to process the implications.


Me: That's what a skill looks like in a neural network isn't it.

Future Me: Uh huh.

Me: And not just a physical skill whose output is movement. Mental skills are the same way. They just output further cognitive processes instead.

Future Me: Yep.

Me: This is all very interesting, but what does it have to do with my boring conversation with PIP and my relationship with empathy?

Future Me: Empathy is a collection of skills.

Me: You're going to have to spell it out more than that.

Future Me: Highly empathetic people have developed very strong pathways through their minds that input certain social experiences and output certain emotions. Or perhaps they were born with brains organized like that. Contrast this to your "swirling soup of personality traits" model.

Me: Your version involves a manipulable causal mechanism. It suggests I can identify and strengthen key associations and end up with new patterns of thought and behavior without gaining "empathy" all at once as an essential personality trait. Just like when I memorize stuff.

Future Me: You can indeed. Let me tell you an interesting fact about my current self. In the past week, I've had three one-on-one conversations. They happened on different days. Two of them were with people you've tried and failed to interact with productively before. All of them were over an hour long, and one lasted an entire afternoon. I agreed to these meetings because they seemed like fun. In the moment, they were fun. At no point did I feel any of the things you're currently feeling toward PIP. No boredom, no exhaustion, no trapped feelings, no annoyance, no wishing I were somewhere else. Instead I mostly felt a wide range of positive emotions. That thing that causes other people to want to interact with each other on purpose? There's a very good chance I felt that.

Me: That's... astonishing. It's hard to believe that happened to me. I honestly cannot imagine what that might have been like.

Future Me: Haha, you can, actually. You just don't have a lot of practice imagining being other people from the inside. Even future versions of yourself.

Me: And you say this startling change is caused by strengthening neural pathways comprising empathy?

Future Me: Yep.

Me: How? I'm not seeing that.

The sound of the party returns, and everything plays backward again for a moment.

PIP: Cool! What did you teach?

Me: I taught people how to deliberately strengthen associations between ideas. If you want to remember "pen" and "orange juice", you explain to System 1 that pens and orange juice have a relationship it considers extremely important.

PIP: You mean like a funny story involving a pen and some orange juice?

Me: That would work. "Funny" and "story" are both things System 1 pays attention to. There are a lot of other things it pays attention to as well, like hedonic rewards, rhyme and rhythm, and strong emotions of almost any kind. Primed with the techniques "funny" and "story", I'm currently imagining a puddle of orange juice and a pen trying to walk to the store together. Though the pen's just hopping right along, the orange juice is struggling to keep up because every time it manages to gather itself into a solid shape, it collapses back onto the pavement with wet squishy sounds. The pen is exasperated by this inefficiency and is trying to drag the orange juice along, but there's not really anything to hold on to and it doesn't like the smell of oranges. The orange juice feels guilty that it can't keep up with the pen. And then... well, stuff like that.

PIP: laughing I'm definitely going to remember "orange juice" and "pen" now! I don't know why I'd want to but I bet if you see me in a year and ask me what goes with "orange juice" I will know.

Everything freezes again.

Future Me: I would like to point out that you just experienced your model of the puddle of orange juice so vividly that you became uncomfortable and ended the story.

Me: ...Maaaaybe.

Future Me: You totally did. I was there. You felt it struggling to pull itself together and you felt a whole complex bundle of unpleasant emotions when the pen started yelling at it.

Me: Yeah ok fine.

Future Me: How do you think PIP is feeling right now?

Me: Um, I'm not sure. I hadn't considered it.

Future Me: Yeah I know. That's ok. Consider it.

Me: I guess maybe he's feeling sort of happy, and probably nervous to be talking to me.

Future Me: Whatever you did to generate that answer, did the process involve feeling any of those emotions? Or even simulating them as more than abstract ideas?

Me: I... don't think so. They just seem like the most likely things given the context and his body language and so on.

Future Me: Correct me if I'm wrong. (I'm not.) You pretty much see everybody as mindless walking bags of meat.

Me: Well no I don't actually think that -

Future Me: I know you don't think it. What you explicitly think is that their internal experiences are as rich and interesting as your own. But your models of them are not nearly as rich in emotion or sensory experience as your model of the orange juice or the pen.

Me: That makes me sound like a terrible person. Not that it's inaccurate.

Future Me: Whether you're a terrible person is irrelevant. You're thinking in essentialist terms about personality traits again. Tell me what is fun about walking bags of meat.

Me: Nothing. Bags of meat are boring.

Future Me: That they are. System 1 does not find walking bags of meat the least bit important.

Me: You're saying that if I could use the same processes to model conversation partners that I use to model orange juice, then people would suddenly feel a lot more interesting.

Future Me: It seems to be working pretty damn well for me so far.

Me: I see. I'm... not sure I know how to do that.

Future Me: I'm sure you don't. It'll take about two months of practice.

Me: jawfloor Two months is not a long time.

Future Me: Well you'll have developed the method first, and that part will take a lot longer. But since this is a fictional reenactment of something that never happened, I'm going to cheat and get you started right now. The first step is curiosity.

Me: That seems to be the case for an awful lot of things.

Future Me: Verily. I'm still not quite sure what that's about, but I'm pretty sure it's important. Anyway, you don't know it yet, but you spend a fair amount of time being curious about people.

Me: If I don't know it, how do you?

Future Me: The next thing you need to learn is how to identify high-leverage intervention points in your default cognitive patterns. Places in a string of firings where you can intervene to strengthen one crucial connection over another. I identified "curiosity about other people" as a high-leverage intervention point, and started paying attention to sensations of curiosity about other people.

Me: How can you pay attention to something that's below the level of conscious awareness?

Future Me: You can't. But you can build bridges that draw it up toward conscious awareness. For example, close your eyes, and tell me the color of the paint on the walls.

Me: I'm... not sure. Blue maybe?

Future Me: Now open your eyes and look around.

Me: They're pink. I definitely saw them but their color failed to register. Your point is that... No, I'm not quite sure what your point is.

Future Me: My point is that if I hadn't said anything, you'd have gone through the whole party seeing the color of the walls and not being aware of it. But I prompted you to search for your experience of the color of the walls in future consciousness. I flagged that experience as important, and flagging things as important, as something to watch for, does two things. I'm not quite sure how it does them. But it both leads your attention to the experiences you're searching for, and it probably causes you to have more of them (you'll spend more time looking at the walls when you're trying to find out what color they are).

Me: I can become aware of things that usually remain beneath the level of conscious awareness just by paying attention to them.

Future Me: Exactly. And you've just dismissed a thought about observation bias because you know it's not actually relevant. Good job. Now look around the room at all of these people, and when you notice anything even vaguely like a feeling of curiosity about any of them, tap your fingers together.

Me: Looks around. Sees someone who looks familiar. Taps fingers.

Future Me: What happened right before you tapped your fingers?

Me: I wondered whether I'd seen that girl over there in person before, or if she'd just commented on my Facebook wall at some point. But I think I only wondered it because you primed me to be curious about people.

Future Me: That will often be the case: You'll only notice the thing you're watching for when you're explicitly thinking about the fact that you're watching for it. You might only do the thing you're watching for because you're thinking about it. That's ok. It still strengthens the relevant pathways. Surprisingly quickly, in my experience. Empathy took a comparatively long time. Other cognitive habits I've learned by this method took a couple weeks or even days.

Me: So I just... tap my fingers whenever I'm curious about somebody?

Future Me: Yep. For now. I started with trying to tap my fingers whenever I noticed that I was experiencing empathy, but this turned out to be a prerequisite. You'll eventually realize that you've been trying to answer your curiosity about people by making System 2 inferences about their external appearance and behaviors, and that you get a lot farther a lot faster if you activate System 1 the way you do while making mnemonic associations and then just let it do the thing it evolved to do.

Me: Wouldn't that give me a lot of wrong answers? I mean, people are complicated. If I project a bunch of rich experiences into their heads I'm going to be mostly wrong most of the time.

Future Me: Yes, but you're not going to forget that while you're doing it. Also it's not exactly the same as "projecting a lot of rich experiences". There are extremely important sensations of "possibility" and "uncertainty". You'll see. And you're also going to be able to update your models way more quickly.

Me: looks incredulous

Future Me: That shouldn't surprise you; you'll be sticking your neck out and making riskier predictions. Furthermore, you really did evolve to model the minds of other humans accurately enough to predict incredibly complex behaviors. Those instinctive tools are a hell of a lot more powerful and accurate than you currently realize. Do you remember the second time you took Val's Againstness class, the one where you didn't have social anxiety?

Me: Yeah, that was super weird. He had somebody stand in front of the group and asked them to do something uncomfortable, like sing "happy birthday", while relaxing. Then he just watched them and told them all about their internal experiences based on ridiculously subtle external cues of anxiety or concern or whatever. And I predicted everything he said before he said it.

Future Me: Yep. You were doing the thing. You let System 1 be in charge of modeling the other person.

Me: Why do you think I was able to do that?

Future Me: Probably you were hypnotized.

Me: Heh, well Val was in the room, so yeah probably.

Future Me: Anyway, do the curiosity thing first. Tap your fingers when you're curious about other people. I'll get back to you.

sounds start again

PIP: laughing I'm definitely going to remember "orange juice" and "pen" now! I don't know why I'd want to but I bet if you see me in a year and ask me what goes with "orange juice" I will know.

Me: It's not unlikely. It depends a lot on how strong your emotional response to the story was. But I suppose - taps fingers - this is your first big party in the Bay Area right? First time meeting a bunch of people you've wanted to meet for a long time? You're probably already in a context where your emotions will solidify a lot of memories. ...Hey, snaps fingers, can I ask you something sort of personal?

PIP: Sure!

Me: What are you feeling right now?

PIP: Um... sort of nervous, excited, extremely happy, slightly drunk, worried about how you're judging me, concerned I won't display enough introspective skill in this answer to gain your respect, a few other things that aren't as obvious right now.

Me: Heh. I love it that you can answer that so transparently. Rationalists are great.

PIP: I've read some of your things about Tell Culture. I consider you one of my allies, so I want to be as transparent to you as possible.

Me: That's... really incredibly touching, actually. Thank you.

PIP: smiles

scene freezes

Me: That was... different.

Future Me: How was it different?

Me: When I noticed I was curious about him, I found that I cared more about his answer, that I paid more attention to what it meant.

Future Me: To get the next part I had to practice just this "noticing curiosity about other people" thing for a month. But if I can selectively freeze time, then I can also accelerate your learning process. Let's put your brain in a state where it's mastered noticing curiosity about people, and pick up from there.

a small popping sound

Me: Woah...

scene unfreezes

PIP: smiles

Me: taps fingers Why do you want to save the world?

PIP: That's an interesting question!

scene freezes

Future Me: Tell me what happened just before you snapped your fingers.

Me: I wondered why he wanted to save the world.

Future Me: What was it like to wonder why he wanted to save the world? What was happening in your head that you translated into the words "I wondered why he wanted to save the world"?

Me: I imagined him doing tasks for CFAR, I imagined him sitting in front of a computer coding and then donating money to FHI, I imagined him smiling while doing these things, and I felt an emotional sensation I want to call a "question mark".

Future Me: What is it like to know why I want to save the world?

Me: If you haven't changed too much, you have feelings of joy and fulfillment when you imagine tiling the universe with flourishing sentience, and you have feelings of loss and despair when you imagine none of that sentience coming to exist.

Future Me: When you thought about PIP wanting to "save the world", you had a visual simulation of his body going about world-saving tasks, and you imagined a smiling facial expression. You then thought about me wanting to "save the world". What's the crucial difference?

Me: I imagined you from the inside, and him from the outside!

Future Me: Yes.

Me: With him, it's like I'm interacting with his mechanical interface. I think that even my model of him is actually mostly of his mechanical interface. When I wonder about him, when I feel curiosity, it doesn't penetrate his exterior. taps fingers

Future Me: What was that tap for?

Me: I just wondered what it's like to be PIP from the inside. What it's like to be the thing that produces all of those behaviors and experiences all of those situations. What internal experiences motivate the activities of the mechanical interface. I wondered who is inside the machine.

Future Me: What's it like, the kind of curiosity that penetrates the exterior?

Me: The question mark sensation is there again, but I also have this almost spacial sensation where my attention is located about where his head is, and it's flipped to look out from that perspective at the world. There's a feeling of... not of specific emotions, but of something like the possibility of emotions, and the possibility of other kinds of experiences. And now I'm automatically starting to try to answer the question of what it's like to be him, and I'm filling in the "possibilities" with specific emotions and experiences, and I'm experiencing those things as I do it in just the way that I experience my memory palace when I walk around in it. It's all actually there in my head.

Future Me: Congratulations! You are empathizing.

Me: Really?

Future Me: I think so. You're doing the thing that I do when I interact with people and feel the human connection that makes interactions worth having. Call it whatever. You're imagining him as a person, instead of a walking sack of meat.

Me: It's a little bit difficult and uncomfortable.

Future Me: It takes practice. ... Though not for you, I guess.

another small popping sound

Me: ...I know Kung Fu.

Future Me: grins Show me.

Me: So you want me to start tapping my fingers when I notice I'm empathizing?

Future Me: Yep.

Me: I can do that.

Future Me: How would you feel about it if I left you alone at this party to talk to some random rationalist you've never met before?

Me: I'd... I'd like that, actually. I really want to know more about him. More about what's inside his head, I mean. It would be fun to try to learn. looks around I'd probably like to talk with any of these people, really.

Future Me: That's what I thought. I'll leave you to it then.

scene unfreezes

Wednesday, May 6, 2015

The Simulation Calibration Formulation

It's a fairly common practice among rationalists I'm familiar with to train epistemic calibration by making predictions, quantifying their confidence in those predictions, tracking their success rates, and betting actual money with each other along the way. The idea is that with repeated practice, you'll eventually get a gut sense that "sensation x", which you used to associate with "80% certainty", actually occurs more reliably in the presence of 60% probability predictions than 80% probability predictions. That is, 60% of the time when you feel "80% certain", the prediction comes true, and this eventually cause that to sensation to feel like 60% certainty instead of 80% certainty.

I'm going to call this approach "the observation correlation calibration formulation" (because I can). [The Credence Game] ( allows rapid-fire execution of the the observation correlation calibration formulation.

But there's a second approach to epistemic calibration that I don't hear people talk about so much, and I think at this point in my development, it's more valuable to me.

From Luke's summary of How To Measure Anything:

"Suppose you’re asked to give a 90% CI for the year in which Newton published the universal laws of gravitation, and you can win $1,000 in one of two ways:

1) You win $1,000 if the true year of publication falls within your 90% CI. Otherwise, you win nothing.

2) You spin a dial divided into two “pie slices,” one covering 10% of the dial, and the other covering 90%. If the dial lands on the small slice, you win nothing. If it lands on the big slice, you win $1,000.

If you find yourself preferring option #2, then you must think spinning the dial has a higher chance of winning you $1,000 than option #1. That suggest your stated 90% CI isn’t really your 90% CI. Maybe it’s your 65% CI or your 80% CI instead. By preferring option #2, your brain is trying to tell you that your originally stated 90% CI is overconfident.

If instead you find yourself preferring option #1, then you must think there is more than a 90% chance your stated 90% CI contains the true value. By preferring option #1, your brain is trying to tell you that your original 90% CI is under confident."

I call that the Simulation Calibration Formulation, and I think it's brilliant. Especially the part about how to identify underconfidence. It's relatively easy to humbly admit your overconfidence, but dropping your credence after that by exactly the right amount is hard.

I haven't tested this, but I expect I'd gain skill more quickly through a rapid-fire Simulation session than through a rapid-fire Observation Correlation session. You can also do a calibration simulation in any real-life instance where you might otherwise make a bet.

I think the Observation Correlation method assumes either that you already have pretty good reflective awareness of your credence-related subjective experiences, or more likely that reflective awareness of those experiences isn't all that important. Especially in the online-training version of Observation Correlation, improvement is expected to happen below the level of awareness. It's a quiet shifting of gut feelings.

I think reflective awareness of credence experiences is probably hugely beneficial. The simulation method trains exactly that, making it a good candidate for something earlier in a calibration training program than the observation method.

The other reason I suspect it should come before observation is that it isn't tied up with social feelings like wanting to protect your reputation or social stigmas surrounding gambling, or personal insecurities related to intelligence and ego. In the moments of real-world prediction and prediction-checking, any of those sensations is likely to be so salient that it blots out credence feelings both at and below conscious awareness. And when you turn out to be wrong, you'll probably be punished (in a behavioral psychology sense) for making a prediction in the first place, if you're not already very skilled.

If I'm right about these things, then it would be wise to practice Simulation Calibration until the mental movements of balancing overconfidence and underconfidence are fast, easy, and nearly automatic, and to do that before you get really serious about Prediction Book or similar things. At that point, you'll be armed with sharper phenomenological weapons to cut through counterproductive ego preservation/20th century science virtue ethics of skepticism, and you'll actually be able to hear your "80% confidence" feeling ringing clear above the noise. You'll know what you're listening for, and you'll store the feelings in memory for later comparison.

You can practice this offline using the Credence Game I mentioned before, performing the simulation for each question, and not keeping score. When that gets easy, pay attention to the score again. And when that's easy, stop doing the simulation.

I don't mean that you should stop making real-world predictions if calibration simulation isn't easy yet. I just mean that early on, Simulation should be the focus of your epistemic calibration training, rather than Observation. I'm certainly going to make it the focus of mine.