Your domain intuition imports the wrong physics

Everyone assumed golf-ball physics

Every conversation we'd ever had about Quantum Caddy's throw physics included spin, and every one of them treated spin the way golf treats it. Spin means lift. Backspin holds a golf ball in the air and moves where it lands, so naturally spin would shape where a cornhole bag lands, and naturally a serious training product would measure it carefully. I believed this. So did everyone I talked the product through with, because the intuition arrives pre-installed: cornhole is a projectile sport, golf is the famous projectile sport, and golf says spin is king.

Then the July 5th audit priced it. The session was auditing our throw simulator, a training-data engine we've since deprecated, and as part of correcting the physics it worked out what spin actually does to a regulation bag in flight. Magnus lift, the golf mechanism, tops out around 3 inches of effect on the landing point. Whether the bag flies flat or on edge, its presentation to the airflow, moves the landing point 23 inches at an identical launch.

Both mechanisms are real. Both are "spin matters". But they differ by 7× in the units that count, and the error wasn't in how much spin matters. It was in which mechanism does the mattering. A bag at cornhole rotation rates never gets anywhere near golf-ball spin ratios; what the rotation actually changes is the cross-section the air sees.

The practical cost, had nobody priced it: the next tranche of measurement effort was pointed at spin rate. We'd have built the rig, run the sessions, and come home with a beautifully measured quantity worth 3 inches, while the quantity worth 23 went unmeasured. I'd have spent our measurement budget on the wrong quantity and probably concluded from the weak results that physics-based prediction doesn't help.

Intuitions get borrowed from the nearest famous domain

The pattern runs well beyond bags. Practitioners in an unglamorous domain import their mechanisms from the nearest famous one, because the famous domain has the books, the vocabulary, and the pre-built explanations, and the unglamorous domain has none of those.

A sales-led B2B product gets run on viral-growth intuitions, with the founders watching referral loops and K-factor, when the mechanism that actually moves revenue in their regime is rep capacity and sales cycle time. A crypto portfolio gets built on diversification intuitions imported from equities, where the mechanism assumes assets that don't move together; in a market where everything is correlated to one coin, the mechanism never operates, however sound it is on an exchange full of unrelated businesses. And golf-ball physics gets applied to a one-pound fabric bag.

What makes borrowed mechanisms dangerous is that they're directionally plausible. Spin really does affect a bag's flight. Word of mouth really does bring some signups. Holding many assets really is different from holding one. The sign is right, so the intuition survives every casual test anyone runs on it, and nobody prices the magnitude, because the intuition doesn't feel like a guess. It feels like knowledge, imported from a domain where it genuinely was knowledge.

Price both mechanisms in the same units before the intuition spends money

An intuition is harmless while it's just talk. It becomes expensive the moment it steers a spend: a measurement campaign, an instrumentation build, a feature quarter, a hiring plan. That is the moment to stop and make the agent, or the analysis, or the intern with a spreadsheet, price both candidate mechanisms in the same units.

Same units is the discipline. "Spin matters" and "presentation matters" can coexist forever as opinions. "3 inches of landing movement" and "23 inches of landing movement" cannot; one of them just lost the argument about where the budget goes. The exercise forces the mechanisms to compete in the currency you're about to spend.

The pricing didn't require new data, which is the part that surprised me most. The QC session did it in one afternoon with dimensionless numbers and literature priors: a spin ratio showed the bag never rotates fast enough for lift to be meaningful, and published drag behavior for flat-versus-edge presentation put a number on the cross-section effect. No experiment and no new dataset, just an afternoon of arithmetic against a measurement campaign that would have cost weeks and returned 3 inches of value.

If the same afternoon had come back saying lift dominates, we'd have measured spin rate with confidence instead of by inheritance. The pricing pass wins either way. The only thing it kills is spending on an unpriced mechanism.

Sometimes the import is right

The honest limits. Borrowed intuitions are often correct, which is why the discipline is pricing, not reflexive inversion. Deciding the famous domain must be wrong is the same failure with the sign flipped, and it is just as unpriced. Plenty of imports are load-bearing: projectile motion itself came to cornhole from every other thrown-object domain, and it holds fine.

The pricing pass has limits of its own. Ours ran on priors and dimensionless arguments, not on measured throws. A 7× gap sits comfortably outside that kind of error, but a 1.5× gap would not have settled anything, and the honest move there is to measure the cheaper mechanism first rather than declare a winner from priors. A pricing exercise can also only compare the mechanisms someone thought to name; if the true driver is a third mechanism nobody listed, both prices are beside the point. The afternoon buys you an allocation, an order-of-magnitude ranking of named candidates. It does not buy certainty, and it is not a substitute for the measurement it aims.

Apply this

Do the written version today. It takes less time than defending the intuition in your next planning meeting will.

• Write down the three intuitions currently steering your roadmap. Not the official strategy, the intuitions: the sentences people say in meetings that end debates, the way "spin matters" ended ours.
• Next to each, name the famous domain it was borrowed from. Growth intuitions tend to come from consumer virality, risk intuitions from equity portfolios, physical intuitions from whichever sport or system got studied first. If you can name the source domain, you can ask whether your regime actually matches it.
• For each intuition, name at least one competing mechanism that would produce the same surface behavior. This is the hard step and the valuable one. Ours was hiding inside the same word, "spin", which is exactly why nobody had separated the two.
• Estimate what pricing the candidates in the same units would cost. In our case it was an afternoon with priors and dimensionless numbers. The estimate alone tells you whether you've been avoiding a cheap question with an expensive default.
• Gate the next spend on the price. If an unpriced intuition is about to direct a measurement campaign, an instrumentation build, or a feature quarter, price it first and let the loser of the comparison off the roadmap.