The Principles System

This is my decision hub. When a problem comes up, I come back here for the answer.

It grows along two paths: principles harvested from handwritten reviews, and entries prompted by reading Principles. Anything new goes first into the candidate area at the end.

Hello, player. Welcome to the game called "life." Every problem you meet is a puzzle to crack, and each solved puzzle earns a gem — a principle — that will help you with the same problem next time. The more gems you collect, the better you decide. The game also moves you up to harder challenges, but don't worry: the gems you've gathered will help.

Enjoy it. 😊

1. Think independently and decide

(1) What do you want; (2) what is true; (3) given what is true, how do you get what you want.

2. Manage energy and time

Anything necessary: don't hesitate, do the first fifteen minutes.

Keep a "don't-do list": no sugared drinks; no phone in the fifteen minutes before sleep or after waking; no short video, other social media time-boxed to twenty minutes a day in total; never complain about the present; nothing deep-fried; eat to seventy or eighty percent full; never do a thing in inner haste — better not to do it; no useless socializing.

3. Declutter: leave room for thought and self

Clear and archive content regularly.

For objects, feel the relation between yourself and the thing: do I really need it? Will I really use it? For information, put social media on a diet; regularly empty the bookmarks and likes and fold them into reviews and ideas.

Leave enough blank time every week. That is where non-linear ideas are born.

4. Stay radically open-minded and humble; let others point out what I may have missed

Case (2026-07-27): I was sure a certain editor plus terminal was the optimum and looked down on graphical clients as "for non-developers." After actually trying one — conversations archived on the left, files on the right — it was very good.

Assuming you are right closes the information inlet in advance.

Always look for people better than yourself, and imitate and learn from them.

5. Embrace reality and deal with it

Run the goal-management system strictly; carry out each day's tasks and plans.

Pain + reflection = progress.

Reality is gray. The human factor, unwritten rules, the right amount of ambiguity and technique all matter.

Don't cling to your view of how things "should" be — it keeps you from seeing how they are. Case: I assumed Pilates followed by Baduanjin would be enough to shape up and build muscle; only after talking it through with someone did I see that the goal and the means were mismatched — keep the Baduanjin, but building muscle requires adding strength training.

6. Maintain and extend the network; tend the inner circle

Meet one person you respect for a conversation each week. Reach out actively: professors, older students. Talk repeatedly with the smartest people — that can include an AI.

7. Be a person of influence

More important than any grading system such as GPA is your influence. It shows indirectly as paper citations, repository stars, the reports and essays you have written.

8. Do the counter-instinctive, antifragile things

I keep a list of these and maintain it often.

9. The process for reaching a goal

The great majority of problems come from two points.

Planning: it must be concrete, down to what to do each day. Absurdly, after saying for so long that I wanted to work in a certain direction, I had never seriously researched how to prepare for that path.

Execution: separate yourself as designer from yourself as executor. As the executor, don't raise too many objections; trust yourself fully and go do it — the time will not be wasted.

Concrete planning over a stage (a month or two) firmly executed is a necessary condition of getting anything done. And there must be a tracking-and-feedback mechanism so you can see how long you have kept it up, which reduces the chance of quitting halfway.

Counterexample (an exam I prepared for): I made an overall plan at the start but never broke it down to daily actions; the result was good enough but not better. Three mistakes — the target was set too low (meeting the bar rather than the ceiling of my ability); not concrete enough (not broken down to "to reach this sub-score, how much must I practice each day"); no tracking (the check-in sheet lived somewhere I rarely opened). The lesson: put the tracking sheet somewhere conspicuous — by the desk or on the wall, not on the first page of a notebook.

10. Abilities for the AI age

Record ideas — your own, or ones seen in media — then talk them over with an AI, then execute.

Material that has not been reflected on and folded into your own body of knowledge is as good as unlearned.

Negotiate with the gray rules; cultivate your own scarcity; play infinite games.

11. Put delayed gratification first

I need to trust my first instinct about the thing — I know it is bad, that I should not do it, that it harms my more distant happiness — and not the various "reasons" the brain produces afterward.

12. Make the key life decisions well

Compound thinking: big results come from small daily compounding.

Look for choices with asymmetric payoff: the upside is unbounded, much of it beyond my current field of view.

Learn all the possibilities, consider the worst case, then find a way to eliminate what cannot be tolerated. Diversify risk.

---

Candidate principles: how this system grows

New entries do not go straight into the body. They enter the candidate area first, with their source and test criterion written out, and are promoted to a numbered principle only after being used and validated in real decisions.

A ledger records every promotion. The line under 5 above — "don't cling to your view of how things should be" — was candidate 1, promoted.

Below are a few still under test.

Candidate: eat the hardship others won't, and let it attach to a compounding asset

"Can't take it" is a contest of endurance — the most crowded moat there is, which is to say the rat race. "Won't take it" is a contest of judgment: the unglamorous, undignified work that clever people route around, but which compounds.

Choosing the hardship takes judgment; enduring it takes execution; judgment comes first.

Test sentence: is this hardship unpopular and compounding, or popular and non-compounding?

Candidate: don't fight the environment — spend willpower on designing it, not on resisting it

Don't trust human nature or willpower too much. Hand recurring daily situations to environmental design: defaults, removed temptations, physical routes. Willpower is scarce; reserve it for the exceptions that cannot be designed — interviews, negotiations, emergencies.

Boundary: environmental design is for spending willpower sparingly, not for declaring it useless. "The environment is bad" may not become an excuse for not executing.

(This one unifies the logic beneath several earlier entries: the don't-do list, decluttering, designer and executor, delayed gratification.)

Candidate: the attention layer holds one thing; the options layer may hold several — divide by timescale, not by camp

"Focus on one thing" and "keep several irons in the fire" are each right on their own timescale. From minutes to weeks it must be single-tasking: one top-level idea, no more than two things in progress. From months to years a few projects may rotate slowly, switched by the week, and only when stuck.

The failure mode is not having many things but many things occupying working memory at once.

The one-sentence criterion: does this need my continuous attention, or only my calendar? The first kind allows one; the second allows many, and they must be locked into a fixed administrative slot.

Boundary: "the options layer may hold several" is not a license for hoarding. Options-layer items are allowed to surface only in the weekly administrative hour.

Candidate: manage the career path by verifiable output, not by title

Company names and job titles are result labels; what is actually controllable is ability, work, network, and credible endorsement.

Rewrite "become such-and-such" as an observable change of state: can I independently carry research question → experiment design → engineering implementation → analysis of results and leave evidence a third party can verify? Update the route at each stage with real feedback.

Boundary: "output-oriented" may not be used to dodge the necessary mathematical foundations and long-term research training, nor to pile up low-quality projects because they are easy to count. The standard for evidence is — can it change the judgment of a strong researcher or a recruiter?