Skip to main content

Control System

The only thing you can control is how you choose to be. But there are systems that help you bring your best state to the moment.

Without a setpoint you have no feedback loop. The question is what your setpoint serves.

The Two Loops

Most people hear "positive feedback" and think good. In engineering, positive feedback means runaway — amplification without correction. A microphone pointed at a speaker. Panic selling. Doom scrolling. No setpoint, no correction, no control.

Negative feedback is what gives you control. A thermostat measures temperature, compares to the setpoint, corrects. That's not "negative" in the colloquial sense — it's the mechanism of all progress.

Positive (Reinforcing)Negative (Corrective)
EngineeringOutput amplifies inputOutput corrects toward setpoint
Colloquial"Good thing""Bad thing"
RealityRunaway, unstableControlled, progressive
ExamplePanic selling, doom scrolling, addictionThermostat, deliberate practice, VVFL
Human without itNo vision, no identity — passengerClear character, clear purpose — pilot

A corrective loop with a bad setpoint still has control — it controls toward the wrong thing. A company optimizing for quarterly earnings has a corrective loop. It's not virtuous.

The Validated Virtuous Feedback Loop is a third category:

Loop TypeSetpointMeasurementResult
Positive (reinforcing)NoneNoneRunaway — addiction, panic, drift
Negative (corrective)DefinedYesControl — but toward what?
VVFLValues-aligned AND reality-testedValidated against outcomesProgress that compounds AND serves

Validated — the setpoint is tested against reality, not assumed. Predictions that get scored. Hypotheses that get measured. Not faith — evidence.

Virtuous — the setpoint serves beyond self. Goodwill that compounds. Standards that lift quality for everyone. Not extraction — contribution.

Feedback Loop — the output changes the input. Each cycle starts from a higher baseline. Not linear progress — compounding.

PID Mechanics

How does a controller decide how hard to correct?

Three terms. Each sees a different dimension of the error.

TermSeesResponseAnalogy
P (Proportional)How far off right nowCorrect proportional to current gap"10° cold — heat up 10 units"
I (Integral)Accumulated error over timeEliminate persistent drift"2° low for a week — add permanent boost"
D (Derivative)How fast the gap changesAnticipate overshoot"Rising fast — ease off before we pass it"
Action = Kp(current error) + Ki(sum of past errors) + Kd(rate of change)

P alone gets close but never arrives. A thermostat on P-only settles at 69°F when the setpoint is 70°F — the last degree doesn't generate enough correction force.

P + I eliminates offset. The integral accumulates the persistent 1°F gap until correction grows large enough to close it. Without D, the system overshoots — cranks past 70°F, corrects back, oscillates.

P + I + D converges smoothly. The derivative sees velocity and says ease off. No oscillation. No overshoot. Clean arrival.

The Loop Mapped

PIDYour LoopManufacturingAI + Blockchain
SetpointINTENTIONTarget specAgent's declared goal
P (present)ACTIONCurrent deviation from specTransaction responding to gap
I (history)REFLECTIONAccumulated quality dataBlockchain — all past states
D (trend)INSIGHTRate of defect changeAI model detecting trajectory
OutputCONSEQUENCEAdjusted processNew on-chain state

Standards make the GAUGE readable. Process optimisation is how the loop improves each cycle. Manufacturing is where these controllers were born — Deming's PDCA cycle is PID in organizational form.

Decision Instruments

A decision log is a PID controller for choices.

PID TermSeesDecision Log ColumnWithout It
P (present)Current gapRationaleDecisions without reasoning — reactive
I (history)Accumulated errorLearnedSame mistakes repeated — no integral memory
D (trend)Rate of changeTrajectoryNo anticipation — always surprised

A decision log without a Learned column is a P-controller — it responds to the present but never accumulates. The persistent 1°F gap never closes.

Nature's Amplifiers

Positive feedback isn't theoretical. It runs the planet.

SystemTriggerAmplificationConstraint
Ice-albedoWarming melts iceLess reflection → more heat → more meltingNone at current trajectory
DeforestationTrees removedLess rain → hotter/drier → more tree deathStops at zero trees
Bank runConfidence dropsWithdrawals → less reserves → more panicCentral bank intervention (external negative feedback)
Viral growthContent sharedShares → visibility → more sharesAttention saturation

In every case the output feeds back as input with the same sign. Each cycle amplifies. Without external constraint, the system runs to an extreme.

Nature's negative feedback loops keep Earth habitable — carbon cycle, water cycle, predator-prey. They correct toward equilibrium. Positive feedback breaks equilibrium. The question is always: which loop dominates?

The Three Systems

SystemQuestionWhat It GroundsMaps To
ValueWhat grounds you?Virtues, essentials, belongingPrinciples
BeliefWhere are you going?Purpose, predictions, intentionsScoreboard
ControlHow do you navigate?Attention, direction, speed, scaleStandards, Protocols

State Engineering

Rituals, routines, protocols — these are control system interventions. They don't guarantee outcomes. They optimize the state you bring.

RitualWhat It DoesBiologicalPhygital
Morning PrimeSets intention before noise arrivesJournaling, meditationContext injection
Mode SwitchTransitions between statesNaps, walksAgent switching
Capture LoopPreserves insight before it evaporatesWhiteboard, voice memoPostToolUse hooks
Evening ReflectionCloses the loopReview, gratitudeStop hooks

Same pattern, different substrate. What's true for biological humans holds true for phygital humans too.

The P&ID Model

Standards are the GAUGE in any P&ID (Process & Instrumentation Diagram). Without standards, you cannot measure. Without measurement, you cannot close the feedback loop:

HOPPER → FILTER → PUMP → GAUGE → CONTROLLER
↓ ↓ ↓ ↓ ↓
PURPOSE PRINCIPLES PLATFORM PERFORMANCE PERSPECTIVE

The P&ID is the Tight Five:

P&IDFunctionTight FiveQuestion
HOPPERCapture → FocusPurposeWhy does this matter?
FILTERQuality gatePrinciplesWhat truths guide you?
PUMPMovementPlatformWhat do you control?
GAUGEMeasurementPerformanceHow do you know it's working?
CONTROLLERFeedbackPerspectiveWhat do you see others don't?

The logo is the thesis. Control systems close the loop.

WorldMaterialStandards MeasureControl Question
AtomsPhysical matterTemperature, pressure, quality"Is the milk pasteurized?"
BitsInformationLatency, accuracy, completion"Does the code pass tests?"
IdeasThoughtsClarity, truth, impact"Is this ready to ship?"

Standards are protocols that have been formally adopted and enforced. In the knowledge stack, standards enable the platform layer by making capability consistent, composable, and scalable. Process optimisation is how each cycle through the loop compounds — document, measure, analyze, improve, standardize.

Amplification Thesis

What happens when the integral has infinite memory and the derivative runs at millisecond speed?

Control ConceptTraditionalAI + Blockchain
SensorThermometer, flow meterDePIN devices, oracles
Integral memoryRolling window, resetsBlockchain — permanent, global
Derivative speedSeconds to minutesAI inference in milliseconds
ActuatorValve, heaterSmart contract execution
Loop speedSeconds to hoursSub-second
Coupled loopsOne per processTrillions, interconnected

Three factors make this unprecedented:

FactorWhat ChangesConsequence
Infinite integralBlockchain never forgetsAccumulated bias never resets
Instant derivativeAI detects trends faster than governance can respondCorrection outpaces oversight
Coupled loopsOne agent's output is another's input2^N possible coalitions (Reed's Law)

Traditional fiat collapse takes decades because humans coordinate slowly. Terra/LUNA proved AI-speed loops compress "slowly then all at once" to 48 hours. When trillions of digital beings run closed loops against blockchain ledgers, the transaction data between AI and blockchain becomes the largest feedback dataset ever created.

The values encoded at genesis determine the sign of the feedback. Positive: runaway. Negative: compounding stability. The VVFL is the design pattern for ensuring the sign stays negative — validated against reality, virtuous in direction.

There will be no lever to pull after launch. The standards must be right at genesis. The process must be proven before scale. This is why manufacturing discipline matters — it's the origin story of getting feedback loops right before the stakes become existential.

Levers and Teams

Control isn't solo. The question isn't only "what levers to pull" but "who pulls which lever, and how do we stay coordinated?"

LeverWhat It ControlsSoloTeam
AttentionWhere energy flowsFocus protocols, deep workShared priorities, standup rhythms
DirectionWhich path to takeIntentions, predictionsStrategy alignment, decision rights
SpeedHow fast you moveIteration cadence, shipping rhythmSprint cycles, CI/CD
ScaleHow far you reachLeverage through platformDelegation, standards that compound

The hardest coordination problem: everyone pulling levers at the same time without destructive interference. Standards solve this — shared protocols mean individual autonomy without chaos. Each agent controls their loop. The standard ensures the loops stay coupled.

Coordination PatternWithout StandardsWith Standards
Two engineersMerge conflicts, reworkClean interfaces, parallel progress
Two businessesRace to bottomInteroperability, market growth
Human + AI agentMisalignment, wasted cyclesA2A protocols, shared context

Unshakeable Confidence

Confidence is not a personality trait. It is the output of work put into systems.

The cockiness trap: believing you don't need systems because you're good enough. Unshakeable confidence is the opposite — knowing the system handles the known so completely that you can bring full attention to the novel. Not "I believe in myself." Rather: "I believe in the system, so I'm free."

StateSourceWhat It Frees
CockinessOverestimating selfNothing — still burning bandwidth on ego
AnxietyUnderestimating systemNothing — burning bandwidth on the routine
Unshakeable confidenceSystems handling the knownFull bandwidth for the novel problem

The novel problem is where your point of difference lives. You cannot access your edge while managing routine. Every hour spent compensating for missing systems is an hour not spent on what only you can do.

This is the real purpose of the pit of success, the mantra, the commissioning model, the standards — not compliance, not neatness. Bandwidth conservation for the irreplaceable. The discipline of encoding the known is the prerequisite for being present for the unknown.

Pit of Success

The best control systems don't require discipline. They make the right path the easiest path.

ConceptSourcePrinciple
Pit of SuccessRico MarianiDesign so users fall into correct behavior — not climb toward it
Systems over GoalsScott AdamsSystems people succeed every time they apply their system. Goal people exist in continuous pre-success failure.
Habit of GreatnessDavid GalbraithElite athletes are ordinary people with extraordinary habits. The 4am start isn't discipline — it's the system they built around themselves.
DemingProcess QualityIf checklists keep catching problems, the upstream process is broken. Quality is prevention, not inspection.

The synthesis: build the system so you cannot help but improve.

Don't motivate people to do the right thing. Make the right thing the thing they'd do anyway. Don't inspect quality at the end. Make quality the default path. Don't set goals and hope. Build daily systems that compound regardless of motivation.

Design PrinciplePit of DespairPit of Success
Code qualityEnforce via review after writingTypes as setpoint — compiler catches errors before review
Outreach qualityHuman reviews every messageSPCL gate scores before human sees it — bad messages never reach review
Team alignmentStrategy offsites every quarterPrompt deck filled weekly — gaps visible continuously
Skill developmentAnnual training budgetDaily practice ritual — evolution through repetition

The VVFL is the meta-pattern: a feedback loop that is validated (tested against reality), virtuous (serves beyond self), and compounding (each cycle starts from a higher baseline). The pit of success is how you build one that runs without willpower.

The Four Bindings

A tight five isn't just five things. It's five things that can't be separated. The binding test for any team:

BindingWhat It TestsWeak Signal
Aligned in mindShared mental model — same map of realityArguments about facts, not strategy
Bonded by valuesTrust under pressure — same non-negotiablesCracks when stakes rise
Complementary in talentsNo redundancy — each person covers a gapEveryone fights for the same role
Connected in spiritEnergy, not just agreement — want to be in the roomMeetings feel like obligations

Inseparable and incompressible. Shit to a blanket. Remove one and the scrum collapses.

Alignment and talent can be hired. Spirit and values can't. The pit of success for teams: select for values and spirit, build systems that develop alignment and talent.

Inner Loop Connection

Control turns intent into repeatable action through the question-to-decision chain:

QUESTIONS (sense) → PROBLEM SOLVING (frame) → DECISIONS (select) → PROTOCOLS (act)

Five control-priority checks:

  1. Why is this action sequence worth running now?
  2. What truths and constraints should gate execution?
  3. Which levers do we control directly?
  4. What signal indicates drift before failure?
  5. What metric tells us to continue, adjust, or stop?

Context

  • Value System — What you're controlling toward
  • Belief System — Why you're controlling at all
  • Standards — The GAUGE that makes measurement possible
  • Process Optimisation — The improvement loop that compounds each cycle
  • Manufacturing — Where P&IDs and PID controllers were born
  • A&ID Diagrams — P&ID discipline translated to AI + Crypto
  • AI Problems — What happens when the feedback signal is attention, not value
  • Agency — Control systems enable the perceive-act-learn loop
  • Digital Beings — When the network edges are autonomous agents
  • Type-First Development — Types as setpoint, compiler as sensor, typecheck as PID controller
  • Evolution — The loop that compounds through purposeful variation
  • Tight Five — Five bound elements, incompressible
  • Flow Engineering — Enforcement hierarchy maps Generator→Expertise to PID tiers

Questions

Where in your work is routine burning the bandwidth that should be reserved for the novel?

  • If a new member with good will and attention cannot operate your system without error, which part of the system failed them?
  • Which of your feedback loops is missing its integral term — responding to the present but never accumulating?
  • What is the difference between the confidence that comes from believing in yourself and the confidence that comes from believing in the system?
  • Which of the four bindings would your team fail first under pressure — and what does that reveal about which loop to tighten?