Chapter 16

The Depth Deficit Index

Measuring What the Dashboards Miss

“When a measure becomes a target, it ceases to be a good measure.”

— Goodhart’s Law, as phrased by Marilyn Strathern

Everything in this book so far has asked you to notice something invisible. The deficit hides by design: output stays impressive while capability erodes underneath, and the fluency illusion — remember, it is self-concealing — guarantees that the people losing depth fastest feel it least. Against an invisible problem, good intentions lose. What wins, reliably, is measurement. A large body of behavioral research — including a meta-analysis of more than a hundred studies led by Benjamin Harkin — finds that simply monitoring progress toward a goal materially increases the odds of reaching it. Not because the number is magic, but because measurement converts drift into a decision.

So this chapter gives the book’s argument a number: the Depth Deficit Index. It is a structured self-assessment you can take in ten minutes, retake quarterly, and — in its organizational form — run across a team. It will not tell you anything the previous seventeen chapters haven’t argued. What it will do is tell you where you specifically stand, which dimension is eroding fastest, and whether ninety days of practice moved anything. That is the difference between reading a book and adopting one.

The Five Dimensions: DEPTH

The Index measures five dimensions, one for each letter, each anchored in the evidence you have already met.

D — Direct Engagement. Do you think before you prompt? The sequence rule as a habit: your own position formed before the tool arrives. This is the dimension the MIT essay-writers lost first.

E — Evaluation. Can you independently judge output — verify the load-bearing facts, catch the confident error, defend the conclusion without the tool? The dimension that failed in the Avianca courtroom and separates the jagged-frontier winners from the ones asleep at the wheel.

P — Presence. The human infrastructure: undivided conversations, a mentor who tells you the truth, someone you are teaching. The thirty-four-fold channel, kept open.

T — Tenacity of Attention. The daily deep block, the ability to stay with a hard problem past the reach-for-the-tool reflex. Gloria Mark’s forty-seven seconds, resisted on purpose.

H — Held Capability. The Question Six test made operational: what remains when the tools are removed? The dimension the endoscopists discovered too late — and the only one that ultimately matters, because it is what the other four are protecting.

The Instrument

Score each statement from 0 to 5: Never (0), Rarely (1), Sometimes (2), Often (3), Usually (4), Always (5). Answer for the last thirty days, not your best month ever. The instrument only works at the speed of your honesty.

D — Direct Engagement

1. On work that matters, I spend meaningful time — fifteen minutes or more — forming my own thinking before opening an AI tool.

2. I write my own rough position or draft before prompting on significant tasks.

3. When a problem is hard, my first move is to sit with it rather than immediately delegate it.

4. I can name a difficult task from this week that I deliberately completed without AI because the struggle builds capability.

E — Evaluation

5. I personally verify the load-bearing facts before AI-assisted work ships under my name.

6. When AI output disagrees with my expectation, I investigate the disagreement rather than deferring — in either direction.

7. I can defend the reasoning of my AI-assisted deliverables without reference to the tool.

8. I actively hunt for what is wrong with a polished output before accepting it.

P — Presence

9. I hold regular, undivided conversations — device down, fully there — with the key people in my working life.

10. At least one person tells me uncomfortable truths about my work, and I have heard one in the last quarter.

11. When the stakes are relational, I take the conversation live and in person rather than asynchronous.

12. I am genuinely mentoring someone — recurring engagement, not occasional advice.

T — Tenacity of Attention

13. I complete at least one 60–90 minute uninterrupted deep-work block on most working days.

14. I can work thirty minutes on a hard problem without switching tasks or reaching for a tool.

15. Deep blocks exist on my calendar and survive contact with other people’s demands.

16. I notice the reach-for-the-tool reflex when it fires, and I can delay it at will.

H — Held Capability

17. Within the last month, I completed one significant task in my domain fully unassisted, as a deliberate capability check.

18. If my AI tools vanished tomorrow, I could still perform the core of my role to a professional standard.

19. I could teach the fundamentals of my domain from my own understanding, without lookups.

20. Compared to a year ago, my unassisted skills are stable or improving — not quietly eroding.

Scoring and Interpretation

Add your twenty answers for a Depth Score out of 100. Your Depth Deficit Index is 100 minus that score — so a higher DDI means a larger deficit. Then read your band.

DDI 0–20 — Deep practice. You are the second kind of professional. Maintain, and take the organizational version to the people you lead.

DDI 21–40 — Early drift. Nothing visible has broken; one or two dimensions are quietly slipping. Tighten the lowest dimension now, while ninety days is all it takes.

DDI 41–60 — Active erosion. The pattern this book describes is underway in your working life. A structured intervention — below — is warranted this quarter, not someday.

DDI 61–80 — Advanced deficit. Under novel conditions — the template-free room, the tool-less crisis — your performance is already compromised, whether or not it has been tested yet.

DDI 81–100 — Critical. The capability the tools are masking is largely gone. The honest response is not despair but a rebuild: the same practices, applied with the urgency the number has finally made visible.

Two warnings from measurement science, before you trust your number. First, self-report inflates — everyone believes they verify more, focus longer, and depend less than they do. So anchor the score with evidence: pull your calendar and count the deep blocks that actually happened; find the position briefs that actually exist for your last ten significant tasks; name the date of your last fully unassisted piece of work. Where the evidence contradicts the self-rating, believe the evidence and rescore. Second — Goodhart’s Law, this chapter’s epigraph: the moment a measure becomes a target, people optimize the measure instead of the reality. The DDI is a mirror, not a weapon. The instant an organization ties it to ratings or pay, it will be gamed into uselessness.

The Organizational Index

Businesses run on dashboards, and nothing on any of them measures whether the organization’s human capability is compounding or hollowing behind its tools. The organizational DDI closes that gap with two readings taken together. The first is the aggregate: individual Depth Scores across a team, collected anonymously, reported only as a distribution — anonymity is not a courtesy, it is what keeps the data honest. The second is a five-condition audit of the environment, because individual depth cannot survive an organization designed against it. Rate each condition 0 to 4, and — critically — have leadership and staff rate them separately, because the gap between the two ratings is usually the most diagnostic number in the whole exercise.

One: sequence norms — do standard workflows require a human position before AI drafting, or reward whoever prompts fastest? Two: verification standards — is there an explicit ownership rule for load-bearing facts, with time budgeted for it? Three: attention infrastructure — do protected deep blocks exist and survive, or does the interruption culture eat them? Four: capability maintenance — are there scheduled unassisted drills, the knowledge-work equivalent of the manual flight hours aviation finally mandated? Five: mentorship and presence systems — is anyone pairing juniors with seniors on real work, and does the organization still convene unrecorded rooms?

Treat the resulting number as a leading indicator — the technologist’s framing. Output metrics tell you how the organization performed under conditions the tools handled. The O-DDI predicts how it will perform under the conditions they won’t: the novel problem, the outage, the situation outside the frontier. Toyota’s craftsmen were a leading-indicator investment. So were aviation’s manual-flight mandates. The organizations that track depth before it fails will own the decade over the ones that discover the deficit the way Air France did.

Your dashboards measure everything your tools produce. The Index measures what remains of the people producing it.

A Worked Example: One Professional’s First Cycle

Here is the Index in use, start to finish. A senior analyst takes the twenty statements on a Sunday evening and scores an honest-feeling 71 — DDI 29, early drift. Then the evidence check, which is where the instrument earns its keep: the calendar shows two deep blocks actually completed in the past two weeks, not the ‘most days’ her answers claimed; the last fully unassisted piece of work she can name is seven months old. Rescored against evidence: Depth Score 58, DDI 42 — active erosion. The eleven-point gap between felt and actual is not a character flaw. It is the fluency illusion, personally measured, and almost everyone’s first honest scoring finds one.

Her subscores locate the problem: Direct Engagement 13/20, Evaluation 14/20, Presence 12/20 — but Tenacity 9/20 and Held Capability 10/20. The cycle targets Tenacity, the lowest, and only Tenacity: ninety days of the daily block, run on Chapter 3’s protocol, with the reach-tally on the card. Nothing else changes — which is what makes the change sustainable.

Ninety days later the retake shows Tenacity at 14/20 and — unplanned — Held Capability at 12, because protected attention quietly funded deeper engagement everywhere. New DDI: 33. Still drifting, now visibly reversing, and the next cycle’s target is already obvious. Two numbers, one quarter apart, did what no amount of reading could: they converted a book’s argument into her own trend line — and a trend line, unlike an intention, demands an answer every ninety days.

An Honest Note on This Instrument

Intellectual honesty, one last time: the DDI is a structured self-assessment built on the research in this book, not a clinically validated psychometric instrument. Its value is directional and personal — your trajectory across quarters matters far more than any single score, and your dimension subscores matter more than the total. Used that way, with the behavioral evidence check, it does the one thing an invisible problem cannot survive: it makes the drift visible, regularly, in a number you cannot unsee. Measurement will not build your depth. It will make it impossible to pretend you are building it when you are not — and for most professionals, that is the missing piece.

The Practice

1. Take the Index today — twenty statements, ten minutes, evidence check included. Record the five subscores, not just the total. Put the retake on your calendar for ninety days from now before you do anything else.

2. Run one 90-Day Depth Cycle on your lowest dimension, exactly as prescribed. One dimension. Ninety days. Then re-measure.

3. If you lead a team, run the organizational version this quarter — anonymous individual scores plus the five-condition audit, leadership and staff rating separately. Present the gap between the two ratings at your next leadership meeting, and let it start the conversation this book has been preparing you to lead.