Grain

01. Bottom Line Up Front

02. The Problem: Stable People, Unstable Behavior

Two things are true about people at once, and most simple models only account for one.

First, personality and cognitive ability are relatively persistent -- the grain doesn't move. The same person who's disciplined at 25 is usually disciplined at 45.

Second, that same stable person behaves very differently across situations and across time -- the same board of wood makes a clean cut here and splinters there. A model that just says personality + intelligence → behavior can't explain the second fact, because it only accounts for the material and ignores the cut being attempted.

Something has to sit between the fixed grain and the behavior it produces: a read of the current situation that decides what cut is actually being attempted, and whether it's running with the grain or against it. That read is the "+1" in this framework's name, and it's the piece that keeps the model dynamic instead of a static personality test.

03. The Nine Stable Dimensions: The Actual Grain

The stable side of the model -- the grain itself -- has two parts: an emotional-processing architecture and an intellectual-processing architecture.

Emotional Processing: HEXACO

Six factors, each a continuous spectrum rather than a type: Honesty-Humility, Emotionality, Extraversion, Agreeableness, Conscientiousness, Openness to Experience. This part of the model isn't new -- HEXACO has a real empirical track record (Zettler, Thielmann, Hilbig, & Moshagen, 2020, ran a large-scale meta-analytic investigation of the model's nomological net) and is used here rather than invented from scratch.

Intellectual Processing: Three Dimensions

Abstraction Ability -- pulling the general principle out of a specific case, rather than just banking the specific case. Expected to correlate heavily with fluid intelligence, and that's fine; abstraction may simply be one of the observable faces of g.

Structured Generalization -- recognizing the same underlying relational structure across domains that look nothing alike on the surface: noticing that a tax-planning problem, a piece of software architecture, and an organizational-design problem are, underneath, the same problem wearing different clothes. This is the least validated construct in the framework, and it started life under a different name.

It was first called "far transfer." That name got dropped on purpose. In the cognitive-training literature, far transfer specifically means training in one activity causing improvement in a different activity -- and the evidence for that kind of transfer is weak. That's not the claim here.

The claim is narrower and, honestly, more mundane: some people just seem to retrieve structurally similar patterns from distant domains faster and more often than others, as a standing trait rather than a trained skill. Whether that trait actually survives statistically once you control for intelligence, Openness, and plain domain expertise is an open question -- current honest estimate is somewhere around a coin flip to two-thirds odds that it holds up as its own thing.

Need for Cognition (NFC) -- the willingness to engage in effortful thinking, independent of the raw capacity to do it. This distinction is the whole point of naming it separately: ability to think and willingness to think are not the same variable. Someone can have real horsepower and just not spend it. NFC has decades of research behind it, dating to Cacioppo & Petty (1982), but the framework stays agnostic about whether it earns a permanent seat -- if it turns out to add nothing once personality and cognitive ability are already accounted for, it gets cut. That rule applies to every dimension in this model, not just NFC.

04. The "+1": Reading the Grain in Real Time

The tenth piece isn't a seventh personality trait or a fourth intellectual one. It's categorically different: it's the read -- the dynamic act of assessing, in the moment, which way a given situation's grain actually runs before committing to a cut.

Call it the Environmental Factor (EF). It is not a property of the environment itself -- it's the person's read on that environment: their sense of what's controllable, what actions are actually available, what those actions would cost, and whether the likely payoff is worth it.

This started under a more familiar name -- locus of control -- and moved through general self-efficacy before landing here, because neither of those existing constructs quite fit what was actually being described. Both have real research behind them: Judge & Bono (2001) meta-analyzed self-efficacy, locus of control, self-esteem, and emotional stability together as "core self-evaluations" and found each one an independent predictor of job satisfaction and performance -- evidence that this general territory matters, even though EF itself carves it differently than any single construct in that cluster does.

The clearest way to see why is a thought experiment: imagine being dropped in the middle of the ocean with no life raft. A highly agentic person in that situation doesn't decide they're helpless. They correctly conclude there are very few effective moves available -- tread water, conserve energy, watch for debris, signal if possible -- and hold two things as true at once: I am doing everything that can reasonably be done, and my available actions have almost no real influence on the outcome.

That's not learned helplessness. It's an accurate read of the grain in front of them. The right EF isn't maximum perceived control -- it's a read that's actually calibrated to the situation, the way a good woodworker doesn't force a straight cut through a knot and call it confidence.

EF is provisionally treated as one dimension for modeling simplicity, but there's no strong reason yet to believe it's really just one thing rather than several related constructs (self-efficacy, perceived control, expectancy, goal appraisal) bundled together. The one claim held with real confidence is the broader one: some dynamic read has to exist, because the fixed grain alone can't explain how the same person behaves so differently across situations.

EF Is Recursive -- This Is Where Learning Enters

EF updates through experience: EF(t) → action(t) → outcome(t) → learning → EF(t+1). Success is evidence a domain is more workable than assumed; failure is evidence it's less workable, or that the cut was wrong for this grain. The stable nine shape how much gets learned from any given outcome -- two people can live through the identical business failure and walk away with completely different lessons, one concluding only "that didn't work," the other extracting the specific structural reason it didn't work and recognizing the same structural risk sitting in three other businesses they're watching. Identical experience, unequal learning yield.

This is also where age belongs in the model, and where it doesn't. Age is a proxy for accumulated iterations through the environment -- but two 50-year-olds can carry radically different amounts of relevant experience, feedback, failure, and deliberate practice behind an identical birthdate.

Some people are 25 years old, twenty-five times in a row. Other people are actually 50, because they spent every one of those years learning and changing instead of running the same year back on a longer loop.

Chronological age isn't the mechanism. Accumulated, interpreted experience is.

05. From Grain to 20,000 Profiles

Every one of the nine stable dimensions is continuous -- there's no such thing as a "type" here, only regions in a nine-dimensional space. But for the sake of being able to talk about it at all, each dimension can be crudely binned into low, medium, or high. Nine dimensions at three bins apiece: 3⁹ = 19,683 -- call it roughly 20,000 coarse grain patterns.

That number is a communication device, not a claim that 20,000 personality types exist in nature, and the combinations won't be evenly populated since the underlying dimensions aren't fully independent of each other. But it's a useful gut check against something like MBTI's 16 types, which most people can beat by simply listing more than 16 kinds of people they personally know. Twenty thousand is coarse enough to be usable and fine enough to be worth taking seriously.

06. Cutting With the Grain: Person-Problem Congruence

This is where the model stops being descriptive and starts being useful. Instead of asking how talented is this person, ask talented cutting which way.

John Taylor Gatto put a blunter version of the same intuition: "genius is as common as dirt." The claim here is narrower and more mechanical than that -- it doesn't assume everyone is a genius, it assumes everyone has a grain, and that apparent talent is inseparable from which cut you're testing it against. A board that splinters on a crosscut can still be perfectly sound wood for a rip cut. The question was never how good the wood is. It's which way you're cutting it.

Every job, business model, or recurring activity demands a specific bundle of things: sustained social contact, tolerance for conflict, repetitive execution, comfort with ambiguity, long stretches of solitary thought, rapid decisions, persuasion, precision, persistence, procedural compliance. A person's grain can be compared against that demand profile, and at least three separate questions fall out of the comparison:

A person can pass Performance Fit and fail Sustainability Fit -- capable of the work, chronically worn down by doing it -- and that distinction is exactly the one most "strengths" assessments skip.

One implication worth taking seriously: job titles are probably too coarse a unit of analysis. Two people with the title "accountant" can spend their careers on completely different activities -- one processing hundreds of standardized returns, the other diagnosing unusual structures under real uncertainty -- with two very different grains suited to each. The more honest unit isn't the job. It's the recurring activity or problem class the job happens to bundle together, and each one is its own cut.

07. When the Business Itself Cuts Against the Owner's Grain

The same logic applies to entrepreneurship: a business is just another recurring set of cuts, made daily, by whoever runs it. When the business's core demands run against the operator's own grain -- constant selling forced onto someone who finds it draining, constant people-management forced onto someone built for heads-down solitary work -- the friction doesn't stay abstract. It shows up in the financials, usually filed under a different name entirely: "undercapitalized," "couldn't scale," "bad product-market fit."

The standard advice given to struggling founders -- get better at cold calls, learn to manage people, network harder -- assumes the business model is fixed and the operator needs to change grain, which isn't actually possible. The alternative is business architecture, not a personality makeover: sorting a business's recurring functions into Perform, Develop, Delegate, Automate, or Avoid, and building around the grain that's actually there instead of the one the owner wishes they had.

This is speculative enough, and consequential enough, that it gets its own full treatment rather than a subsection -- a companion piece, The Mismatch Nobody Diagnoses, is forthcoming, covering the two specific failure chains, real cases, and the obvious objection worth answering before taking this seriously.

08. "Everyone Is World Class at Something"

I've used a version of this line for years, and I don't mean it literally. There are, in fact, perfectly mediocre people. What the model actually supports is a narrower, defensible version of the aphorism: every piece of wood has some grain that's excellent for some cut, even if it's the wrong grain for a different one. Given nine dimensions of real variance and the sheer number of distinct problem classes an economy contains, an unusual number of people are sitting on some uncommon combination of only moderately unusual traits that happens to be extremely well matched to a specific, narrow class of problem -- and once that match exists, specialization and accumulated experience compound whatever edge it started with.

Nobody needs to be in the 99.9th percentile on any single dimension for this to work. They need an uncommon conjunction, paired with the right problem, given enough time to compound. This is the same ground Gatto's line was standing on above, and the honest, hedged version of my own claim is close to his: most people probably have the potential to become genuinely exceptional at some sufficiently specific class of problem; the real bottleneck is finding the cut their particular grain was built for and staying on it long enough for specialization to do its work. That also means a lot of what gets scored as "low capability" in a classroom or a job might actually be a bad cut against a perfectly good grain -- institutions that expose people to a narrow band of demand profiles will systematically misjudge ability outside that band.

09. What's Actually Being Claimed Here, and What Isn't

Most of the individual pieces in this model aren't new. HEXACO, general cognitive ability, Need for Cognition, self-efficacy, and person-environment fit all have real research bases behind them independently. What's actually being proposed is the architecture: the specific compression into nine stable dimensions plus one dynamic read, the recursive loop connecting EF to learning to future EF, Structured Generalization as a possibly-distinct trait, and the extension of all of it into person-problem and operator-business congruence.

None of this should be read as validated science. The individual claims sit at genuinely different confidence levels, and it's worth being specific rather than hand-wavy about that:

The falsification conditions are worth stating plainly, because a framework that can't fail isn't saying anything: if Structured Generalization disappears entirely once you control for ordinary cognitive ability, drop it. If NFC adds nothing once personality and cognition are already in the model, drop it. If person-problem congruence adds no real predictive power over cognitive ability, Conscientiousness, and plain experience, the practical matching claim is substantially weaker than argued here. And if a smaller model predicts just as well as this one, use the smaller model. That last rule matters more than any of the nine dimensions individually -- the point was never to defend nine, it was to find the smallest model that still explains the thing worth explaining.

10. Where This Leaves Me, Personally

Since this is my framework, the honest move is to run it on myself instead of just gesturing at the exercise. Here's my own grain, self-scored -- which comes with the same caveat I raised when I ran a version of this self-scoring exercise in Intellectual Agency: self-report is circular and prone to flattering itself, and this should be read as illustrative, not as data.

• Honesty-Humility: medium, sitting right on the border with high.
• Emotionality: low.
• Extraversion: low.
• Agreeableness: low.
• Conscientiousness: high.
• Openness to Experience: high.
• Abstraction Ability and Structured Generalization: both high -- and this is where my actual outliers live. Most of the six HEXACO scores above sit closer to the boundary of a quartile than out at a real tail. The two intellectual dimensions don't. On standard aptitude measures I test in the 98th-99th percentile, and I'd expect abstraction and structured generalization to sit near there too if anyone ever measured them directly.
• Need for Cognition: high.

Read as a cut, that grain runs toward solitary synthesis -- building a model, finding the structure underneath three unrelated fields, pushing an idea until it's internally consistent -- and away from the sustained interpersonal push that turns an idea into a validated, sold product. Low Extraversion and low Agreeableness don't make that work impossible. They make it exactly the kind of cut Section 6 describes: technically survivable, chronically higher friction than the cut the grain actually favors.

Case in point, and current as of writing this: I've spent the last six weeks traveling heavily and attending four different conferences. I can do that work, and I did. What I want right now is to sit in my basement, read a book, and not talk to anyone -- which is the Friction Fit cost showing up on schedule, not a complaint about the conferences themselves.

Which is the honest admission this section exists to make: I enjoy building this kind of framework and iterating on it. I have close to zero interest in doing the empirical legwork to validate it, or in the sustained selling and relationship-building it would take to commercialize it. If the framework is right, that's not a character flaw to apologize for -- it's Person–Problem Congruence describing its own author in real time, in public, using his own numbers. The honest move isn't pretending I'm about to go run a factor analysis against my own grain. It's naming what cut I'm actually built for, and leaving the rest of it -- explicitly -- as an open, falsifiable claim for someone with a different grain, better suited to that particular cut, to go test.

The Environmental Factor section's ocean thought experiment isn't hypothetical for me either. Asked directly what I'd do dropped in open water with no raft, my honest answer was: tread water as long as possible and hope to get lucky, because there's nothing else meaningfully available to do. That's not the read of someone with low perceived agency -- it's a fully internal locus of control that still updates accurately based on what the actual situation allows, which is closer to a calibrated "can-do attitude" than to textbook locus of control. That distinction is exactly why this framework needed its own term for it instead of reusing an existing one.

11. This Connects to Intellectual Agency

This piece is the direct continuation of Intellectual Agency: Measuring Intelligence Through the AI Corpus, which closed by naming a ten-dimensional model built from HEXACO, locus of control, and the three intellectual-agency dimensions described there. That post used a different metaphor -- raw intelligence as the engine, intellectual agency as the steering wheel -- for the narrower question of capacity versus drive. Grain names the larger architecture both metaphors sit inside: the engine is part of what the grain is made of; the steering wheel is one instance of reading it in the moment.

The model itself has moved since that piece was written: locus of control has become the Environmental Factor, "cross-domain transfer" and "willingness to engage" have separated into Structured Generalization and Need for Cognition as distinct constructs, and Abstraction Ability has taken its own explicit seat alongside them.

The core instrument from that piece -- reading abstraction, transfer, and engagement out of an AI chat corpus -- still maps directly onto three of the nine dimensions here, and remains one of the more practical ways to actually estimate where a given person's grain runs without a formal instrument.

12. Where This Meets Tax Sherpa

Grain isn't a Tax Sherpa framework, and it doesn't need a hard pitch bolted onto the end of it to earn its place here. But it isn't disconnected either.

A business built to run with its owner's grain, per Section 7 and The Mismatch Nobody Diagnoses, is a business more likely to actually survive its own early years and grow -- because the owner isn't spending a fixed, finite supply of energy fighting their own configuration every single day just to keep the doors open. Growth is exactly the point where tax planning stops being an afterthought and starts being a real lever: more revenue, more entities, more structural decisions, more at stake in getting it right the first time instead of the fourth.

Grain is a framework for building a business that gets that far. Tax Sherpa is for the business that did.

References

• Cacioppo, J. T., & Petty, R. E. (1982). The need for cognition. Journal of Personality and Social Psychology, 42(1), 116-131. https://doi.org/10.1037/0022-3514.42.1.116
• Judge, T. A., & Bono, J. E. (2001). Relationship of core self-evaluations traits -- self-esteem, generalized self-efficacy, locus of control, and emotional stability -- with job satisfaction and job performance: A meta-analysis. Journal of Applied Psychology, 86(1), 80-92. https://doi.org/10.1037/0021-9010.86.1.80
• Kristof-Brown, A. L., Zimmerman, R. D., & Johnson, E. C. (2005). Consequences of individuals' fit at work: A meta-analysis of person-job, person-organization, person-group, and person-supervisor fit. Personnel Psychology, 58(2), 281-342. https://doi.org/10.1111/j.1744-6570.2005.00672.x
• Lee, Y., Berry, C. M., & Gonzalez-Mulé, E. (2019). The importance of being humble: A meta-analysis and incremental validity analysis of the relationship between honesty-humility and job performance. Journal of Applied Psychology, 104(12), 1535-1546. https://doi.org/10.1037/apl0000421
• Oh, I.-S., Le, H., Whitman, D. S., Kim, K., Yoo, T.-Y., Hwang, J.-O., & Kim, C.-S. (2014). The incremental validity of honesty-humility over cognitive ability and the Big Five personality traits. Human Performance, 27(3), 206-224. https://doi.org/10.1080/08959285.2014.913594
• Schmidt, F. L., & Hunter, J. E. (1998). The validity and utility of selection methods in personnel psychology: Practical and theoretical implications of 85 years of research findings. Psychological Bulletin, 124(2), 262-274. https://doi.org/10.1037/0033-2909.124.2.262
• Zettler, I., Thielmann, I., Hilbig, B. E., & Moshagen, M. (2020). The nomological net of the HEXACO model of personality: A large-scale meta-analytic investigation. Perspectives on Psychological Science, 15(3), 723-760. https://doi.org/10.1177/1745691619895036