Haute Lumière

Commerce · II.09 · MMXXVI · daylight

La Bourse  /  Volume II  /  Nº II.09  /  Workbook — the Gainshare employee

WORKBOOK — THE LUMINOUS GAINSHARE EMPLOYEE

Chapter II.09 · Evolution and Economic Selection

For the person working inside a gainshare arrangement — where a defined share of verified improvement returns to the people who created it. This chapter is about selection, and a gainshare is a selection environment with your name in it. What follows is how to read it, how to compute your own terms, and what to ask for.

---

WHY THIS CHAPTER IS YOURS

A gainshare has four parts: a baseline, a measure, a share, and a period with a verifier. Chapter II.09 says that a population moves toward whatever covaries with persistence, at a rate set by variance and transmission fidelity.

Put those together and the consequence is direct. Your scheme's measure is a fitness function, and it is selecting a population of working practices — yours — at a computable rate. Practices that raise the measure carry weight forward. Practices that do not, however good they are, quietly disappear from the floor over a few cycles.

So the question is not whether your scheme is generous. It is: what is it selecting for, at what boundary, and how fast? You can answer all three with arithmetic you already have.

---

PART ONE — DISCOVERY

Days 1–30: read the fitness function you are inside

Exercise 1.1 — The five questions (2 hours)

From the scheme document, in writing:

1. What exactly is the measure, as a formula?
2. At what boundary is it computed — the individual, the shift, the team, the site, the company?
3. What is the share percentage, and is it of gross or net improvement?
4. What is the period, and who verifies?
5. What happens to the baseline once a gain is realised?

Question two is the one this chapter adds, and it is the one almost nobody asks. A scheme measured at the individual level and a scheme measured at the team level select for entirely different behaviour from identical people.

Exercise 1.2 — The Muir test (1 hour)

William Muir's hens were selected two ways. Keep the offspring of the best individual birds and you get superb individual layers who are also efficient aggressors — mortality in those cages ran at 68.0 percent. Keep the offspring of the best cages and six generations later mortality is 8.8 percent and eggs per hen housed have gone from 91 to 237, a rise of 160.4 percent.

Same birds. Different boundary.

Now answer honestly about your own workplace: does succeeding on your measure ever require taking something from the person next to you? Time on the good machine, the easy jobs, the experienced hand, the parts, the help. If the answer is yes even occasionally, your scheme is measured at the wrong boundary, and the evidence is behavioural rather than financial: you will see it in who helps whom in the last week of a period.

Exercise 1.3 — Find what is already working (one week, out loud)

Ask three colleagues: what is the best thing anyone on this floor does that nobody outside this floor knows about? Write down the answers.

You are doing two things at once. You are locating the positive core, and you are measuring transmission fidelity — because a practice that three people can name and nobody outside can is a practice with h² near zero, which means it will die with the shift that holds it.

---

PART TWO — THE ARITHMETIC

Days 31–45: compute what the scheme is actually buying

Exercise 2.1 — Your own selection term (one day)

Take the teams, shifts or cells in your scheme as the population. For each, over the last four periods, record:

• z — a practice trait you can observe: handovers documented, checks completed before the run, time spent training somebody else, jobs finished by the person who started them.
• w — weight carried forward: hours allocated, headcount, share of the work.

Then compute, exactly as the chapter does:

  z̄ = mean of z        w̄ = mean of w        E(w·z) = mean of w·z
  Cov(w,z) = E(w·z) − w̄·z̄
  selection term = Cov(w,z) / w̄     in trait-points per period

The chapter's worked case gives Cov(w,z) = 0.720, w̄ = 1.100, a term of 0.6545 points a generation, of which selection is 88.5 percent and transmission 11.5 percent.

If your term is negative, your scheme is selecting against a practice you believe in, and you now have the number to say so.

Exercise 2.2 — The group term you are owed (half a day)

Where a practice costs you and helps the team — covering someone, training someone, writing the handover properly — the chapter gives the exact condition for it to survive:

   b / c  >  1 + V_within / V_between       →  1 + 2.6667/9.0000 = 1.2963

With b/c = 2.0000 the trait rises at +0.1652 points a period: between-team selection at +0.2348 beating within-team selection at −0.0696, 3.37 times harder. Drop the group benefit to b = 0.020 and the same six teams give −0.1789 — the practice is selected out.

This is the arithmetic behind the ask. You are not asking for fairness. You are pointing out that without a group term above the threshold, the scheme will destroy the cooperation it depends on, and the number is computable in advance.

Exercise 2.3 — What your practice is worth (2 hours)

Using R = h²·S: with S = 4.00 points and h² = 0.35, the response is 1.40 points a generation. Over 3.33 generations in a ten-year horizon that delivers 4.67 points against a 10.00-point requirement — 46.7 percent. At a value of £310,000 per point per year, one generation's response is worth £434,000 a year against a cohort cost of £1,020,000: a 42.5 percent return, 2.35-year payback, £728,571 per point.

Note what raises R fastest and costs least: h², from 0.35 to 0.750, doubles the response. h² is you. It is documentation, secondment, training the next person, writing the method down. That is the part of the value chain nobody prices, and this arithmetic prices it.

Exercise 2.4 — The baseline ratchet, tested (1 hour)

If the baseline resets to the improved level each period, every gain raises the bar you are measured against, the same effort yields less each cycle and eventually nothing. Compute your own: take the last four baselines and plot them. A baseline rising as fast as performance is a treadmill with a verification procedure attached.

A well-made scheme holds the baseline fixed for a stated term — three to five years — or ratchets on a published schedule everyone can see coming.

---

PART THREE — DESIGN

Days 46–70: make the uncounted countable

Exercise 3.1 — The three practices worth measuring (one week)

From Exercise 1.3, choose three practices that are real, valuable, and currently invisible to the measure. For each, write: what it is, what it prevents, and a number that could be collected in under five minutes a shift by the people doing it.

The five-minute rule is not a convenience. A measure that costs more than that will be gamed or abandoned, and either way it stops being a measure.

Exercise 3.2 — Propose the group term (2 hours)

One page. It says: the practice, the boundary it should be measured at, the number, who collects it, and the share. Then the arithmetic from Exercise 2.2 — b/c, the threshold, and the selection term with and without the group term.

A proposal with a computed threshold in it is a different document from a request. It can be checked, argued with and adopted. A request can only be granted or refused.

Exercise 3.3 — Raise h² deliberately (one week)

Pick one thing you know how to do that nobody else on the floor can do, and transfer it. Properly: one written page and one person trained to competence, within the period. Then record it.

This is the most undervalued move available to you. From R = h²·S, doubling transmission fidelity doubles the response to selection at zero selection cost — and unlike the other two levers it requires nobody's approval.

---

PART FOUR — DESTINY AND DELIGHT

Days 71–90: make it hold

Exercise 4.1 — Ask for the amendment window (1 hour)

The one clause worth more than a point of share: how much notice must be given before the measure changes. Without it, no practice with a payback longer than a period is rational to invest in — including every practice in Exercise 3.1. Ask for one generation of notice, in writing, in the scheme document.

Exercise 4.2 — Protect the variance (2 hours)

When every team is made to work the same way, V_between falls and the scheme loses the ability to tell a better method from a worse one — from the threshold condition, the required b/c rises without bound as V_between approaches zero. Standardisation past that point does not make the floor better. It makes improvement unmeasurable.

Argue for a named tolerance: teams may differ in method, provided the outcome measure and the safety standard are met. That sentence, in the scheme document, protects every future gain in it.

Exercise 4.3 — Seal a prediction (1 hour)

Before the period, write down what you expect the measure to do and by how much, with an interval — the chapter's form is 0.6545 ± 0.25, dying outside [0.4045, 0.9045]. Give a copy to your representative.

Being right about a number you wrote down in advance changes how you are heard in a way that no amount of being right afterwards ever does.

---

WHAT TO CLAIM, AND HOW THE LEDGER READS IT

A gainshare pays on verified improvement. Three kinds of contribution are easy to verify and routinely go unclaimed, and each one maps onto a term in this chapter's arithmetic.

Variance you created. You tried a method that was not the standard method. Whether it worked or not, it widened V_between and gave the scheme something to select on — and from the threshold condition, a population with no between-team variance cannot improve at all. Claim it as a variant: what you tried, over what period, what it cost, and what the outcome measure did. A scheme that pays only for successes is paying for luck; a scheme that funds variants at a published kept-to-funded ratio — three of twelve, an intensity of 0.250 — is paying for the mechanism.

Selection you sharpened. You made the measurement better: a count that was not being kept, a cause that was being misattributed, a baseline that was wrong in your favour and you said so. This raises the accuracy of Cov(w,z) and therefore of every gain computed afterwards, including other people's. It is the least visible contribution on any floor and the most durable.

Transmission you built. You wrote the method down while it was working. You trained the next person to competence. You took your practice to another site and it survived the move. This is h², and every figure downstream of it is linear: at h² = 0.35 the response is 1.40 points a generation, and at 0.750 it is 3.00.

How to record it so the ledger can read it. One line per item, in the period it happened: the date, the kind (variant, measurement, transmission), one sentence of what it was, and the number it moved. Nothing longer. A ledger entry that takes three minutes to write gets written; a case study does not, and an unwritten contribution is one that did not happen as far as any verifier is concerned.

And the honest part. Not every period will contain one. A period in which you ran the standard method well, and it worked, is a good period and the scheme should pay for it through the ordinary measure. The three kinds above are what you claim when the ordinary measure cannot see what you did — which is exactly the case this chapter exists to make visible.

---

KNOW YOUR SCHEME — A CHECKLIST

Mark each one. Any no is a conversation, not a complaint.

---

THE CONVERSATION, SCRIPTED

Opening. "I have computed what our scheme is selecting for. Can I show you one page?"

The evidence. The selection term for the practice, over four periods, with units. The b/c ratio against its threshold. The h² estimate and its method.

The ask, in three parts. A group term above the threshold. An amendment window of one generation. A tolerance for method variation between teams.

The close. "None of this changes the share. It changes what the share selects for, and the arithmetic is on the page."

---

APPRECIATIVE QUESTIONS FOR YOUR TEAM

1. What is the best thing anyone on this floor does, and who else could do it by the end of the quarter?
2. When has something we invented here travelled to another team or site? How did it travel, and who carried it?
3. What would we be willing to predict about our own measure before the period, and be scored against afterwards?
4. Where does succeeding on the measure cost the team next to us — and what would the group term have to be worth for that to stop being true?