The Record

A year after a contested decision was won, the executive who won it gives a clean account of how. They knew who decided, they knew what would move that person, they made the case and it went their way. Asked what they expected the decision-maker to object to, or how sure they were of the sign-off they needed, they are vaguer, and the answers tend to agree with what happened. Nobody is lying: memory keeps outcomes and rewrites the reasoning behind them to fit, so a year on, the executive remembers having foreseen what they only lived through.

The organization's documents do little better. What survives a decision is mostly the paper that records it: the minutes, the approval email, the revised budget. These say what was decided. None of them says what the person who brought it about expected along the way, or what they spent getting there. And most are kept as current versions, edited as things change, so the plan of March has been written over by June.

In The Lam System for Political Interaction, a maneuver leaves two things behind. One is the result: the target selects, and the actor obtains a resource or does not. The other is the record, every entry written from the moment the cycle opened, in the order it was written. One discipline governs the record: nothing in it is ever overwritten. Supersession is additive. When one entry takes the place of another, the replacement is written as a new entry and the original stays where it was.

What the system keeps track of

While a maneuver runs, LSPI keeps track of four things. Three of them describe where the maneuver stands: which cycle has control at this moment, which cycles are waiting beneath it for a sub-cycle to hand back, and what the actor holds, counted. These change constantly. Control passes down to a sub-cycle when a gate fails and back up when the sub-cycle closes. What the actor holds shrinks when a lever is executed and grows when resources flow back.

The fourth is the record, and the record only grows. Every change in the other three has a cause, and the cause is an entry: the gate that failed, the execution that spent a resource, the flow back that returned one. So the other three can always be accounted for from the record. If what Lena Ortiz holds at the end of her maneuver looks wrong, the record shows what she started with, every resource she spent and on which lever, and every resource that came back and from which cycle.

Most entries in the record are declarations, one for each constituent as it fires: the interest, the detection, the objective, each lever and its required resources, each execution, the target's perception and selection, the outcome, the target's obtained utility and the actor's obtained resources. Other entries mark moments that no declaration would show on its own, when the cycle itself moves: a cycle opening, a gate passing or failing, a cycle closing, resources being returned. From the declarations alone a reader would know what Lena said at each step, and would not know when she found she was short of Priya Natarajan's sign-off.

The record is also a sequence. When cycles nest, their entries interleave: while Lena works on Priya, the sub-cycle's entries are written in the middle of the cycle with Daniel Okafor. Each entry is written at the address of what it records, so every cycle's story can still be read out of the whole.

Nothing is overwritten

Most working documents in an organization are kept as the latest version. The plan, the forecast and the organization chart are edited in place; a version history may exist, but nobody reads it, and the document's authority comes from being current. The record's authority comes from the opposite fact: nothing in it has been improved after the event. An entry, once written, is never edited and never removed. A later entry that supersedes it is written after it, and the supersession is itself on the record. This is one of the guarantees the system makes: whatever the actor decides, no move rewrites an earlier entry.

Escalation shows the plainest case. When Lena's second lever fails its gate, the failure is written. When Priya's sign-off comes back and the gate is tested again and passes, the pass is written after it. The failure is never erased. A reader a year later can see that the lever stalled, what it lacked, the sub-cycle that went to get it, and the pass that followed. A latest-version document would show only the pass, and Lena would come to remember the sign-off as something she had all along.

The discipline has a cost. The record is longer than a summary, and it keeps entries that turned out wrong: a projection that missed, an aim that had to be revised, a gate that failed. Those are the entries an actor learns most from, and a record that dropped them would be an account of the maneuver as the actor would have liked it to go.

It also fixes what can be done with a cycle once it has ended. A closed cycle stays closed, and so does one that was given up. If Lena abandoned a maneuver and later wanted another attempt, that attempt would be a new cycle with a record of its own, and the first attempt's record would sit beside it unchanged.

The projection stays beside the realization

Much of a maneuver is written twice. The objective projects that a particular selection by the target will yield a particular position: Lena projects that Daniel will approve the move. Each lever projects how the target will see the comparison once it has been played. Later the realized versions arrive: Daniel's perception as it came out, his selection, the outcome. The outcome is the realized mirror of the objective, and the perception is the realized mirror of what the levers projected.

At the close, the record holds both. The realization is written beside the projection, and the projection stays. In the system's notation the two differ by a single mark.

Read aloud: the first is the preferred position as the target's selection was projected to yield it, written into the objective near the start of the cycle; the prime marks it as a projection. The second is the same position as the selection did yield it, written into the outcome at the end. Both sit at one address: under the only interest, under its target, in the target's selection, the only preferred position.

The system never treats the two as the same thing, even when they say the same thing. When Daniel approves exactly as Lena projected, the record holds two entries that agree, and the agreement is information: Lena called it. If a projection were allowed to stand in for its realization whenever the two matched, a correct forecast and a missing result would look identical on the record.

This is what makes forecast calibration possible. Calibration compares what was expected with what happened, and it needs both, kept apart. A record in which the realization overwrote the projection could say how a maneuver ended and could never say how well the actor saw it coming. The second is the more useful thing to know, because the actor will run another maneuver and needs to know how far their own expectations can be trusted.

Keeping both also tells two kinds of failure apart. Suppose Daniel had handed dispatch to the automation program. If the record showed both levers read as Lena projected, the desks ranked above the depots and then above automation, and Daniel choosing automation anyway, her levers worked and her picture of his choice was wrong: something weighed with him that her decision matrix never held. If it showed the signed figure read differently from her projection, with automation still ahead, the picture may have been sound and that lever missed. The first calls for better detection next time, the second for a better lever, and without the projection on the record the two failures look the same.

Cycle: Lena and Daniel

1. OpenThe cycleA new cycle opens under Lena's agenda.
2. InterestActorLena wants the dispatch budget and the authority over the desks.
3. DetectionActorDaniel will approve the move, leave dispatch with the depot managers, or hand it to the automation program.
4. ObjectiveActorProjected: Daniel approves the move, and the desks and their budget come to Lena.
5. IntentionActorFirst lever, aimed at the depot managers. Projected: shown that one desk cuts the overtime he has been criticized for, Daniel ranks the desks above the depots.
6. IntentionActorSecond lever, aimed at the automation program. Projected: with the savings figure signed by finance, Daniel ranks the desks above automation.
7. Required resourcesActorThe first lever needs the overtime data and her standing with Daniel.
8. Gate passesThe cycleFirst lever: Lena holds the data and the standing.
9. ExecutionActorThe overtime case goes to Daniel.Spent: the overtime dataSpent: some of her standing with Daniel
10. PerceptionTargetRealized: Daniel now ranks the desks above the depots, as projected. The first lever's projection stays beside it.
11. Required resourcesActorThe second lever needs Priya's sign-off on the savings figure.
12. Gate failsThe cycleSecond lever: Lena does not hold Priya's sign-off. This row stays after the gate later passes.
13. EscalateThe cycleThe cycle waits at the second lever's gate. Nothing Lena holds is spent.

    Sub-cycle: Lena and Priya · 14 entries

    1. OpenThe cycleA sub-cycle opens under the lever that failed.
    2. InterestActorLena wants exactly the sign-off the cycle above lacked.
    3. DetectionActorPriya will sign the figure now, or hold it until the year-end audit.
    4. ObjectiveActorProjected: Priya signs now.
    5. IntentionActorProjected: walked through how the savings figure was built, Priya finds it sound.
    6. Required resourcesActorHer standing with Priya.
    7. Gate passesThe cycleLena holds that standing.
    8. ExecutionActorLena walks Priya through how the figure was built.Spent: her standing with Priya
    9. PerceptionTargetRealized: Priya reads the figure as sound, as projected.
    10. SelectionTargetPriya signs.
    11. OutcomeThe cycleRealized: Priya signs now, as the objective above projected.
    12. Obtained utilityTargetPriya gets a savings figure she has checked herself before it reaches Daniel.
    13. CloseThe cycleThe sub-cycle closes.
    14. Flow backActorThe sign-off joins what Lena holds.Returned: Priya's sign-off
14. ResumeThe cycleControl returns to the cycle with Daniel, at the gate where it stopped.
15. Gate passesThe cycleSecond lever, tested again: Lena now holds the sign-off. The failure above is still there.
16. ExecutionActorThe signed savings figure goes to Daniel.Spent: Priya's sign-off
17. PerceptionTargetRealized: Daniel now ranks the desks above automation, as the second lever projected.
18. SelectionTargetDaniel approves the move.
19. OutcomeThe cycleRealized: Daniel approves, and the desks and their budget come to Lena. It mirrors the objective written near the top.
20. Obtained utilityTargetDaniel gains lower overtime and a cleaner organization.
21. CloseThe cycleThe cycle closes.
22. Flow backActorWhat Lena set out for joins what she holds.Returned: the dispatch budget and the authority over the desks

In the trace of Lena's maneuver, the objective near the top and the outcome near the bottom are the same position written twice, once as an expectation and once as a fact, and each lever's projection has its reading written below it as the lever lands. The failed gate is still there after the pass. No row summarizes an earlier row: each was written once, at the moment it records, and left alone.

A name and its definition

Every objective, lever and outcome in the record has two parts: an address, which names it, and a body, which says what it is. Lena's second lever has an address, the second lever under her objective, and a body: bring Daniel the savings figure with Priya's signature on it. Most of the time the record, and anyone discussing the maneuver, refers to the lever by its name. When the body is needed, the name stays on the record and the body is bound beside it, and the expansion is itself an event in the sequence.

A name and its definition look interchangeable, the way either side of an equation can stand in for the other. An equation runs both ways and has no history. The record runs one way, and which form was on the page at which moment is part of what it records. If a name could be swapped for its body and back without a trace, a reader could no longer tell whether a conversation was about the second lever as a commitment or about the specific case it stood for at that moment, and the record could not say when the commitment was first spelled out.

Keeping the two apart is also what lets a cycle's aim change without losing its identity. A revised interest is the same name with a second body. Because the name and the body are distinct, the record can hold both bodies under the one name and say which was in force at every point in the cycle. The same separation keeps a projection and its realization distinct when they share an address: the address names the position, and the mark says whether the entry is an expectation or a fact.

A revised aim keeps the old one beside the new

Actors change course in the middle of a cycle, often for good reason: they learn something that changes their reading of what the target can be brought to do. LSPI lets the cycle's aim be revised while the cycle is still open. The interest, what the cycle was opened to get, keeps its address and takes a new body. The old body stays on the record, marked as superseded, and the revision is an entry of its own, so the record holds the first aim, the second, and the moment between them. Detection runs again, since a different aim may mean the target is choosing between different things, and the objective follows, projecting the selection that would yield the revised aim. Levers already declared continue.

Illustrative. The people and the company are invented. Tomás Reyes runs regional sales at a building-products distributor and wants the chief executive, Hana Lindqvist, to move the national accounts to his team. His interest is every national account, and his objective is that Hana transfers all of them. His first lever, a comparison of his team's service record on the national accounts it already supports, is played. Then Tomás learns from the finance team that most of the national contracts are in renewal, and that Hana will not move an account during a renewal. He revises: the interest becomes every national account won from the start of the next fiscal year. Detection runs again, because Hana's choice is now a different one: give him the new accounts, leave all new accounts with the key-accounts team, or split them by region, and the objective becomes that Hana assigns him the new accounts. Tomás adds a second lever, his team's pipeline of prospects, and Hana assigns him the new accounts.

Cycle: Tomás and Hana

1. OpenThe cycleA new cycle opens under Tomás's agenda.
2. InterestActorTomás wants every national account for his team. Superseded later; never removed.
3. DetectionActorHana will transfer the national accounts to Tomás, leave them with the key-accounts team, or split them by region.
4. ObjectiveActorProjected: Hana transfers every national account to his team.
5. IntentionActorProjected: his team's service record on the national accounts it already supports makes the transfer look safe to Hana.
6. Required resourcesActorThe service figures.
7. Gate passesThe cycleTomás holds the figures.
8. ExecutionActorThe service comparison goes to Hana.Spent: the service figures
9. ReviseActorTomás learns that most national contracts are in renewal and that Hana will not move an account mid-renewal. He revises the cycle's interest at the same address. First body, kept: every national account. New body: every national account won from next year.
10. DetectionActorRun again for the new aim: Hana will give Tomás the new accounts, leave all new accounts with the key-accounts team, or split them by region.
11. ObjectiveActorProjected, for the revised interest: Hana assigns him every national account won from next year.
12. IntentionActorProjected: his team's pipeline shows it is closest to the prospects most likely to sign.
13. Required resourcesActorHis team's pipeline data.
14. Gate passesThe cycleTomás holds the pipeline data.
15. ExecutionActorThe pipeline goes to Hana.Spent: the pipeline data
16. PerceptionTargetRealized: reading the service record and the pipeline together, Hana sees new accounts under Tomás as the better option.
17. SelectionTargetHana assigns him the new national accounts.
18. OutcomeThe cycleRealized: it mirrors the objective declared after the revision. The first interest and its objective are still on the record, above the revision.
19. Obtained utilityTargetHana gets an owner for new national business without disturbing a renewal.
20. CloseThe cycleThe cycle closes.
21. Flow backActorWhat Tomás won joins what he holds.Returned: every national account won from next year

A year later, Tomás may well remember the maneuver as a bid for the new accounts from the start, while the record holds the larger aim he began with, the lever he played against it, the news that changed his reading, and the narrower aim he won. If his aims keep shrinking in the middle of his cycles, that is something about how he sets them, and only a record that kept the first versions can show it.

The decision matrix is read from the record

The decision matrix is the target's choice laid out: the preferred position, the adverse alternatives, and the benefits and costs the target weighs in each. It changes more than anything else in a maneuver, since every lever is aimed at it, and the system keeps no copy of it. Detection names its members. Each execution records the effect its lever had. The matrix at any moment is what those entries add up to, read from the record's history up to that moment.

So any earlier state of the matrix can be rebuilt by reading the record up to that point: what Daniel was choosing between before the overtime case went in, and how the choice looked after it. There is also only ever one account of the past. A matrix stored and updated in place would lose its earlier states as it changed. A matrix stored alongside a history would be two accounts of the past that could disagree, and someone would have to decide which to believe. With one record and the matrix read from it, there is nothing to reconcile.

The same holds for what the actor holds and for which cycles are waiting. Each is read from the record, and each can be rebuilt for any moment. Asked what Daniel's options looked like before Lena's second lever, a machine keeping the record replays it to that point, since there is no snapshot to consult.

Exact, and exact up to order

Order is part of what the record says, and for most of a maneuver the order is a fact. Detection comes after interest because it reads the interest. The sub-cycle with Priya opens because the gate failed, and the gate is tested again because the sign-off came back. Each of Lena's levers is read as it lands, after its execution, and Daniel selects only after both have been read. Each of these moves reads what the one before it wrote, so the order in which they appear could not have been otherwise, and the record states it exactly. The same goes for the actor's choices that cannot be undone, such as mooting a lever, abandoning a cycle or revising its aim: each changes what can follow, so where it falls in the sequence is part of what happened.

Some moves have no real order between them. Lena has two levers in one cycle, and once both have passed their gates, she can play the overtime case and Priya's figure in either order. Each spends its own resource. Neither reads anything the other writes. In her office she may present them in one meeting, interleaved. The record is a sequence, so it has to write one execution before the other, but the order it writes is not a fact about the maneuver. For moves like these the record is exact up to order: written in either order, it is the same record.

Insisting on one order for parallel moves would falsify the record. It would state that one lever came first where nothing in the maneuver made it so, and a later reader would be entitled to conclude that the second lever was a response to the first, or that the first was meant to prepare the ground for it, and neither would be true. A faithful record keeps exactly the order of everything that was sequenced and treats the parallel moves as interchangeable, which is how it says they were parallel. Whichever order they are written in, the maneuver ends the same way. Only the actor's choices at the points where the maneuver forks change how it ends, and that is where the boundary of computation falls.

Reveal allSelect each item to see the verdict.

• Lena declares her interest, then names Daniel as the one who decidesThe order is a factSelect to see

  Detection reads the interest. Nobody can ask who controls a thing until the thing has been said.

• The second lever fails its gate, then the sub-cycle with Priya opensThe order is a factSelect to see

  The failure is what opens the sub-cycle, and fixes what it wants.

• Priya's sign-off flows back, then the second lever's gate is tested againThe order is a factSelect to see

  The retest reads what came back. Written the other way around, the gate would have failed again.

• Lena plays the overtime case, then plays Priya's signed figureEither order, same recordSelect to see

  Two levers, each spending its own resource, neither reading what the other wrote. The record has to put one first; which one is not a fact about the maneuver.

• Lena declares the overtime lever, then declares the signed-figure leverEither order, same recordSelect to see

  Both levers are planned at the outset, and her budget of levers has room for both. Neither declaration reads the other, so either could have been written first.

• Both levers are read, then Daniel approvesThe order is a factSelect to see

  The selection waits until every live lever has been read or set aside. A selection written earlier would leave a lever's reading with nothing to explain.

• Daniel approves, then the outcome is recordedThe order is a factSelect to see

  The outcome is the realized result of the selection, so it cannot come before it.

• The cycle closes, then the dispatch budget flows back to LenaThe order is a factSelect to see

  What flows back is what the closed cycle obtained. Until it closes, there is nothing to return.

The order is a fact

6 of 8

Either order, same record

2 of 8

In Lena's maneuver, two of the eight pairs in the figure could be written in either order, and both are pairs of moves on her two levers. The other six are sequenced. Parallel moves are the narrow case: separate levers in one cycle, each working on its own resources.

The record as capital

A result is used once. Daniel approves, Lena obtains the desks and the budget that runs them, and those resources go into what she holds for the next maneuver, to be spent there. Her standing with Daniel, spent on this ask, is not available again for the same ask. Everything the record counts is consumed by use.

The record is not used up by being read. The entries about how Daniel read the overtime case are as available for the next maneuver with Daniel as they were the day they were written, and for the one after that. Over many maneuvers they add up to a history no single maneuver contains: what the actor expected and what happened, target by target and lever by lever.

Patterns in that history are visible only because the record kept both halves. An actor whose projections run ahead of what happens whenever a lever spends their standing with the target is overrating that standing, and will go on doing so until something shows them. A run of failed gates for want of the same kind of resource, a finance sign-off, say, points to a relationship worth building before it is needed, and the essay on detection describes the actor whose lists keep leaving off the option of doing nothing. No single record contains any of these patterns; they appear across many records, and only where every entry was left as written.

Every lever of this kind read as projectedAt least one read otherwise2Levers read otherwise than projected

Levers that spent her standingRead as projected in 5 of 12 months · 8 levers read otherwise than projected

1. Jan0
2. Feb1
3. Mar1
4. Apr0
5. May1
6. Jun0
7. Jul2
8. Aug1
9. Sep0
10. Oct1
11. Nov0
12. Dec1

Levers that spent something elseRead as projected in 10 of 12 months · 2 levers read otherwise than projected

1. Jan0
2. Feb0
3. Mar0
4. Apr1
5. May0
6. Jun0
7. Jul0
8. Aug0
9. Sep0
10. Oct0
11. Nov1
12. Dec0

The figure shows one invented actor's year, a maneuver closing each month. Levers that spent her standing with the target were read as she projected in five months of the twelve, with eight levers in all read otherwise. Levers that spent something else, such as data, a sign-off or a budget, were read as projected in ten months, with two read otherwise. A single miss in any one month reads as bad luck. Across the year, eight of the ten misses fall in the row for standing.

The record can also be read by someone other than the actor. An adviser the actor trusts with it reads the same entries the actor wrote, at the same addresses, so when the two disagree about what happened in a maneuver, the disagreement is about particular entries, and entries can be checked. An adviser is often better placed to see a pattern the actor is inside, and the record gives them something exact to see it in.

That is why the record, more than any result in it, is the practitioner's durable capital. Resources obtained in one maneuver are spent in the next. The record keeps growing, and what it holds is an exact account of the actor's own judgment beside what followed from it, which nobody else can supply.

Everything in that account can be kept by a machine without error: what was written, at what address, in what order, and beside what. None of it requires knowing whether Daniel was persuaded. The record holds that Lena judged he would be, when she judged it, and what he did next.