YOUR NEXT DISCOVERY

Reliable control: trace, test and explain automation

Follow a program from its inputs to a confirmed result, and find where its guarantees stop.

Complete the Course to Earn a Certificate and Join the Hall of Fame

A sensor connects to a controller and actuator, with feedback and a separate failed path.

See it. Try it. Make it yours.

  1. Trace the logic

  2. Test and diagnose a program

  3. Connect a complete algorithm

For beginners. · What to know before you start

Bring with you: Add, subtract, multiply and divide ordinary numbers; compare numbers and read labelled values.

I want to reason about simple automated programs and tell the difference between a requested action and a valid, confirmed result.

Follow a program from its inputs to a confirmed result, and find where its guarantees stop.

By the end, you can…

  • Follow a calculation through functions with clear inputs and returned results.
  • Use a failing example and a trace to choose a targeted program repair.
  • Check whether a list-processing program meets its complete stated requirement.
  • Decide whether a requested action satisfies every stated permission.
  • Tell whether a reading is fresh enough for a decision without confusing a screen refresh with a new observation.
  • Explain whether a simulated action was permitted and actually confirmed, including what remains unknown.
  • Diagnose order-dependent test evidence caused by shared mutable setup and choose an isolated fixture.
  • Use a stated left-to-right short-circuit rule to protect a partial operation and distinguish truth from safe evaluation order.
  • Trace a bounded recursive function using a base case, a decreasing input and pending returned work.
  • Apply an exclusive mode selector and a defined mode-change policy without combining requests from inactive sources.
  • Trace a consecutive-sample input filter that resets on contradiction and distinguish stable confirmation from total positive samples.
  • Detect a circular resource wait and evaluate a shared acquisition order without claiming that deadlock freedom implies fairness.

For another Course

  • Deploying physical machinery or providing equipment safety certification.
  • A particular programming language, continuous-control mathematics or distributed-systems completeness proofs.

YOUR LEARNING PATH

One idea leads to the next.

0 of 40 Missions completed or proven
Unit 1

Trace the logic

Follow stored values, conditions, branches, repeated work and function results.

  1. Follow a changing valueMission 1 of 40Locked. Trace sequential assignment and distinguish a stored value from an unevaluated formula.
  2. Evaluate a precise conditionMission 2 of 40Locked. Evaluate comparisons and combined Boolean conditions under explicit truth rules.
  3. Choose exactly one branchMission 3 of 40Locked. Trace an if/else-if/else chain, including overlapping conditions and branch order.
  4. Accumulate repeated workMission 4 of 40Locked. Trace a bounded loop and distinguish iteration count from the value accumulated.
  5. Follow inputs and returned valuesMission 5 of 40Locked. Evaluate parameterised functions and trace a returned result through another operation.
Unit 2

Test and diagnose a program

Use explicit list identity, isolated test data, traces and assertions to evaluate and diagnose program behavior.

  1. Read and update indexed dataMission 6 of 40Locked. Use zero-based list indices and preserve unmodified entries during an update.
  2. Know when two names share one listMission 7 of 40Locked. Distinguish two references to one mutable list from an independent shallow copy of a flat list.
  3. Keep an execution traceMission 8 of 40Locked. Construct and interpret intermediate-state traces to locate when an output was produced.
  4. Choose tests that distinguish behaviourMission 9 of 40Locked. Select normal, boundary and contrasting inputs that discriminate competing program behaviours.
  5. Check a claim while a program runsMission 10 of 40Locked. Evaluate explicit assertions and distinguish a failed property from a complete fault diagnosis.
  6. Make a test independent of yesterday’s stateMission 11 of 40Locked. Diagnose order-dependent test evidence caused by shared mutable setup and choose an isolated fixture.
  7. Repair the first wrong decisionMission 12 of 40Locked. Use a requirement and counterexample to choose a targeted logic repair.
Unit 3

Connect a complete algorithm

Respect types, evaluation order, termination, recursive returns, interfaces and event history when evaluating algorithms.

  1. Keep numbers and text distinctMission 13 of 40Locked. Trace explicit numeric conversion and diagnose operations applied to the wrong data type.
  2. Check before evaluating a risky expressionMission 14 of 40Locked. Use a stated left-to-right short-circuit rule to protect a partial operation and distinguish truth from safe evaluation order.
  3. Explain why a loop stopsMission 15 of 40Locked. Trace a while-loop guard and identify whether its update makes progress toward termination.
  4. Finish the work waiting inside a callMission 16 of 40Locked. Trace a bounded recursive function using a base case, a decreasing input and pending returned work.
  5. Connect small calculationsMission 17 of 40Locked. Trace data through composed functions while respecting input and output contracts.
  6. Remember state between eventsMission 18 of 40Locked. Trace a finite-state transition table and distinguish a repeated event from a new state transition.
  7. Check a complete bounded algorithmMission 19 of 40Locked. Evaluate a list-processing algorithm against its stated filtering, counting and output requirements.
Unit 4

Control a stated simulation

Separate observations from commands and combine sampled state, repeated requests, permissions and exclusive operating modes.

  1. Separate observations from commandsMission 20 of 40Locked. Distinguish sensed input, commanded output and independently observed response in an automated system.
  2. Follow the read–decide–act cycleMission 21 of 40Locked. Trace a sampled controller cycle using a fixed input snapshot and identify when later changes can be observed.
  3. Make repeated commands predictableMission 22 of 40Locked. Distinguish idempotent set-state commands from repeated incremental actions.
  4. Avoid switching at every tiny fluctuationMission 23 of 40Locked. Trace a two-threshold controller whose intermediate band preserves the previous output state.
  5. Require every permission before actingMission 24 of 40Locked. Evaluate a stated permissive interlock and keep missing or invalid inputs separate from permission.
  6. Choose which request source is in chargeMission 25 of 40Locked. Apply an exclusive mode selector and a defined mode-change policy without combining requests from inactive sources.
Unit 5

Check the operational evidence

Use stable-input rules, time, logs, transition coverage and controlled tests to distinguish confirmed behavior from missing information.

  1. Respect minimum intervalsMission 26 of 40Locked. Calculate elapsed time and evaluate a bounded cooldown without confusing elapsed time with clock values.
  2. Wait for a genuinely stable inputMission 27 of 40Locked. Trace a consecutive-sample input filter that resets on contradiction and distinguish stable confirmation from total positive samples.
  3. Explain a decision from its recordMission 28 of 40Locked. Interpret a controller log that binds the input snapshot, decision, command and observed result to one operation.
  4. Test the whole control pathMission 29 of 40Locked. Choose end-to-end simulated tests that exercise decision boundaries and verify observed output rather than commands alone.
  5. Test the paths between statesMission 30 of 40Locked. Compare a recorded test path with reachable controller transitions and identify an untested edge despite visiting every state.
  6. Learn from a controlled faultMission 31 of 40Locked. Interpret one-at-a-time simulated fault experiments and distinguish observed resilience from untested conditions.
  7. Notice when an input is too oldMission 32 of 40Locked. Evaluate an explicit freshness contract and distinguish stale input from a valid unchanged reading.
Unit 6

Coordinate and audit actions

Respect output limits, late events, exclusive resources, fair turns, dependencies and restart records when defending controller behavior.

  1. Recognise an action limitMission 33 of 40Locked. Evaluate controller requests against explicit output limits and distinguish a clipped command from achieved regulation.
  2. Handle events in a defined orderMission 34 of 40Locked. Use explicit operation versions to reject obsolete results without overwriting a newer accepted state.
  3. Prevent two tasks claiming the same resourceMission 35 of 40Locked. Find an unsafe check-then-claim interleaving and replace it with a single conditional resource claim.
  4. Coordinate dependent actionsMission 36 of 40Locked. Evaluate a bounded action sequence with explicit preconditions and completion checks.
  5. Resume without repeating completed workMission 37 of 40Locked. Choose restart behaviour from durable operation records and distinguish uncertain from confirmed effects.
  6. Give every ready task a turnMission 38 of 40Locked. Evaluate bounded controller scheduling for starvation and apply a stated cyclic service rule under explicit duration assumptions.
  7. Avoid a circular waitMission 39 of 40Locked. Detect a circular resource wait and evaluate a shared acquisition order without claiming that deadlock freedom implies fairness.
  8. Defend a controller’s resultMission 40 of 40Locked. Evaluate a complete simulated controller trace against freshness, permission, sequence and confirmation requirements.Mastery Challenge

A Debsie Course · beginner