Diagnosis: Company
Part of the cognition series. Applies the diagnostic from Diagnosis LLM to public companies.
Every public company is a self-recursive loop. Revenue in, product out, revenue in. The loop either compounds or decays. The Natural Framework diagnoses which cells are functional and which are broken. The Handshake predicts the consequence: broken contracts compound.
Warren Buffett calls the functional loop a “moat.” The framework names its six walls.
The six cells of a company
| Step | Contract | Healthy | Dying |
|---|---|---|---|
| Perceive | Reads environment accurately | Customer research, market signals, competitive intelligence flow in | Ignores signals, falsifies internal data, executive bubble |
| Cache | Stores and retrieves institutional knowledge | Knowledge survives turnover, systems of record work | Knowledge walks out the door with employees |
| Filter | Rejects what doesn’t matter | Says no to most projects, kills bad bets early | Tries everything, no focus, no kills |
| Attend | Ranks priorities with diversity guarantee | Multiple revenue streams that repel each other | One product, one market, one bet |
| Consolidate | Updates procedures from experience | Post-mortems change process, playbook evolves | Runs the playbook from ten years ago |
| Remember | Persists durable output for next cycle | Brand, patents, culture, ecosystem compound | Each quarter starts from zero |
The dim cell test: count the cells that exist but don’t meet their contract. More dim cells, shorter time horizon before the loop degrades.
Attend is the moat
Functional Attend means diverse revenue streams that suppress each other through lateral inhibition. iPhone doesn’t compete with Mac. Insurance doesn’t compete with railroads. When one region contracts, the others persist.
Broken Attend is one bet. Film. Phones. Search ads. The single axis inflates until it’s the only thing the company knows how to do. When the axis shifts, there’s no diversity to fall back on. Kodak, Nokia, every conglomerate that “focused on its core business” until the core rotted.
The Handshake’s diversity guarantee: survivors are dissimilar. Healthy business lines repel. Unhealthy ones cluster.
The only cell that matters twice
Can the company fix itself? That question lands on one cell: Consolidate, the procedure that writes procedures.
Intel under Grove: “only the paranoid survive” was a consolidation event. Near-death changed how the company attended to threats. Christensen called the failure mode the innovator’s dilemma. The framework names the broken cell: stale Consolidate runs yesterday’s playbook in tomorrow’s market.
When Consolidate is the broken cell, the fix must come from outside the loop: new management, a crisis, an activist investor who reads the diagnosis.
SOAP note format
Same structure as Diagnosis LLM. A company may run multiple stacked pipelines: each subsidiary has its own six cells, and the holding company runs a meta-pipeline on top. Diagnose each stack separately. The scores are generated via double loop between two models, not one-shot. For each company:
Subjective. What management claims about each cell. Earnings calls, shareholder letters, public statements.
Objective. What the filings show. Revenue concentration (Attend), R&D allocation (Consolidate), employee turnover (Cache), market share trajectory (compounding or decaying).
Assessment. Which cells are functional, shallow, dim, stale, or broken. Whether Consolidate can self-correct. The death conditions: broken step, closed loop, or decaying input.
Plan. Which cell to strengthen. Whether the fix requires internal initiative or external intervention.
Three examples
Berkshire Hathaway: all six cells functional
The framework incarnate. Buffett reads every annual report, talks to managers, visits factories. Each subsidiary caches its own institutional knowledge, decentralized, surviving turnover at the top. The filter is brutal: 99% of deals get a no. One criterion.
Insurance, railroads, energy, candy, furniture, jewelry. Diverse. Repelling. No business line competes with another. That’s functional Attend. And the playbook updated: young Buffett bought cigar butts, Munger taught him to buy quality, the procedure changed. He never sells. Holdings compound across decades. The output of each cycle feeds the next.
Assessment: all cells functional. The loop compounds. The moat is the pipeline itself.
Risk: Consolidate is personality-dependent. Greg Abel inherits the loop. Whether he inherits the Consolidate is the open question.
Kodak: broken Attend, stale Consolidate
Kodak invented the digital camera in 1975. Perceive worked. Deep institutional knowledge, good research judgment. Three healthy cells.
Then the pipeline broke. Every revenue stream clustered around film. One axis, one chemistry, no repulsion between business lines. Attend was broken. And the playbook from the 1960s never updated: “Razor and blades. Sell cameras cheap, make money on film.” The digital signal arrived through Perceive and died at Consolidate.
Assessment: the company saw the future and couldn’t restructure to meet it. Stale Consolidate couldn’t update broken Attend. The fix required external intervention (bankruptcy, 2012).
The framework in 2000: Attend is single-axis (film revenue > 70%), Consolidate hasn’t updated since the 1980s. When the axis shifts, no diversity and no procedure to create it. Structural short.
Apple: functional but fragile
Deep design research, supply chain intelligence, developer ecosystem signals. The institutional design language survives individual departures. Apple kills more products than it ships. “Focus is about saying no.” Three strong cells.
iPhone, Mac, iPad, Wearables, Services. Five axes, but iPhone is > 50% of revenue. The diversity guarantee holds; the weight makes it fragile. Under Jobs, Consolidate was the strongest cell. He changed the playbook repeatedly: computers, iPod, iPhone, Services. Under Cook, the procedures optimize. They don’t reinvent. Switching costs lock in the next cycle’s Perceive, but whether Cook can write a new playbook is the open question.
Assessment: five of six cells functional. Consolidate is the watch cell. Operational Consolidate (improve the process) keeps the loop spinning. Strategic Consolidate (change what we attend to) is what writes the next playbook.
The investment rubric
- Read the 10-K. Score each cell: functional, shallow, dim, stale, broken.
- Check Attend diversity. Revenue concentration by segment. Do the segments repel or cluster?
- Check Consolidate activity. Has strategy changed in the last five years? In response to evidence (functional) or crisis (shallow)?
- Check for the death conditions. Broken step: which cell? Can it self-correct? Closed loop: is the company ignoring its market? Decaying input: is the market itself shrinking?
- The dim cell count is the diagnosis. Zero dim cells: compounder. One or two: watch list. Three or more: the loop is decaying. If Consolidate is one of the dim cells: the company cannot self-correct. That’s a structural position.
Count the dim cells. The loop either compounds or decays.
Written via the double loop. More at pageleft.cc.