Notes.
Working thoughts, fragments, propositions, and things I’m still figuring out.
The questions I keep returning to
in my work never change: How do we perceive what matters, discover the relationships that govern it, decide what ought to exist, give that judgment a working form, and recognize when reality proves us wrong?
When to articulate. When to scan.
Some things are easier to describe than to find. Others are easier to recognize than to describe. Reordering groceries may take one sentence. Choosing a chair may require seeing twenty. Reviewing an agent's changes may require a visual comparison of what existed before and what exists now. The mistake is to assume that conversation should replace the graphical interface. Conversation reduces the effort of navigation
The Art of Seeing
Someone glances toward the door. We notice the glance. We infer impatience. We may be wrong. The art of seeing is learning to distinguish what we observe from what we make of it. We cannot know reality directly, but we can improve our ability to notice what matters, interpret what we notice, and recognize when we are mistaken. We must become observationists. Four abilities we can cultivate Attention — What am I notic
Try to see inevitability before revelation.
What is hard to see coming, then impossible to see any other way. How we see changes what we make. What we make changes what there is to see.
Notions are cheap. Insights are scarce. A coherent concept is hard-won.
Product judgment is the learned ability to make good decisions under ambiguity.
Product judgment is the learned ability to make good decisions under ambiguity about what software should do, for whom, how simply it can do it, and whether the resulting artifact is actually good. In the agentic era, that matters even more. When AI agents handle more of the implementation, the scarce skill is no…
High-value questions to ask when scoping which opportunities and desired outcomes to prioritize
Is this a proven behavior or a new one? How large is the opportunity, and at what rate is it growing? What are the baseline outcomes today? What assumptions are we making? Does this problem/opportunity recur? What…
Opportunity Judgment Loop
1. Sense — notice a signal in the world. Look for anomalies, workarounds/misfits, unmet needs, changing behavior, or new capabilities that make previously impossible things possible. Treat need-signals and capability-signals differently: a new capability is an opportunity hypothesis, not…
Intent–Discover–Align–Execute Method
Begin with intent, not commands. Explain the outcome you want, your current understanding, and the relevant context. Ask the gen AI model/LLM interface to challenge the framing, surface assumptions and unknowns, identify blind spots, and explore alternative approaches. Use that…
Distribution should not sit outside the product as a later marketing problem.
It has to be designed into the first use and proven from the beginning. The moment a user gets value, the product should naturally create something worth sharing, inviting others into, or building on. Otherwise, the company is not running…
A director is a navigator who has learned to leave the first-person view without abandoning what the first-person view knows.
The navigator encounters the system from inside it. Friction is not theoretical there. A beautiful design can still be frustrating. A coherent strategy can fall apart in use. A feature can…
Coding is the production of software.
Product is the ongoing selection, shaping, coordination, and evolution of software under changing conditions. That is why “build” has two meanings. Sometimes you build to learn. Sometimes you build for quality, reliability, and scale.
Shape Up’s appetite is a value judgment, not an estimate.
Shape Up’s appetite is a value judgment, not an estimate: decide how much team time a problem is worth, then shape scope to fit that creative constraint. Agents keep that principle, but now you have to budget machine execution and human review separately. Before the work starts, set: • Appetite: what the problem is…
An AI-native product means the features on the critical path require intelligence.
People who use AI without enough language risk becoming products of its defaults.
If you can’t name what you mean, what’s missing, or why something is wrong, the model starts supplying not just your answers, but your ways of thinking.
One way to understand taste is to notice its absence.
You lack taste when you can tell something is off but cannot say precisely what, why it matters, or what you would change. Taste closes that gap.
When describing a dream to a generative image tool
Describe only what you actually remember. Prefer concrete particulars to interpretive labels. Preserve what is strange, even if it feels impossible. Do not invent what memory leaves blank. Let arrangement, scale, light, posture, and action carry the feeling. Keep uncertainty where memory is uncertain. Add nothing merely to make the image richer. Stop when ad
The new contemporary always arrives late.
Software has never been about writing code.
It has always been about giving people more hands, more mind, more reach. Watch what people actually do. How their days unfold. Where they stop. What they repeat. What they work around. Where the hassles are. What they’ve simply learned to tolerate. Where the emotion is. Notice when something becomes easier. Relief. Satisfaction. Delight. Study awe. Find whe
I do not hate many things, but I do hate avoidable impedances.
Know when a way of thinking has stopped paying off.
All of life is reinvention.
A life is not one stable identity waiting to be understood. Work changes us. So do failure, age, relationships, and circumstance. Any fixed account of who we are eventually becomes incomplete. A final account of a person would require that person to stop changing. The task is not to preserve an earlier version of yourself, but to carry enough continuity forw
Judgment Framework Epistemically examine the principles.
Do not inherit or assume them. Understand where they came from, what reasons support them, what assumptions they contain, how they hold up against counterexamples, and how they relate to competing principles/opinions/positions. Which principles need this scrutiny is usually clear only once the problem
I can get non-determinism from humans.
On Writing
2026 —————— Good writing can be generated from linguistic competence. Great writing often begins earlier—with a particular consciousness encountering the world. Write from perception, attention, and discovery. Writing is not merely the reporting of a conclusion you already possess. It is a way of thinking. Pay attention to what is actually there, especially
When writing is done through a surrogate, you bypass your own perspective.
Make your writing arise from things encountered rather than presenting the finished conceptual system from the beginning.
AI writing frequently tells us the distilled insight and then demonstrates it. For example: “Doing is a way of knowing.” That's a strong formulation, but it arrives as a polished proposition. A more lived version might first stay with the experience: You start writing because you think you know what you mean. Then the sentence refuses to work. You change it
What Human Practice Is For
AI may separate making good things from becoming capable of making and judging them. Judgment usually develops by being wrong in useful ways. You predict. You choose. Something happens. You discover what you failed to notice. AI can remove the mechanical work and leave that cycle intact. Or it can remove the cycle. Then the problem is not merely that machine
Repertoire: encounter many examples.
Perception: notice differences. Naming: make those differences explicit. Mechanism: understand what produces the effect. Judgment: decide which effects matter most here. Calibration: discover whether reality agrees. See broadly. Notice finely. Name precisely. Understand causally. Choose deliberately. Update honestly.
Good writing is answerable to something beyond its own fluency.
A writer looks until the thing stops being generic. What moved? What made the sound? What shape did it have? What was it made of? What is it called? If you do not know the name, find it. A vague adjective is not a substitute for having looked. The sentence can be part of the looking. It can reve
Zeroth Principles Thinking
Most of the time, we meet the world through self-consciousness, habit, convention, ambition, judgment, preconception. Sometimes attention clears that away. Often in nature. You stop thinking about yourself having the experience. You receive what is there. I’m tired of treating ideas like math. Clichés, homilies, formulas: answers arrived at before the lookin
Don’t dumb it down, smarten it up to simple.
To determine a property of a work is to fix what that property is rather than leave it to another process to supply.
A person authors a property only if the person determines it. Therefore, when a generative system supplies a property that the person’s choices left undetermined, the person does not author that property. A person authors a generated work only
Power is the ability to create irreplaceable value, see reality without dogma, choose sincerity over perfection, and never let fear govern your decisions.
Great ideas don’t come cheap, and the people who generate them are exceptionally rare; however, people often learn how they emerge through the attempts to realize them.
Ideas often arrive slowly and serendipitously, but not without work, openness, and the ability to recognize when disparate parts add up to something great and when they do not. Great ideas ar
How to Build Taste That Isn’t Just Opinion
Taste is judgment, not objective proof. But developed taste is more than preference: it is perceptive, informed, intentional, comparative, context-sensitive, and accountable to consequences. 1. Separate Mechanics from Aesthetics “Competence is the price of admission for preference.” Fix execution errors before debating artistic choices. Sloppiness is not a s
Conventional problem-solving asks, “What solution will achieve the goal?” Fit-oriented design asks, “Where is the current form fighting its context, and what change would help the whole work together better?”
AI is a computational executor at speed.
Understanding is not necessarily the same as correctly following rules. A system can apply rules, produce correct answers, and explain its steps without thereby demonstrating that it grasps what the rules mean or why the answers are true. Gödel’s First Incompleteness Theorem shows that any consistent, effectively axiomatized formal system powerful enough to
Squeezing intuition into the right idea, then pulling that idea out of abstraction and into a shape that can be made real.
Every idea has a load-bearing part.
Find the part with the fewest good answers and the greatest effect on everything else. That is the center. It might be the first action, the data model, the trust moment, the permission boundary, the feedback loop, the handoff, the interface metaphor, or the final artifact. Shape that first. Do not let AI build the easy outer parts while the center remains u
Can you subtract an element from the object without destroying the system?
In elite design and style, every line, texture, and seam performs structural, functional, or visual work. If you can remove a detail without the object losing anything, that detail was superfluous.
Let novelty be outnumbered, not outranked.
An intuition or opportunity, once found and clearly defined, is already halfway to a solution.
What you have is probably not the answer. But it may be the right intuition. Get that right first.
Good Business
Make something excellent, durable, and true. Charge enough to make it without exploiting people. Pay workers and suppliers fairly. Earn profit honestly, instead of underpaying people and making up for it with charity later. Grow slowly enough to protect quality, craft, culture, and community. Use technology only where it helps the work and the people doing i
The ability to judge reality and the ability to operate inside machine-mediated systems without being controlled by them.
Put things together when they change together. Separate things when they change for different reasons.
The planet is AI’s embryo.
We are the placenta. Its hardware is not metaphorically earthly. It is Earth: gold, silver, copper, silicon, dug up, refined, wired, cooled, and taught to think. Our primate fingers are doing what tectonic pressure, glaciers, and volcanoes could not do fast enough: arranging dead matter into circuits. Maybe we are not inventing a m
The mystical experience is this ordinary awareness.
Generating Breakthrough Ideas
Think hard, stop thinking, notice what appears, then test it brutally. No breakthrough? Acute desynchronization of the Default Mode Network can act as a kind of biological catalyst, which may open a window of neuroplasticity. Pregiven answers stop thinking. Base perception is a controlled hallucination. Reduce confidence in priors. Increase confidence in sen
The Emergent Layered Ontology
The bit and the Planck scale should not be treated as rival candidates for the single foundation of reality. The it from bit or bit from it nonsense. A bit belongs to an informational logic of distinction, while the Planck scale belongs to a physical logic of measurement and limit. These may be connected, but neither clearly eliminates the other. Reality app
Minimum Ideal State
Founders should not innovate across every dimension of a product at once. In mature or semi-mature categories, they should preserve proven patterns unless they have strong evidence that those patterns are wrong for their target users. They should concentrate novelty in the one or two areas where they have a real insight, and that novelty should produce a mea
intuition ideas
The default is the decision someone else made for you.
We can learn about the world through tools in two broad ways: by understanding before use, or by understanding through use.
The first says: “Learn the tool, then use it.” The second says: “Use the tool, then let experience teach you what matters.” Some tools you learn before you use. Others you use to learn.
The fastest path to reading code well is usually: 1.
Read tests before reading implementation 2. Read small amounts of code 3. Verbalize your prediction before running 4. Run it 5. Compare reality vs prediction — precisely, not approximately 6. Deliberately induce a failure and predict how it breaks 7. Modify it surgically — one change, one observation 8. Fi
How to write a problem statement to design well.
DISCRETE SUB OPPORTUNITY Dev shop owners need a way to keep project work moving toward billable milestones without manually chasing team members for updates. Under tight time pressure and uneven cash flow, even one completed task sitting unnoticed for days can delay the next step and push out client billing. The system should automatically notify the right p
A Semantic Form I Love
Appositive fragment. Noun-phrase fragment with relative clause. Temporal opener + main clause + content clause. Proper-noun fragment. Adjective appositive fragments. Equative setup + dash + explanatory clause. Plain Language Explanation Appositive fragment = Rename the thing. Noun-phrase fragment with relative clause = Say what kind of thing it is. Temporal
The internal “feelings,” “tension,” or “recognition”, as if AI had introspective access to them, are at most metaphors for patterns in its outputs or processing.
Understanding does not equal computation.
If I Could Give Only Two Writing Tips
Avoid ready-made phrases. They are the enemy of good writing. Do not surrender to language. Command it. Notice one specific detail and record it in a short, declarative sentence. Strip away explanation, transition, and commentary.
Interaction design asked: Can people use this?
New Lines for Every Sentence
Treat each sentence as a self-contained unit of thought. Give it physical space and silence. Use the break as a tactical tool for revision. Unspool dense paragraphs into a vertical list of sentences. Then inspect the internal logic and clutter of each line without the distraction of the surrounding paragraph. Break the flow, and you must confront the actual
Logic is the skeleton; architecture is the flesh, blood, and (too often) the sweat.
Ideology is premature closure, a prison of prefabricated language.
A civilized empiricism means: You know your concepts are socially shaped. You know your tribe distorts your perception. You know evidence does not simply “speak for itself.” But you still believe reality exists outside your frame and can correct you.
Path Analysis
Alternatives are not merely different ways to execute the same idea. They are distinct ways to address the same opportunity. Without this discipline, you are not doing good design. And when a requirement specifies the solution, the design is a tautology. Separating requirements from solutions is key to doing design.
The Force-Multiplying Challenge and Opportunity
The hard part of collaboration is this: we all work differently. There is no single right way to work. But when we have to work together, and my way of working interferes with yours, something has gone wrong and needs to change. It is not enough for me to say, “This is how I work, and it works.” When I work with you, something has to change in both of us, wi
Figure out what’s actually happening, decide what’s possible, draw the line, spell out what matters, then choose what to give up.
Pragmatics and epistemic reasoning are back and not going away.
Opportunity Leveling
In a just society, people can reasonably expect the benefits attached to their roles under fair rules. But it is a mistake to think we inherently deserve a society that highly rewards the talents or traits we happen to possess. Inequalities are justified only when they make the least advantaged better off than they would be under a more equal arrangement. Si
Argument Reconstruction
First, separate what was said from what would need to be true. Ask: “Assuming all these premises are true, does the conclusion necessarily follow?” Do not argue with the conclusion until you know what unstated premise is doing the work. Add the missing premise explicitly if needed. Once you spot the gap, write the missing bridge as a separate line and mark i
Just Barely Enough Product
Planning + Serendipity Principles
Plan for direction, not prediction. Decide where you are going, not how every step will unfold. - Do not try to control the world; prepare to respond to it. The mistake is trying to control the system and context instead of building the ability to adapt within them. - Use plans to reduce confusion, not eliminate uncertainty. A plan should clarify action, n
2013, the summer of my life. The summer I fell in love. The first time we met, I knew I was in trouble. Derya. Radiant. Charming. Beautiful. And one of the sexiest things about her: she made me laugh.
It’s not enough to know how to decompose an object or problem.
To recompose with imagination, you also need to know how to ask of each part: Does the system fail without this? If yes, it’s an epicenter. Does this part have the most "unknowns"? If yes, it should be sequenced first. Could this part work in a different product? If yes, it has transformational or adjacent possible potential.
AI Literacy = E × P
Our Jobs
AI won’t take all jobs, but it will absorb many tasks inside most jobs. It will eliminate some roles where the remaining tasks can be fully standardized. In response, organizations will (maybe should) concentrate human responsibility at decision gates: the points where someone must approve an action because the costs of being wrong are meaningful. This shift
We don’t need politics.
Whether mechanical or intellectually demanding, many forms of work contain a variety of toil—some that deepen thinking and skill, others that merely consume time. Wisdom lies in discerning where automation serves us and where preserving human engagement matters more.
My Modified OODA Loop: The PC²UDA Loop 1.
Problem: Notice an inadequacy (empirical or conceptual): contradiction, surprise, goal failure, explanatory gap, unexplained regularity, new constraint/aim, or a mismatch between prediction and outcome. 2. Conjecture: Propose 2–3 explanatory candidates (not just predictions). 3. Criticism: Attack candidates hard: - N
Once you’ve developed enough experience to make sound judgments, approaching strategy through rational inference, imagination, written analysis, and visual reasoning gives you a distinct advantage over those who rely primarily on tools or research.
You are free to do what you want, but you are not free to want what you want.
A great problem solver does two things: (1) tracks truth under uncertainty and (2) finds structure in…
A great problem solver does two things: (1) tracks truth under uncertainty and (2) finds structure in complexity. Everything else is support skill: language, values, self-control, and coordination. The Two-Part Engine: 1) Probability + Causality So you don’t fool yourself. (Calibration, base rates, evidence quality, confounding, causal graphs / counterfactua
Judgment skills are central to critical thinking: deciding what’s true, what matters, and what should be done.
These forms of metacognitive knowledge should be taught to every kid. These are high-stakes agency skills that can atrophy if delegated. A citizen who can’t form an independent perspective is easy to steer—by ideology, by media, or by AI. Short list
All the following mental models are handy. But when the map and the territory disagree, trust the territory.
80/20 Rule (Efficiency): Focus on the vital few, ignore the trivial many. Territory check: Often, there are more opportunities in the under-managed 80% of tasks than in the over-managed 20%. Signal: We focus excessively on the high-payoff 20% while ignoring the untapped potential in the remaining 80%. Inversion (Safety/Reliability): Avoid stupidity rather th
Think on paper.
If you’re aligning on what the space looks like and how pieces connect (opportunities) → use a tree.
Don’t Atrophy Via AI
Keep a manual spine. Automate limbs, not judgment. Own the call. Delegation isn’t abdication. Earn the question. State it, constrain it, name what changes your mind. Do the first 15%. Frame, bound, estimate, sketch the system…then prompt. Match tool to risk. Brainstorming can hallucinate; decisions can’t. Testable beats believable. Never trust smoothness. Co
What was it like to wake up after having never gone to sleep?
Explore Being Right We need not be right to know; what matters is reaching a place where no further criticism…
Understanding Memorizing
You might find hidden truths through experience that you can generalize into heuristics. Then, you can create a philosophy around that. Any pursuit you master will lead to philosophy, i.e., will lead you to becoming a philosopher, if you go deep enough and generalize those truths back out.
Physics of the Problem
Delimit the situational boundaries in both space and time and make the context specific. In other words, clarify where and when the situation occurs, and describe the context (when and where the unmet/unrealized need or desire occurs, and the current approach/workaround and its result) precisely. Doing so narrows the focus of the problem so tightly that it i
Problem-Solving Skills We Should Teach Kids
Teach them to model situations as discrete structures (graphs, sets, constraints). Teach them families of algorithms (search, greedy, DP) and when each fits. Teach invariants, extremal reasoning, and impossibility proofs via puzzles. Teach them to reason about complexity and trade-offs (exact vs heuristic). Do it through rich, varied problems that feel like
Creative Problem-Solving Sequencing Heuristics
1. Frame the change: Before doing any work, look at real examples and write a simple statement of the core problem and what “better” will look like for real people. 2. Dump and cluster the work: List every relevant task, constraint, and idea, then group items that obviously must change together and give each group a clear, plain-language name. 3. Choose scop
Don’t Violate These Product Development Principles
Get feedback from real use, not projections. Iterate on actual problems, not hypothetical ones. Direction is a filter, not an order-dependent roadmap. Decide as late as you can without losing options. Favor near-action decisions for reversible, loosely-coupled choices where fresh information is highest quality. Decide earlier when a choice is hard to unwind
Logic hits incompleteness, algorithms hit undecidability, physics hits interpretational gaps, cosmology hits measure ambiguity.
The Hill Chart
Basecamp’s Hill Chart is a great example of form-context fit. It maps a universal reality of knowledge work—unknowns → knowns—into a dead-simple shape that exposes risk and progress without fake certainty. It eliminates the core misfit (lists/burndowns hide uncertainty) with a form (a single hill) that any team can grasp in 30 seconds. Forces → Form → Fit (a
Simple Difference Principle Examples
Paying doctors more If higher pay attracts talent and expands clinics in poor areas so the poorest get better care than before, that inequality passes. If it just boosts doctor income with no improvement for the poorest, it fails. CEO bonus plan If bonuses come with raising entry-level wages and training so the poorest workers end up better off, okay. If bon
Describe observable effects and regularities in the way things behave (laws).
This method marks a clear boundary between describing observable effects and guessing at causes beyond empirical evidence (hypotheses).
The universe is the answer; our questions—their quality and fit—matter.
Our goal is to find the right ones. When a problem stays insoluble, always suspect the question—it’s been asked the wrong way.
Model systems precisely (JSON, state diagrams, invariants).
Test before coding—list what breaks. Read unfamiliar code weekly. Debug with logs, not guesses. Ship to real users and iterate. Master one domain for 2+ years. Repeat forever. Everything else changes. This doesn't.
To design well is to make wrong action costly
Motivation, Trust, and Direction
Summary: Reward real work: collaboration, momentum, learning, and novelty that ships. Set strategy above projects and let teams inform it. Use goals to test strategy, not to manufacture motivation. After each cycle ask if you’d extend the work and whether the team stayed healthy. Share risk with clear bounds. Assume trust, start at 50%, and use accountabilit
Don’t ask what makes you (or others) unique.
Successful decision-making requires getting straight on criteria, conditions, and constraints.
Syntax-craft for LLMs
Begin with the outcome—a one-line task; follow with parallel, single-action steps; define terms at first mention in brief appositives; then foreground scope and edge cases—“If/When…, then…”. These cues align syntax and intention so the model codes to spec.
Competing factors: Figuring out the right things to build often requires identifying the embedded criteria behind the avoidance or the workarounds.
AI cannot replicate the human act of thinking and emotional introspection involved in true writing—the act of meaning-making that is uniquely human and not just text generation or simulation.
If a moral rule is grounded in shared human cognition, reflected in cross-cultural judgments, and justified…
If a moral rule is grounded in shared human cognition, reflected in cross-cultural judgments, and justified by reasons any culture can accept, then it’s true for everyone; those conditions hold for “intentionally killing a non-threatening person is wrong,” therefore, global moral relativism is false. This is an epistemological problem. We have instinctive me
Process-Oriented Success Measurement
Process that is about improving the craft and chemistry between people is much more important than outcomes. The point isn’t to know, it is to do. And for it to be done. The Holistic Approach to Outcomes: Outcomes must extend far beyond simple output measurements. It’s not just about the output and outcome, it’s about how did it feel as we went. When evaluat
The Aftermap: Navigate by wake, not wish.
An Aftermap is the map you draw after outcomes. Start from where you actually are, and revise direction based on what just happened. The longer you wait between intent and review, the more variability you incur—call it variance debt. Keep the loop short. An Aftermap isn’t a log of actions; it’s your updated model of the world: what you now know about, what y
Ask an LLM to teach you about
Limits of math/CS: Incompleteness → computability → complexity → Kolmogorov complexity. - Physics: Classical chaos → fluids/turbulence → QFT basics → quantum gravity puzzles. - Inference: Graphical models → causal DAGs/ID algorithms → experiments under interference. Choose one of these highly complex topics and engage the model to help you deepen your unde
The Nubs of Technology
The present on your watch is represented by a hairline that is as thin as possible as is consistent with visibility.
But the future does not change into the past, where everything is a result of the past. The past is caused by the present. Now. The present. It’s not counted in seconds. It drones.
Treat your confusion as signal, not noise.
As You Go
If you can plan a year out, you can also steer month to month. When you feel sure about twelve months, you can adapt week by week. By late June, you know more about July than you did in February. Decide closer to the moment. What’s so magical about mapping the whole year? Because conditions change and information improves, a decision taken near the action is
Differentiated Learning = (Curiosity + Interest) × (Guided Inquiry + LLM Support)
When a learner’s curiosity and interest are supported by powerful inquiry tools (like ChatGPT), the result is differentiated learning — learning that is tailored to the individual’s needs, questions, and pace.
Invented or Discovered
Mathematical primitives, like prime numbers, are discovered. Something that exists independently, like gravity. Mathematics is too big to be captured entirely by rules. There’s likely an encoded structure underlying all reality and it’s decoded using mathematical models. That decoding is possible given our innate cognitive structures. We know the incompleten
The mind can grasp unprovable truths.
Reality may be a single computational ocean — but it forms waves, organisms, and ideas — each governed by their own emergent rules that cannot be flattened.
Bayes’ Theorem turns belief into arithmetic — and evidence into calibration.
G ≡ ¬Provable(⌜G⌝)
Trust-less
The best thing you can do is to trust yourself and others to be untrustworthy. It’s the only authentic approach toward building trust. To prove your own trustworthiness, you must first trust yourself. But that’s what you’re trying to prove. That’s inconsistent. Everyone knows each person needs to change themselves. But we can’t do that entirely because the s
You need at least the following fixes to begin approximating toward super AI intelligence: Give AI direct…
You need at least the following fixes to begin approximating toward super AI intelligence: - Give AI direct sensory grounding (sight, sound, touch) tied to internal concepts. This is a form of embodied AI — where meaning is tied to the agent’s interaction with the world. - Develop meta-reasoning layers — self-modeling systems that monitor their own rule-base
On Character
Rationality is not about knowing facts, but knowing which facts are relevant, i.e., knowing what to ignore—a kind of wisdom.
Get stable, repeatable AI answers
Turn off randomness: - Use temperature = 0 and always use the same seed (like 12345). Force a reply structure: - Tell the AI exactly what format to use, e.g., “Answer as a JSON with ‘summary’ and ‘tags’ only.” Set a stopping point: - Add “END” at the end of your question and ask the AI to stop when it gets there. Limit answer choices: - Tell the AI: “Only an
Mechanical Rituals
🕰 Mechanical Watch Why: Trains appreciation for complexity, rhythm, and maintenance. A model of how systems can keep time through nothing but springs, gears, and ingenuity. It also teaches respect for patience and precision. ⸻ 🚲 Bicycle Why: The most elegant form of locomotion—simple, efficient, and entirely human-powered. Learning how to ride and maintain
Respect Space
As a writer, especially when using digital tools, assume the blank page is not infinite. Train yourself to say a lot in a little bit of space. That’s what the act of being succinct looks like in action. To say no more and no less than is required. Respect space; even when it’s unlimited.
Shaping is not about all the details.
It is like the schematics, not the specs, of a thing, and a way to prevent time bombs later when real implementation work occurs. It is the stuff behind the walls. The possible locations of the walls. Not the beautiful stuff that covers up the walls and gets in the way of quickly understanding what needs to happen. And w
Teams require resourceful, cooperative, perceptive, curious, and sincere people.
Physical heirlooms endure because materials + workmanship + culture align.
Software can match that durability when licensing + spec discipline + community stewardship play the same role that bronze, spruce, or hand‑wired turret boards do in the physical world. The moment a program is designed with those constraints in mind, it’s no longer disposable tech—it’s a candidate family heirloom waiting to happen.
Spontaneous intuition.
Shared Risks
Goals are guesses. Risks are real. Teams that align on what could go wrong—not just what they hope goes right—move faster, adapt smarter, and own the outcome. Progress is de-risking, not just chasing goals. Prep Plan We achieved goals differently across teams, but share risks along the way. Risks and contexts are what unite us and the work we do.
Unified Argument for an Innate “Universal Moral Grammar”
P1 (Empirically limited moral input). Across languages and cultures, children encounter only a sparse, fragmentary stream of moral examples—speech, observation, and feedback that is orders of magnitude smaller than the space of possible adult moral judgments. P2 (Early & cross-cultural moral convergence). Nevertheless, infants and young children everywhere d
A First‑order Value Judgment
Snapshot A society is just iff its rules (a) are chosen from a stance of equal moral standing and (b) never evolve into material patterns that re‑create domination or humiliation. Therefore, whatever advantages you hold remain legitimately yours only so long as three empirical guardrails simultaneously hold: 1. Floor‑raising – the least‑advantaged keep gaini
The problem is not knowing the right problem…often the result of asking the wrong question or accepting the wrong answer or idea direction.
Once you figure that out, get things tight and narrow. A problem well-framed is a problem half solved.
As product managers, our role mustn’t be to focus on the problems people have—that’s the role of a psychologist. Rather, our primary concern should be understanding customer forces and addressing the specific challenges within our design space—challenges we can refer to as “misfits.”
On the one hand, we must actively identify opportunities presented by custo
A Natural Order of Product Development
Healthy product discovery and delivery involve moving fluidly between high‐level goals and lower‐level concrete decisions. Below is a way to read the diagram attached here through the lens of Abstraction Hierarchy: 1. Wishing (High Abstraction) - Submit raw ideas - Very high level: The team is exploring the opportunity space and the seed of an idea without c
Crux-query.
Too many “ands” are the enemy of small, simple things.
Internal Language to AI Prompt
1. Crafting Deterministic, Precise Prompts: To instruct AI deterministically, follow these principles: • Explicit Clarity: Specify exactly what to do, how to do it, and clearly define expected outcomes. Example: Instead of: “Summarize this article” Say: “Summarize this article in exactly 100 words, focusing specifically on key facts and omitting opinions or
Kill the Spec, Build the Thing
Functional specs are fantasies. Reality is built, not imagined. Words on paper won’t code themselves into existence. Specs appease egos more than solve problems. Tough decisions don’t fit neatly onto pages. True costs remain hidden. Agreement in text is illusion. Everyone nods, but nobody sees the same vision. Reality clarifies differences. Specs force criti
If you leave enough room for your product to grow, it will outgrow your plans altogether.
Make Distinctions
Beliefs are imitations of reality. Representations become the real. Repeat this program: Active interpretation, interacting processing. Subscribe to the master narrative of the actual living world.
Good design happens when you find a solution that satisfies multiple forces at once.
Like finding the sweet spot where: - The page loads quickly enough for users - The server doesn't crash - The business still makes money That's it - forces are just the different pressures you need to balance when designing anything. They are the underlying t
Prose for Business That Won't Bore the Shit Out of Your Colleagues
Be direct, succinct, aphoristic, and rhythmic. Communicate and inspire. Traditional: Our strategic initiative to digitally transform our operations will encompass multiple phases of technological integration and process optimization, aiming to enhance operational efficiency and market competitiveness. My Crystalline Method: Legacy systems devour millions. In
Keepin’ it quiet and confident does wonders.
Some human in some hunter-gatherer group changed.
That person could do what we all do internally (thought) but without externalization (communication) because there was no one to talk to. Whatever is prelinguistic in our heads, bits and pieces of which we get when we think, would have been available to nearly one person. This developed without external pressures, like a snowflake, because that’s the way nat
Good Generalists Generalize
They are antifragile. When ecosystems shift—such as during a market fluctuation—generalists are better equipped to handle change than specialists. The prevailing paradigm, regardless of its scale, may have shifted to the periphery of a specialist’s expertise, leaving them with their tools in a world cast in shadow. Generalists apply governing principles acro
What Is Wind
? Wind is the quiet traveler, Drifting from high to low, Where warmth and coolness softly meet. It carries the sun’s uneven touch A stirring breath, a fleeting whisper. Each gust is a thousand murmurs, A dance of unseen forces, A step in nature’s measured song. In its passing, the world is rewritten, Etched in delicate patterns of motions A quiet
By employing narrative devices, aphoristic language, and abductive reasoning to craft a compelling strategic narrative, we can articulate the purpose of a business or product and its significance to people. This approach enables us to scale autonomously decentralized actions in a coherent direction.
High-level Example: UrbanFlow: Transforming Medellín Throug
Favorite Top-bound Spiral Hardcover Notebooks Pen: Marker:…
If you accept that (1) cognition (including language) is best explained by recursively applied symbolic rules…
If you accept that (1) cognition (including language) is best explained by recursively applied symbolic rules and (2) physical reality is best explained as a manifestation of recursively applied computational rules, then it logically follows that both mind and matter can be seen as “threads” in a unified, generative network of rule-based processes. Thus, the
Looking at Looking
The mind turns back to watch itself, like a river starving for its current— each glance creates another watcher, infinite mirrors in a bare room. I stand at the edge of thought, marking the place where observation begins, but the marker bleeds with every mark, and the beginning keeps beginning again. This is how wisdom comes: not in the looking, but in the r
What first leaps to mind might just be a mistake.
The Unrecognizable
Get off the nod.
No small moments
Friction and Flow
Time moves with flow and friction. We touch it, try to hold it still, but it shifts beneath our hands. We slow what we love, pressing against it gently, hoping it will last— but that only makes it slip away. The hard things, the dull work, the waiting— we smooth their edges, give them flow, and they disappear before we notice. Either way, time goes. It is no
The Way It Comes
You lean too hard against the wind. Let go. Somewhere, it is already waiting, watching from the tall grass, breathing easy. It was never locked away— just resting, just waiting for you to stop pushing so hard. Find the weight that fits your hands, the ache that feels like home.
Our Science-forming Capacity
Our intuitive way of understanding the world just happens to be different from the way the world appears when we apply our science-forming capacities to it. We can only talk about different aspects of the world. Even with very little knowledge about the properties that allow for “thinking matter”, the only thing we are ultimately confident about is our own i
Management Without Command
Command-and-control structures typically increase transaction costs: handoffs, sign-offs, lengthy and questionable meetings, etc. By distributing authority, you move decision-making downstream, which shrinks batch sizes and queue times—boosting the throughput of the overall system. High utilization without local authority leads to congestion (like traffic on
Belief–Desire–Intention Product Model
Definition: This model applies when customers (or users) hold Strong Beliefs, which function like preferences that are not easily changed. - Opportunity: Because these beliefs are deeply rooted, a product or service that effectively aligns with them can achieve especially high engagement and loyalty. - Conclusion: Therefore, a Strong Belief represents a ma
Knowledge Acquisition Algorithm
1. Watch others do it incorrectly, not correctly (i.e., observe others being taught). 2. Do it yourself with guidance and short feedback loops. 3. Teach it to others ( ) See one, do one, teach one.
The only real (type of) social media app I know…
Make Malleable
Three Main Types of Arguments
Deductive arguments are most useful when you can confidently establish the truth of your premises and desire a definitive conclusion. - Inductive arguments help you move from examples or data toward a general (though not certain) rule. - Abductive arguments guide you toward the most likely or most reasonable explanation among several possibilities, especia
Flow Principles for Product Development
1. Time-Limited Cycles with Variable Scope Controlling cycle time is crucial to reducing Cost of Delay and feedback latency. By fixing a short time window (e.g., six weeks) and allowing scope to flex rather than extending the deadline, work batches remain small, queuing is minimized, and results get validated faster. This accelerates learning and prevents pr
Beyond Programming
Computational thinking is about looking for underlying computational processes and modeling them. The key insight of computational thinking: Sometimes, the best way to understand a system is to run it and observe its evolution rather than trying to “solve” it analytically. We reframe problems in terms of computation. This means: - Algorithmic Thinking: Break
It’s Not About Outcomes or Effort Alone
Effort that isn’t aligned with outcomes feels wasteful while prioritizing outcomes over effort leads to conservative, slower decision-making—favoring predictability over progress. True effectiveness comes from balancing both in a way that sustains momentum without sacrificing impact. 1. Lead with Context, Not Just Outcomes Focus on understanding the current
The poorest use of language is to speak without saying anything at all.
You’re not getting less when you cut.
Before and Now
The old plans sit on my desk like fallen leaves. I made them with such care then, believing in what I thought I knew. Strange how we trust what we used to think, as if being further away makes something more true. Walking this morning I saw deer tracks in snow— they don't follow paths we cleared for them. Each step knows more than all our thinking: the weigh
Types of Software Product Operating Rhythms
1. Reactive Work Reactive work is driven by immediate needs within the team or external stakeholders. These tasks are often urgent, labeled as “red alerts,” and managed on a First-In-First-Out (FIFO) queuing basis with Work-in-Progress (WIP) limits. This workflow is a continuous cumulative flow without a defined beginning or end. 2. Batched Work Batched work
Say it Well
Every spoken word is a work in progress. A draft in constant flux. Articulating the clarity of our inner thoughts demands deliberate effort; we must resist the temptation to shortcut the depth of our understanding. Instead, attune yourself to the flow of your thoughts and commit to expressing that inner dialogue, the central idea, with authenticity. Whether
Prioritizing CS Skills: What to Hone and What to Skip
Problems are functional issues or misfits identified within the design ensemble, often rooted in unrealized…
Problems are functional issues or misfits identified within the design ensemble, often rooted in unrealized needs, wants, or wishes. Struggles are social and emotional challenges we experience in moments of our lives, reflecting the tension between what is and what we desire or expect. Vision is a view of a better world—beyond what most can currently imagine
Designing From 0, 1, n
1. Draw a boundary between the part of the world you plan to change and the part you will not change to isolate the forces acting within the "designable" focus area. 2. Identify the forces that influence what you intend to design. 3. Represent and resolve related forces in visualizations (e.g., sketches, state diagrams, etc.). 4. Place the new design (form)
A piece of writing is always, in some sense, in a draft state.
But before you publish your “final version,” consider asking yourself the following questions: 1. Is it clear from the beginning what the paper is about? 2. Does each paragraph advance the subject? 3. Do the important ideas stand out clearly? 4. Are more details, examples, or anecdotes needed? 5.
How Framing (Demand Side) and Shaping (Supply Side) Ping‐Pong Plays Out
Sometimes you have a problem and no solution. Other times you start with a solution, but you haven't yet worked backward to the problem. The reality is that it’s not a linear process but a moving back-and-forth motion between problem (demand) and solution (supply). Example Scenario Let’s imagine you have the germ of an idea for a new product that automatical
A Controversial Take on Lyrics
All too often, aesthetics shadow substance. Even when the lyrical content is strong, form tends to be overemphasized—most commonly through rhyme—rather than harnessing the natural rhythm of artfully crafted sentences.
It feels prudent not to decide whether something is practical or impractical by labeling it as good or bad, or right or wrong. They’re not simply opposite ends of a spectrum.
Opinionated by default, but flexible under the hood.
Skillful Editing
Software has no natural boundaries or edges. There are no obvious laws of physics that tell us something is too big, too heavy, etc. There are no natural pushbacks. Shaping constrains the work so there are clear boundaries around what is in and what is out. Doing the shaping work well achieves a state of elegance, which is characterized by four key elements
A Dialogue Culture Playbook
The purpose here is simple: to learn from each other, to gather new perspectives, and to enjoy the process. But simple doesn't mean easy. Begin with kindness. When disagreeing—and disagreement is valuable—let your words be considerate. Always. This isn't about softening your “truth” but making it more likely to be heard. Take offense rarely. Reserve it for g
Counterintuitively, sometimes—especially in creative problem-solving endeavors—modeling the messiness of reality requires being imprecise or approximate. Imprecise wiggle room often better reflects what’s truly happening than basing conclusions on a precise but false premise.
The image presented here shows how this idea can be applied to understanding the pr
Healthy Sources for General News
Appetites, not Estimates
Estimates start with a design and end with a number. Appetites start with a number and end with a design. What the graphic below shows are two very different ways of deciding how (and how much) work gets done: 1. Estimate (Fixed Scope, Variable Time) - Here, you begin by locking down everything you want delivered—the full feature set, requirements, etc. - Be
Personalization in a World of Standardization
Adaptive learning systems like Large Language Models (LLMs) bridge the gap between professionals and beginners, creating tools that make knowledge accessible and relatable for all. By tailoring content to students' interests, these systems have the power to transform abstract subjects into engaging, meaningful learning experiences. LLMs can take personalized
The arts must end up saying something explicit with all their implicitness.
Questioning Without Presuppositions
Core Idea: Don’t embed unverified assumptions directly in your main question. - Practical Tip: Turn any unverified assumption into (1) a separate question to establish the fact, or (2) a conditional (“If P then Q?”). - Result: You maintain logical rigor (no hidden premises) yet still can explore scenarios where a presupposition might or might not hold. Thi
Where Power Rests and How Some Norms Form
In Hume’s paradox, it’s puzzling that so many obey so few. Taleb’s “intolerant minority” clarifies it: a rigid, unyielding few can overrule a more flexible majority. Their uncompromising stance sets norms others conform to, revealing how the many often get governed by the few.
Using a custom Transformer with only 21s of fMRI data, researchers predict brain states 5s ahead with 0.997 correlation. Short sequences suffice to capture functional connectivity, enabling accurate short-term forecasts and potential use in time-limited clinical scans.
Predicting Human Brain States with Transfo
Learn to teach to learn.
Planning Evolves Alongside Execution
We expect and long to have the past back and drive what’s next and distrust the present, which is why we create roadmaps (and backlogs). But they create a feeling of debt. They can become order-dependent, making it hard to drop things if things get stuck, and they are often inflexible once they are shared. They treat strategy as a problem in deduction, which
Differentiation Strategy
Over time, our product’s competitive set will emerge in the form of the attributes that define what our product is more and less about and whether the product is in or out of the competitive set when a “hiring” moment arises. For example: Snickers can’t take a feature request to be “more melty” and then win over some new customers. Milky Way can’t “add a lit
Designing Synthesis
Good fit (the fitness function) emerges from the ensemble—the interaction between form and context within the user’s environment. This ensemble reflects a relationship between two parts: form (the proposed solution) and context (the social, emotional, and functional forces shaping a user’s tasks and objectives at a given point in time). We aim to satisfy the
On Developing Taste
One key ingredient: reflection. Engaging with art, philosophy, or criticism deepens one’s ability to judge quality and meaning from experience, fostering a more nuanced sense of taste. In other words, critically evaluating your existing preferences broadens your perspective and, in turn, your sense of taste. It’s like walking a tightrope: one side is ignoran
Social media amplifies our emotions, but not equally.
Research shows that while most posts express joy or contentment, anger travels faster and further. Why? While joyful content tends to stay within close social circles, outrage bridges communities, spreading rapidly across weak social ties between networks ( ). The solution is
Our moral values emerge from biological systems, and ethical norms vary widely across space and time - this is simply an observable fact. These values are shaped by experience, existing within a range of possible options that have both variation and limits.
When discussing human affairs, while we cannot offer mathematical proof, we can observe patterns in ho
Product Management is a kind of perspective-taking.
Becoming (or ideally being) the customer is a powerful technique that moves us beyond mere empathy modeling. It widens our perspective to better shape the form-context boundaries of a solution’s intended use since direct experience can uncover hidden insights. It’s not just one role—it’s every role involved in interacting with the product. Additionally, the
If a system consistently produces a certain result, then that result reveals its true purpose.
Gauge things according to their criticality.
Apply no blanket standards across all situations. The law provides good examples of a system designed to do a good job of gauging things with varying levels of criticality across situations.
Creating dignity inside of a company culture often eludes people.
It’s understandable, as developing culture is a pretty nebulous task filled with noise factors, like human values and behaviors, that are not easily configurable. Add to that, it’s also a moving average along a relatively short time scale. The Westrum Topology is a start (image of the table be
It’s easier to agree on what exists than what is good.
keep change cheap
Yup.
Research in neuroscience (Zeki, 1999) shows that the brain responds more positively to patterns and structures reflecting the Golden Ratio.
Studies by Taylor et al. (2001) reveal that people find fractals with a fractal dimension (D) of 1.3–1.5 most beautiful, as they strike a balance between order and complexity. Kaplan’s studies on restorative environments
Progress Tasks
Tasks as Reminders, Not Milestones: Tasks are placeholders for intentions. They are just reminders. They are tools to manage memory, not direct indicators of progress or value. This is especially true because we discover tasks as we do the real work, i.e., we cannot know them all upfront when doing creative problem-solving work. - App State as the Source o
Scratching an itch on the back of a Möbius strip.
Always loved this.
The ordinary is that which nothing quite is.
Two digital games worth your time
GAME!
A variation of NIM I created with Claude (In like 3 minutes/4 sequential prompts. Bananas). Teach this one to kids. Don’t stop at Tic-tac-toe. Setup and Rules: 1. The game starts with 21 empty cells arranged in a row. 2. Two players take turns - Player X and Player O. 3. On your turn, you mus
Labor Market Predictions
Psychomotor work will be hard to displace and likely durable. Occupations that require high amounts of trust will stick around. Future knowledge worker titles will include familiar words like designer but also new ones like “Arbiter” and “Ombudsperson”. Non-technical roles that guide and evaluate AI-human alignment should fold into a function headed by a Chi
Later to Later
Later makes a pile at the end. Make it so there’s no time for later. It’s now (or in some short time frame from now) or not. This way later isn’t an obligation, but a maybe. Preserve optionality. Later needs to mean: not now, not soon, and not for sure. Important things come back. Decide on what to do now while also creating space to shape what to consider d
Creative Pricing Models
Honey (now part of PayPal): - Honey earns commissions when a member uses Honey to find available savings or to activate PayPal Rewards. They work with affiliates to help confirm customers’ purchases, so they can get a commission from the merchant. What makes Honey unique is that they pass some of their earnings back to their members in the form of their fr
Life’s a vacation from obliteration.
Be on Books…Beyond Devices
Across your space. Leave them by your couch. By your bedside. In a bag. On your desk. Leave your books all around so they’re to be easily found. Scan lines.
[...]the only people for me are the mad ones, the ones who are mad to live, mad to talk, mad to be saved,…
[...]the only people for me are the mad ones, the ones who are mad to live, mad to talk, mad to be saved, desirous of everything at the same time, the ones who never yawn or say a commonplace thing, but burn, burn, burn like fabulous yellow roman candles exploding like spiders across the stars and in the middle you see the blue centerlight pop and everybody
Practical Math
I think we should be teaching optimization word problems in math class to kids sooner than we think (closer to 10…not starting at 16 years old…Perhaps even sooner). Optimization word problems incorporate proportionality, probability, and feedback loops into a framework that is both practical and metaphorically powerful. In other words: useful. Its practicali
Cooperation
Unlike genetic evolution, which operates over millennia, language enables ideas to proliferate in real time, fueling cultural and technological leaps. Language gives us access to use cooperative models, like indirect reciprocity, in unlimited ways. Through storytelling, teaching, and dialogue, humans can build cooperative networks far exceeding the constrain
The Real Iron Triangle
In software feature development (and many other creative problem-solving endeavors), it's wise to trade scope for quality while fixing time—ideally to no longer than six weeks per project since software has no natural boundaries. This keeps the scope variable, which may seem counterintuitive. Time budgets, while seemingly arbitrary, act as creative constrain
Building Ethically
Just a few “should we do this?” questions software companies often don’t ask in the following ways regarding their designs: - Empowerment: Who’s having the fun (should the technology grow — or the person using it)? - Resilience: Does it make us more vulnerable (and more complicated because the system is more complicated)? - Empathy: What is the impact of sim
Users 💉
I find myself growing increasingly uneasy with the term "user-centered design." The word "user" carries a secondary meaning—"consumer of drugs"—which evokes notions of dependence, fleeting gratification, and exploitation. Meanwhile, "centered" inherently excludes everything and everyone outside its focus. I understand that sometimes it’s a placeholder term b
Measurements related to “feel” might give us objective facts about the human mind, but they don’t reveal what…
Measurements related to “feel” might give us objective facts about the human mind, but they don’t reveal what it is like to be such a mind.
Memorwe
We don’t just react like a single-celled organism, we can choose from a range of potential movements. How? We remember. And recalling the right memories sharpens our perception of the present. It is consciousness that connects the current physical structure of our bodies with the temporal structure of our lives. Neither are we predictable or free.
Computers Like Us
Routine work turns life into repetition, tasks divided so finely that the person gets lost in the process. Movements become mechanical, thoughts shaped by the rhythm of the task, until everything feels like part of the same engine. Now machines are given the scripts— AI built to mimic the way humans learn, adapt, and respond. It’s still rules, still patterns
Here are some tips when prompting AI as your Data Analyst copilot: 1.
Define the task clearly and concisely in the prompt. 2. Provide context or rules the data must adhere to. 3. For clarity, include examples of valid and invalid data. 4. Ask the model to reason through potential issues rather than apply fixed rules. 5. Evaluate the model's output critically
Generative Design
Good design defines a space of possible physical configurations, not one specific configuration. Provide so much latitude that it could lead the team or your future self to arrive at a configuration nobody could anticipate (because we can’t know until the realities of a project unfold), while still satisfying the system of relationships that make it fulfill
Wilhelm Von Humboldt said, in so many words, that human beings are born to inquire and create and to create under their initiative.
Any work that a person does under external pressure is inhuman. If a person creates something beautiful under external orders, we may admire what the individual does, but we will despise who they are. That’s classical liberal th
Rationality
What to safeguard: - Empirical testing and peer reviews - Checks and balances in power systems - Fact-checking - Adversarial proceedings - Freedom of inquiry and open debate How to safeguard: - Acknowledge fallibility - Show work (don't rely on authority) - Avoid politicization Why we fail: - Rather win versus learn - Depend only on our intuitions - Distrust
Criteria for a sound education
Independent Inquiry and Creation: A good education fosters the ability to ask questions, think critically, and create new ideas without relying on external guidance. - Resourcefulness: It equips individuals with the knowledge and skills to find information, formulate questions, and pursue their learning paths. - Critical Thinking: A good education encourag
Packaging
Effective document creation requires a thoughtful approach that goes beyond traditional writing techniques. Doing this well can create compelling documents that effectively communicate complex ideas. Always prioritize clear communication over digital convenience and form. And remember, as the reader, always evaluate the content, the concept, not the pitch. T
Generating Ideas
There’s no repeatable or universal method for exploiting the imagination. It comes when it comes. It’s maddening. Generally, the creative cycle has two parts: to take in new or existing knowledge and then let our minds do associative magic. Breaking monotony, or finding refuge in routine. Whatever you do, it’s all a test of fitting habits into form. Though
Comedy “Writing”
Louis C.K. says that comedy is a spoken art and that the words should come from your speech center, not from writing them down. He explains that when you write jokes down, you're using your hand to write, your eyes to read, and then your mouth to say them, which disrupts the natural flow of the material. He believes that the best comedy comes from the moment
Acting Beautifully
Harmony Perfection Finding Keeping Sincerity Seriousness Curiosity Certainty Resilience Fragility Adaptability Rigidity Humility Pride We cannot set knowing the truth as the standard for choosing, or the standard among the models we can produce. T
Many decisions are made unconsciously (see Benjamin Libet’s experiment).
Many things are happening that are just not accessible to consciousness. Why should they be? That’s why we have art, literature, poetry, music, and so on. To say that which can’t easily be said.
Similarities in delegating/prompting to both humans and AI systems
People and models excel at understanding and responding to succinct, clear instructions without the need for extensive guidance. Keep prompts simple and direct. Unfold as you go. Since people and models perform reasoning quite well, prompting either to "think step by step" is unnecessary. Avoid chain-of-thought prompts and instead, you (the human) must provi
Expertise.
There’s nothing scientific about collecting data and trying to make inductive generalizations from it.
That gets us nowhere. You need to inquire about the world. Craft the questions worth uttering you expect will surface valuable data and insights from running experiments. Synthesize learnings into new knowledge about the phenomena observed in the world. htt
If only all research papers were written like this…
We cannot set knowing the truth as the standard for choosing, or the standard among the models we can produce.
These are often tricky endeavors—too tricky for mortals. How do we recognize “the true enough”? The final call is aesthetic. For humans, those things are easier to discern. Developing taste is a skill. To notice the subtle and ineffable. To know wha
Parkinson's Law extends beyond time.
Pathfinding (Fit Check)
Sometimes when we don't know how to go forward in the solution space, we can use a counterintuitive technique. It can be easier to come up with three or more ways to do something instead of just one way. That's because when we make room for many options we can allow ourselves a bad option. Articulating what doesn't work can spark new and reframed ideas simpl
Focal Sketches
Arrangement of elements in a two-dimensional space: Fat marker sketch at large scale = Fine marker sketch at small scale. These annotated focal sketches can be arranged along a flow or be more targeted to a specific page/screen/place (on a lightweight annotated affordance map). This tool is critical when we need to communicate an idea when words can’t, since
Interaction Schematics
Lightweight text-based notation connects interface screens/pages/states with listed affordances ("affordances” are the parts of the interface that users can interact with, such as buttons, fields, and other interactive elements) in each so you can quickly see the macro relationships before worrying about how affordances and other elements must be placed in t
Your gut is a source of intuition to tap into and hone through experience.
Nerve ganglia represent an important evolutionary step - serving as simple but effective centers for managing basic physiological responses like moving towards, e.g., food, or away from potential harm. Inaction is irreversible.
A belief should be understood as a self-limiting function.
Types of bad anecdotal data
Compliments (usually mistaken for commitment). - Fluff (generics, hypotheticals, and the future). - Generic claims (“I usually”, “I always”, “I never”). - Future-tense promises (“I would, “I will”). - Hypothetical maybes (“I might”, “I could”). Go deep and ask story-based (space and time) causal questions to avoid the left-brain interpreter effect. https:/
You can hit the ground sooner than you usually think you can.
Do what you hate not doing, do what scares you, and ignore goals unless they change your present behavior.
Use examples as metaphors.
Generative Design Writing
Describing relationships provides latitude for interpretation and innovation. Specifying relationships also sets boundaries and helps prioritize elements (around the centers) inside the project. Focusing on the relationships between elements, rather than rigidly defining their forms from the outset, leads to more adaptable, creative, and ultimately more succ
PRIORITIZING PRODUCTS/FEATURES: HOW-TOS
A learning goal or research question starts by asking: What are people already doing that we want to better understand? It’s not about what we think they need, but about deepening our curiosity around their current behavior and approach. First, get into the physics of a business opportunity and frame it up. You should ask people in your ideal customer segmen
Don’t Outsource Your Thinking
We'll be signing off on slop as we lean too much on AI writing for us rather than teaching us how to improve. Writing well is thinking well.
The bicycle is a perfect blend of form and function, evolving into a living structure that adapts to human needs. It seamlessly connects us to nature, offers an intimate relationship between rider and machine, and empowers freedom while maintaining ecological harmony.
Another amazing thing about bikes is that reducing effort drastically doesn’t slow you down
Creating is the most honest and highest form of consumption.
The consumption of “doing a process” and finding what’s new from what’s already there is an act of art.
Art/engineering, philosophy/science, and relationships.
Traits that define us—resilience, creativity, and adaptability.
Historically, tools have expanded human capabilities, but there’s always a trade-off—convenience can erode critical skills. The stakes become higher when the tool isn’t just assisting but replacing foundational human experiences like decision-making.
It begs the question: What is the value of imperfection? Decision-making isn’t just about outcomes but also a
It was an uncomplicated making process, sophisticated only in its simplicity.
i
You don't "make" your heart beat any more than you "make" the sun shine. You are not a small, isolated being but a localized expression of the entire cosmos sampling itself subjectively. By realizing that circulating blood and the shining sun are equally involuntary and spontaneous, you begin to see yourself as a continuation of universal phenomena rather th
Psychedelics are now being medically described as tools for de-conditioning from societal constraints,…
Psychedelics are now being medically described as tools for de-conditioning from societal constraints, promoting unity, and broadening one's perception of reality: Desynchronization ( ) Psilocybin desynchronizes brain networks - PMC
Worker-owned cooperatives are an effort to replace wage labor.
In an ideal state, status and prosperity derive from how much one gives rather than hoards.
Virtual reality could allow individuals to share their innermost thoughts, dreams, and visions, akin to inviting someone into your inner world, much like the way octopi display their "minds" externally.
Best use case to this day, which I think is close to the original intent. Dreams of Vibrant Octopisimplecompose01jgsbmzc3e6srj4
Arthur Schopenhauer, in his essay “The Art of Being Right: 38 Ways to Win an Argument,” uses tools that are ideally only used to remove very bad actors when all else (reasoning) has failed.
Very dangerous stuff.
Yikes!
Quantum computers, in theory, have the potential to simulate certain complex systems, particularly those governed by quantum mechanics, much more efficiently than classical computers.
While they may help us understand aspects of life and reality at a fundamental level, their current and near-future capabilities are far from simulating all of life or reality
Agent Action Spaces
In the coming years, we may move away from the human-computer interaction model and toward an “agent-action space” model. We’ll need to start thinking about how action-space design patterns work best for “agentic intent” rather than human actors alone. It’ll be generative plus parametric. Agent Affordances: Interfaces must convey what agents can do in contex
As a product strategist and conceptual software designer, I reach for Miro a lot.
The work I do in it feels effortlessly lightweight. And a sign of work being lightweight is when throwing it away feels like progress. Miro has the right mix of primitives that support thinking in action. I can effectively provide enough latitude in it to enable breakthroughs, i.e., to enable my future self and others to work within rough, schematic constrai
Identity-Based Transformation
Lasting change comes from aligning your habits with your desired identity. Instead of focusing on what you want to achieve, consider who you wish to become. This shift is profound—like the difference between saying "I'm trying to quit smoking" and "I'm not a smoker." To understand this, imagine two people trying to get fit. The first thinks, "I want to lose
Untitled note
Happiness is the easiest thing to take for granted and the hardest thing to get right.
Topic of Emergence Chemistry to Biology: Emergence of Life Key Point: Life introduces concepts like…
Topic of Emergence Chemistry to Biology: Emergence of Life Key Point: Life introduces concepts like metabolism, reproduction, and evolution, which transcend chemical reactions. - Emergent Properties: Life is not just a collection of chemical reactions but a system where these reactions are organized to sustain, replicate, and evolve. - Metabolism: Emerges as
Capitalism needs an operating system update.
Call it, Capitalism 6.0. As I see it, the three main concurrent issues are: Crony capitalism, monopolies, and banks socializing losses and privatizing gains. We need universal access to essential and affordable services, many of which can be provided efficiently through automation and modern infrastructure. These
We should expect to receive rewards and benefits for using our skills and abilities according to the rules of society. However, it seems misguided and arrogant to believe that we inherently deserve a society that highly values the specific qualities we possess.
Any description of the universe which leaves you out is inaccurate.
You cannot prove or disprove anything except in reference to a set of assumptions which you must assume but cannot prove.
Find honest, humble, eager learners who demonstrate persistence, give them a chance.
We must not shrink from the karma of conflict, the essential discord of true argument.
To slash through the thicket of falsehoods, to lay low the towering follies of our time is not a breach of politeness but an act of intellectual hygiene. It is the necessary work to prepare the ground for a society based not on the comforting lie but the uncomfortable trut
Systems can be interconnected at all levels even when each plane consists of differing logic.
When a computer executes a program, the underlying logic gates and circuits don't change in their physical makeup; they follow the paths and states dictated by the abstract software loaded onto the computer.
Works of art are rarely unique.
Co-piloting AI
Ask AI to clarify requests as you work with it: Ask it to help you break down a task step by step by asking follow-up questions, or let the AI know to ask you to make something less ambiguous, or to prompt you for more details before proceeding. That technique is known today as Generative Active Task Elicitation (GATE).
Every piece of evidence that affects our confidence in an outcome changes its probability and should influence our actions. Probabilities reflect our ignorance of the world, not the world itself. As new information reduces our ignorance, it alters the probability.
Specify Relationships Over Forms
Design should capture relationships between elements, not fixed forms. Describing "a place to pause near the entryway" is adaptable across diverse environments, giving room for designers to respond to specific spatial or contextual constraints. This focus on relationships lets the design align with both the intent and the surroundings, balancing guidance wit
Language is the original code for hacking reality
The world is composed of description, and meaning is only found in language. To find the words that map to what we observe is a kind of alchemy. The goal is not to turn the world into words. The goal is to get your words as close as you can to the materiality of the world you’re noticing. Writing isn't just a transcription of pre-formed thoughts, but a colla
Say what you mean and mean what you say, listen to understand, read to consider, and write no more or less than is needed.
Reactive work and planned work are fundamentally different topologies and must live in separate streams or networks.
Objectives
Deciding where to go based on where you are is often wiser than deciding where to go based on where you want to be. We generally do not know the stepping stones in a complex search space to solve a complex problem. Instead, we need to amplify our ability to proliferate novelty. Novelty in business terms can be defined as something like: Information that resu
Metrics do not lie by being wrong about the thing they are measuring, they lie by being wrong about how important something is and how unimportant all the things you can't measure are.
Do things that you cannot justify yourself, which is so important because sometimes those are the most magical things of all.
Beware of enthymemes—arguments with hidden assumptions.
They can mislead unless you reconstruct them. Example "He's rich, so he must be smart." Reconstruct it: Premise: He’s rich. Formerly Unstated Premise: Rich people are smart. Conclusion: He’s smart. Now, you can challenge: Is wealth a measure of intelligence? Enthymemes are everywhere. Fucking everywhere
What you can't automate is to know what you should be trying to do.
Simplicity is achieved by being less abstract and by turning big things into small things.
Ask what needs to be solved together as one whole, and what can be solved separately.
Develop the skill to work on one thing without having to get into the details of another thing, i.e., to set boundaries where you can change one thing without affecting other things.
The more we work on a thing (the more we are "working it") the more bloated and complicated that thing gets.
Debating
First, determine if an argument is worth having. If your interlocutor cannot imagine they would change their mind under any circumstances then the debate is not worth having. You can't reason somebody out of a position they didn't reason themselves into. Ask: Under what circumstances would we change our minds? Ask: What definitions are we using for key terms
Philosophy is what you’re doing when you’re not sure what the right questions are to ask.
It is the craft of thinking. The most durable of all skills. Today, philosophy is more relevant than ever—a growth industry in a world that demands higher-order reasoning. While science progresses incrementally, rarely upending its foundational frameworks, philosophy equips us to confront paradigm shifts and rethink our assumptions. As we face complex challe
Contrast.
We like edges. We like seeing edges in our ideas next to each other that are different. That helps to improve our thinking. Our perception is involuntarily, directed by edges.
There’s knowing how to use a tool, and then there’s knowing when to use a tool.
There’s no nav in software product design and development.
There are phases. You do something, and then you see the next step. It is incremental, not monotonic. You need to be ready to make decisions at any point in time.
Networks not Chains
There is rarely ever a single root cause of a problem. There is rarely ever a single symptom (effect) of a problem. Causality is not a chain, but a network. There may be faster and cheaper places to intervene than the root cause. - Why do I think the root cause is the best place to fix this problem? - Why do I think I should only intervene at a single locati
People get mixed up and think that they should reframe problems.
But effective/useful reframing happens at a level above and below an empirically identified problem.
A Good Loop
The more we learn, the more we feel a slight return to where we started: Puzzlement. The increased quality, relevance, and novelty of our questions is the reward.
Experience is not linearly correlated with expertise.
Zoom in and out, from meta to matter.
The Solution Changes, But the Job Remains the Same
What changes are our expectations for quality and performance. This is tue across seemingly different segments of the population. Once we managed to tame boredom on the subway commute by reading the newspaper. Now, for better or worse, it’s the smartphone. The solution changed, the job/task remained the same, i.e., ward off boredom for commuters traveling by
Good meetings are small working sessions.
Real work is not simply being watched, but work is actively being moved forward and complex questions are cleared.
What things have the fewest configurations where there will be a good fit?
Knowing which things are most sensitive, we can then select things that we know relate to the centers of the form of other structures well or not. The whole comes first. Then gives birth to the parts by differentiating space in a sensible sequence. Design (or prompting, or planning) i
If you sense you’re not making tradeoffs, you’re doing something wrong.
Define your product’s competitive sets and what makes them suitable for different hiring moments.
Avoid defining the “product category”, as that’s outsourcing your strategy to the past.
Four types of product work
1. Improving a feature (increase customer satisfaction). 2. Getting more people to use it (increase % adoption). 3. Getting people to use it more (increase frequency). 4. New feature/product to support a new workflow (increase customers).
A helpful to-do list structure
Dump initial tasks into the Dump without worrying about relationships. - Affinitize tasks (cluster like with like), then name the task groups to avoid contriving the names of your lists/groups. Task groups should generally not exceed 5 business days (think of 4 days as the time box for each slice). - Sequence unknowns/risky task group elements first and cr
Learning happens in phases and moves back and forward.
To incrementally move your understanding forward, ask yourself: - Where am I at? - Where am I trying to go? - What do I know now? - What do I need to know?
Teach kids to
Pull things apart and put them back together - Ask good questions about how things work - Recognize their biases (start with cultivating the curiosity to challenge their beliefs) - Practice kindness via basic rules of manners and spatial intelligence - Create more than consume
Why “Mutual Exclusivity” Is Not Always Desirable
Many strategy consultants highlight “MECE thinking”—ensuring no overlap (ME) and covering all possibilities (CE). While this framework can help structure analyses or categories, it sometimes leads to artificially excluding options that could be combined advantageously. For instance: - Increasing Revenues vs. Reducing Costs: There is no inherent reason you mu
People name things the same thing for opposite reasons all the time.
You gotta weed through a lot of shit-brained thinking and pre-made intellectual distortion before you can even arrive at having a rational conversation about the facts.
Quality cannot exist without humanity, and dignity generates responsibility.
Human beings must be at the center of the work, and valued as the indivisible parts of the product that they are.
The mind spans many branches of history.
It forces a consistency condition to give the impression there is one thread of experience.
Climate scientists correctly identified the problem, but their solutions do not welcome growing countries where people are simply trying to get out of poverty.
Additionally, the metrics are not personal. Here are some high-level ideas: Water/Food quality → GI health, nutritional status, hydration. AQI → Lung function, oxygen saturation, respiratory symptoms.
Time is irreducible.
We have to live the time. No jumping ahead. We can lack free will, but we can never predict what we are going to do. So…I guess it all works out.
Frequentist Hypothesis Testing: In frequentist hypothesis testing, we begin with the null hypothesis (H0),…
Frequentist Hypothesis Testing: In frequentist hypothesis testing, we begin with the null hypothesis (H0), which often represents no effect or no difference (e.g., “A is equivalent to B”). We assume H0 is true and gather data to see whether the observed evidence is sufficiently unlikely under H0. Our goal is not to prove H0 true; rather, we want to see if w