Skip to content
← the lab

All 24 equations

Every number the course uses, with the book's own traffic-light benchmark printed next to it. Each one opens on a real worked example, so you can see what the answer looked like for somebody else before you try it on yourself.

None of these care what you meant. That's the entire reason they're useful. Every calculator is free to embed in your own pages, with no tracking inside the frame.

24 of 24 equations

#1 · Ch 1R_ext

Extinction Rate

How fast generic work goes extinct. Above 5 you're landfill the moment tooling catches up.

Rext=LinesgenericLinesauthentic×AIvelocityR_{ext} = \frac{Lines_{generic}}{Lines_{authentic}} \times AI_{velocity}

🟢 < 2 · 🟡 2–5 · 🔴 ≥ 5 (you are landfill the moment tooling catches up).

Example: A four-person internal tools teamTaught in lesson 02open →
#2 · Ch 1A_hl

Authenticity Half-Life

How long your work resists being replicated by a machine. Higher is safer.

Ahl=Taste+Context+CraftReplicabilityA_{hl} = \frac{Taste + Context + Craft}{Replicability}

🟢 ≥ 4 (months→years) · 🟡 1.5–4 (weeks→months) · 🔴 < 1.5 (days).

Example: A freelance analyst's monthly reportTaught in lesson 02open →
#3 · Ch 2R_t

Theater Ratio

(Press releases + demo days) ÷ production deploys. Count deploys, not announcements.

Rt=Press_Releases+Demo_DaysProduction_DeploymentsR_t = \frac{Press\_Releases + Demo\_Days}{Production\_Deployments}

🟢 < 1 · 🟡 1–3 · 🔴 > 3 (all talk, no ship).

Example: One quarter of an innovation officeopen →
#4 · Ch 2F_g

Vendor Gravity Force

Sunk cost × political capital ÷ migration effort². You can't fight the well; orbit it.

Fg=Sunk_Cost×Political_CapitalMigration_Effort2F_g = \frac{Sunk\_Cost \times Political\_Capital}{Migration\_Effort^2}

🟢 < 1 (movable) · 🟡 1–5 · 🔴 ≥ 5 (do NOT pitch replacement).

Example: A 2009 core policy ledgerTaught in lesson 03open →
#5 · Ch 2E_tax

Entropy Tax

Maintenance × e^(λt). The status quo is never free — read the curve, not a point.

Etax=Maintenance×eλtE_{tax} = Maintenance \times e^{\lambda t}

read the curve, not a point. 🟢 flat · 🟡 bending · 🔴 hockey stick. (Numeric, as a multiple of base Maintenance: 🟢 < 1.5× · 🟡 1.5–3× · 🔴 > 3×.)

Example: A 2014 batch settlement jobTaught in lesson 03open →
#6 · Ch 2μ_leg

Legacy Friction Coefficient

Years since last update ÷ pages of documentation. Obscure + terrifying = nobody will touch it.

μleg=Years_Since_Last_UpdatePages_of_Documentation\mu_{leg} = \frac{Years\_Since\_Last\_Update}{Pages\_of\_Documentation}

🟢 < 0.5 (low) · 🟡 0.5–2 (mid) · 🔴 > 2 (high — obscure + terrifying = nobody will touch it).

Example: A claims-routing service nobody touchesTaught in lesson 03open →
#7 · Ch 2D_comp

Technical Debt Compound Rate

Bugs found ÷ bugs fixed. Above 1.0 the team is underwater.

Dcomp=Bugs_FoundBugs_FixedD_{comp} = \frac{Bugs\_Found}{Bugs\_Fixed}

🟢 < 1.0 · 🟡 1.0–1.5 · 🔴 > 1.5 (the team is underwater).

Example: A six-person platform team's trackerTaught in lesson 03open →
#8 · Ch 2F_cp

Consensus Paralysis Factor

Communication paths in a decision group — n(n−1)/2. Why big rooms can't decide.

Fcp=n(n−1)2F_{cp} = \frac{n(n-1)}{2}

🟢 ≤ 3 (n≤3) · 🟡 6–15 (n 4–6) · 🔴 ≥ 21 (n ≥ 7).

Example: The weekly release-planning meetingTaught in lesson 03open →
#9 · Ch 3V_curator

Curator Value Function

(Signal selected ÷ noise available) × context depth. Sell the taste layer, not the generator.

Vcurator=SignalselectedNoiseavailable×ContextdepthV_{curator} = \frac{Signal_{selected}}{Noise_{available}} \times Context_{depth}

🟢 ≥ 5 · 🟡 1.5–5 · 🔴 < 1.5. Explodes as noise → ∞, **only if** context is real; context_depth ≈ 0 = a Top-10 listicle.

Example: A security analyst's Friday emailTaught in lesson 04open →
#10 · Ch 3T_authentic

Authenticity Token

Human trust × 1/AI-substitutability. Prove you have context AI can't fake — that's the whole moat.

Tauthentic=Trusthuman×1SubstitutabilityAIT_{authentic} = Trust_{human} \times \frac{1}{Substitutability_{AI}}

🟢 ≥ 30 (high trust / low substitutability — a sommelier's pick) · 🟡 10–30 · 🔴 < 10 (AI-washable).

Example: Eleven years inside pharma quality systemsTaught in lesson 04open →
#11 · Ch 4RT

Risk Tax

(P_fail × cost of disaster) + (panic × career cost). The number running in their amygdala.

RT=(Pf×Cd)+(Tpanic×Ccareer)RT = (P_f \times C_d) + (T_{panic} \times C_{career})

🟢 < 1.5× · 🟡 1.5–3× · 🔴 > 3× (career risk > any upside; walk).

Example: A $500K pilot in front of a VP of OpsTaught in lesson 04open →
#12 · Ch 4DF

Decision Fatigue

Decisions made ÷ glucose remaining. Why the 4pm 'no' is automatic.

DF=DecisionsmadeGlucoseremainingDF = \frac{Decisions_{made}}{Glucose_{remaining}}

🟢 < 5 (before 11 AM) · 🟡 5–<10 (post-lunch) · 🔴 ≥ 10 (after 2 PM — default answer is "no").

Example: The 4pm meeting you asked forTaught in lesson 04open →
#13 · Ch 5D_gray

Dorian Gray Index

(Dark-pattern revenue ÷ trust eroded) × time. Trust erodes exponentially, not linearly.

Dgray=Revenuedark-patternTrusteroded×TimeD_{gray} = \frac{Revenue_{dark\text{-}pattern}}{Trust_{eroded}} \times Time

🟢 < 1 · 🟡 1–5 · 🔴 > 5 (the portrait is becoming visible; collapse incoming).

Example: A four-screen cancellation flowTaught in lesson 05open →
#14 · Ch 5S_soul

Soul Score

Σ(human impact × honest intent). Intent is binary: +1 genuine, −1 exploitative. The Mirror Test, quantified.

Ssoul=∑features(Impacthuman×Intenthonest)S_{soul} = \sum_{features}(Impact_{human} \times Intent_{honest})

🟢 > +5 (net positive) · 🟡 −5 to +5 (mixed — fix it) · 🔴 < −5 (net negative; fix the product before you read another page).

Example: Scoring one year of shipped featuresTaught in lesson 05open →
#15 · Ch 6R_r

Reversibility Ratio

Blast radius ÷ rollback speed. Build the Kill Switch before the feature.

Rr=ImpactblastSpeedrollbackR_r = \frac{Impact_{blast}}{Speed_{rollback}}

🟢 ≤ 1 (ship it) · 🟡 1–3 (feature-flag it) · 🔴 > 3 (canary first / do not deploy).

Example: A change to the pricing engineTaught in lesson 06open →
#16 · Ch 7V_w

Wedge Velocity

(Acute pain × trust increment) ÷ adoption friction. If adoption needs procurement, your wedge is blunt.

Vw=Painacute×TrustincrementFrictionadoptV_w = \frac{Pain_{acute} \times Trust_{increment}}{Friction_{adopt}}

🟢 ≥ 4 (viral) · 🟡 1–4 · 🔴 < 1 (it's a platform pretending to be a wedge).

Example: Ninety minutes, three schedulers, every morningTaught in lesson 07open →
#17 · Ch 8GT

Good Trouble Coefficient

(Impact × values-alignment) ÷ (career risk + bureaucratic friction). Structural, not sentimental.

GT=Impactchange×AlignmentvaluesRiskcareer+FrictionbureaucracyGT = \frac{Impact_{change} \times Alignment_{values}}{Risk_{career} + Friction_{bureaucracy}}

🟢 ≥ 4 (go) · 🟡 2–4 (find a Shadow Sponsor) · 🔴 < 2 (vanity — wrong terrain).

Example: The reconciliation that pays suppliers twiceTaught in lesson 08open →
#18 · Ch 8I_insurgent

Insurgent Index

Problems solved quietly ÷ credit claimed publicly. Give the credit away — your manager becomes your shield.

Iinsurgent=Problems_SolvedquietlyCredit_ClaimedpubliclyI_{insurgent} = \frac{Problems\_Solved_{quietly}}{Credit\_Claimed_{publicly}}

🟢 ≥ 5 (invisible and unstoppable) · 🟡 2–5 (effective but watch the ego) · 🔴 < 2 (you're building a LinkedIn, not a movement).

Example: Eighteen months, honestly countedTaught in lesson 08open →
#19 · Ch 9L_legacy

Legacy Liability

P × (1+r)^t. The cheapest option always looks expensive today. Show them what doing nothing costs.

Llegacy=P×(1+r)tL_{legacy} = P \times (1 + r)^t

🟢 < 1.15× · 🟡 1.15–1.40× · 🔴 ≥ 1.40× (and read the 3/5/10-year projection — anything that doubles inside 5 years is RED).

Example: The cost of changing nothingTaught in lesson 09open →
#20 · Ch 9S_fc

Self-Funding Coefficient

(Human cost − AI cost) ÷ pilot cost. Above 10 the CFO is negligent not to buy.

Sfc=Costhuman−CostAICostpilotS_{fc} = \frac{Cost_{human} - Cost_{AI}}{Cost_{pilot}}

🟢 ≥ 2.5 (pays for itself) · 🟢🟢 ≥ 10 (CFO is negligent not to buy) · 🟡 1–2.5 · 🔴 < 1 (don't pitch the CFO yet).

Example: Three contractors and an invoice queueTaught in lesson 09open →
#21 · Ch 10Score

Zombie Scorecard

(Pain Owner ×5) + (Timeline ×3) + (Budget ×5), each 0–5. Is this deal alive, or are you selling to a corpse?

Score=(Pain_Owner×5)+(Timeline×3)+(Budget_Code×5)Score = (Pain\_Owner \times 5) + (Timeline \times 3) + (Budget\_Code \times 5)

🟢 45–65 (live) · 🟡 25–44 (one vital missing) · 🔴 0–24 (corpse — send the Breakup Email).

Example: The deal that's been closing for five monthsopen →
#22 · Ch 10I_v

Vaporware Index

Adjectives in the pitch ÷ live demo minutes. Works both ways — score yourself.

Iv=AdjectivespitchLive_Demo_MinutesI_v = \frac{Adjectives_{pitch}}{Live\_Demo\_Minutes}

🟢 < 1 · 🟡 1–3 · 🔴 > 3 (slides, not software).

Example: Scoring your own last pitchTaught in lesson 09open →
#23 · Ch 11D_audit

Audit Defense Score

(Certifications × scope) ÷ auditor fear. Make saying NO riskier than saying YES.

Daudit=Certifications×ScopeAuditor_FearD_{audit} = \frac{Certifications \times Scope}{Auditor\_Fear}

🟢 ≥ 4 (rubber stamp) · 🟡 1.5–3.9 · 🔴 < 1.5 (rejection).

Example: Security review, and a nervous auditorTaught in lesson 10open →
#24 · Ch 12T_r

Trust/Risk Ratio

(Traffic volume × uptime) ÷ incident severity. Run silent until the math is undeniable, then push the alert.

Tr=Volumetraffic×TimeuptimeSeverityincidentsT_r = \frac{Volume_{traffic} \times Time_{uptime}}{Severity_{incidents}}

🟢 > 10,000 (ready for primary) · 🟢🟢 > 100,000 (System of Record) · 🔴 < 100 (stay in shadow).

Example: Running silent in one departmentTaught in lesson 11open →