mail

  1. body building, cardio, ergonomics, Veganism
  2. what would you be working on if you were me, What are your frustrations with me?
  3. Books to read, What are you up to, what are your future plans? How much time do you spend on DP? What did you do when you took your time off?
  4. How do you keep the grind? how do you stay motivated? Sometimes its boring, sometimes it is just too difficult, how do you deal with variances in feedback?
  5. What is this concrete thinking? what is your source? what is the point?
  6. Quitting job is a plan? What direction to go in career if ‘AI safety is life’? any thoughts?

new

Concrete thinking, career, Goal,

Health

  1. Why do you workout? why do you spend 12 hrs a week on this? How do you keep log? laptop or paper? What do you do when there is no improvement?

  2. Cardio what do you do? 20 mins a day? isn’t cardio boring? or hard? For example, I don’t want to cycle, treadmil is ok!

  3. Ergonomics: I don’t care about it at all

  4. Veganism, I am not sure what is my impact and what is my cost! It was a loose statement I made. I am not vegan… I eat paneer from hotels easily.

DP

  1. Why concrete thinking? Venezuala example, “hightech industries are showing high profits”

    That is better than them just reading the labels and nodding their head without checking it on any examples and thus not finding if their current models were wrong. (It will also help them remember it. Won’t go into details here; coming soon.)

  2. How many hours should I do? How to count DP hrs (days of deadline different from other days of DP). How did you work on your skill examples?

  3. How do I get immediate feedback while DPing? example!

  4. Seeing progress and feedback (discuss how it happens now and plans to amend?)

    Following all this, one can work on his technique but it very important to see progress and get feedback. Otherwise one will stop getting any better and stop caring about the end goal. Feedback is extremely crucial to feel motivated. Sometimes results might need interpretation, trusting your own judgment might not really be the best way to go about it. Is the blog post clear to a new reader? Will people get the main idea? Is it organized well? Paul Graham for example, appears to take this very seriously. Before he publishes a post, he runs atleast a few loops of getting feedback from his friends followed by correction before publishing it.

  5. What does an ideal agent look like? DP, persuation, data science! What is the path you envision for someone like an agent?

  6. How much do you grind everyday? how much should I?

  7. How do you stay motivated with grinding at max force for 4 hrs everyday?

  8. how do you allow persuation to not affect you, I “have to meet friends”, I “have to socialize a bit”?

  9. Measuring progress? what all do you measure? What all do you work on?

  10. Goals for each other? aims future?

Career

  1. How do I go about finding what I should be doing in career?

  2. Is going to US still important? cryonics etc…

  3. Have you subscribed to cryonics?

Why concrete thinking? from mail!

“High-tech industries” is completely vague. Zero examples. A Dutch person may not know what to think about when you say “high-tech”. (Maybe Dutch guys all speak English, in which case substitute some Indian villager.) You may try to explain using other labels: “it is an industry that deals with advanced technology and semiconductors and software”. He still doesn’t know what “software” is.

“Phone and tablets and laptops” are less vague because they are abstract examples. You’ve unpacked “high-tech” into constituent labels. But they are still labels! Again, the aforementioned, hypothetical, ignorant Dutch person won’t know what is a “tablet”.

“iPhone and iPad and Macbook” are more concrete because they are labels for which the above Dutch person can point to examples. Yes, it’s still not fully concrete because is it iPhone 5 or 6 or 17? Not mentioned. Is it a golden iPhone 6 or a white iPhone 6? In fact, is it the one that my friend has with serial number 76235 or the one your friend has with serial number 262093? Not mentioned. But that’s fine. We don’t need that level of detail. The difference in the details we imagine between “high-tech industry” and “iPhone” is much more than the difference in the details we imagine between “iPhone” and “iPhone 6”, at least for our current discussion. We would need more details (i.e., iPhone 6S with upgrade 7.62 and 3 GB RAM or whatever) if this were a product review article. (Imagine a product review of the latest iPhone by a guy like me who’s like “iPhone 7 or 8 or whatever - ah, it’s all the same thing” - that’s not what the consumers want. They want to know which phone will give them greater battery life when playing PUBG, which require concrete examples.)

Why does it matter if someone has a concrete example in their mind? Because, then, they can test your statements on at least one example. That is better than them just reading the labels and nodding their head without checking it on any examples and thus not finding if their current models were wrong. (It will also help them remember it. Won’t go into details here; coming soon.)

Suppose you say that “high-tech industries are becoming more and more powerful in today’s economy and have very high profits”. With “high-tech industries”, the Dutch guy may not know whether or not that’s true. He can’t test it. When you say “phone and laptops”, the Dutch guy may have a better idea of whether they are becoming more and more powerful because he sees phones and laptops around, but he still can’t test it precisely. When you say “iPhone”, he can test immediately that Apple is one of the biggest companies in the world and that it has high rates of profit. Yes, it’s not a highly empirical test, which would point to journal papers with exact percentage growth rates over time and comparisons to other industries, but it’s better than just accepting the words as such. In short, one test done right away is better than zero tests.

Why do we care about this dumbass, hypothetical Dutch guy who doesn’t understand English labels? Because when you’re writing, you are that dumbass, hypothetical Dutch guy. That means when you’re writing with just abstract labels or reading a book with abstract labels, you probably won’t think of a concrete example and you won’t be able to test the statements made. What then are your chances of writing correct statements or learning correct models?

So, when you use a label or see a label, make sure you have at least one example on which you can test statements about that label. How much detail should that example have? Enough to judge the statements.

Test: “People in Venezuela are desperate for food.” - vague or not? Your answer: ___. Why?

Can you **test** the claim that "they are desperate for food"? How? You probably don't know anybody from Venezuela. Now imagine being told: "People in Venezuela go through trash to eat rotting food." (https://www.youtube.com/watch?v=r2pjEYBhJSo) - can you now test the claim that they "are desperate for food"? You bet. If someone else says that "things in Venezuela are fine; it's just a little trouble here and there", you can say "hell no" because you have an example on which to test that statement. Nobody in their right mind would call eating trash "a little trouble". **That** is the power of one concrete example.