Task sheets without answers
Original fictional cases for basic practice. Open judgements have defensible alternatives. Use the separate comparison pages after making your own attempt.
Systems concepts: make your boundary inspectable
Fictional case: Riverbank Library has a computer-booking service. Staff count a booking as successful when a slot is allocated. In one week 80 slots are allocated, 30 are not used, and 20 visitors leave without completing the online application they came to make. The booking form is online. A community group helps people who cannot use it; this work is absent from the library's statistics. Managers propose reducing staff assistance to allocate more slots. Staff predict that visitors will return repeatedly for help. Neither prediction has yet been tested.
Draw an OU-style systems map for the booking manager's system of interest.
Produce: A titled containment map with one stated purpose and boundary.
Construct a second map from a visitor's purpose.
Produce: A second boundary and a short explanation of what changed.
Write a testable feedback hypothesis about reduced assistance.
Produce: One closed causal sentence and one observation that could challenge it.
Locate yourself in the inquiry.
Produce: A two-sentence observer note.
Checks
Does 80 allocated slots establish 80 successful visitor outcomes?
- Yes: an output is an outcome
- No: allocation and completion are different observations
- No: all 80 bookings failed
What does a boundary in an OU systems map primarily show?
- A causal influence
- An editorial inclusion in a stated system of interest
- A manager's authority
Which observation supports emergence without claiming magic?
- Repeated local booking decisions can generate an access pattern nobody intended
- Every large organisation is unpredictable in every respect
- A whole has no relationship to its parts
Repair
The map places every organisation in one circle labelled 'the whole system' and claims that everyone shares the library's success measure.
Changed case
A new manager counts only applications completed on library computers. What exclusion could this introduce, and how would you test it?
Review criteria
- Each map has a purpose and an explicit boundary.
- Containment and causal links are not conflated.
- Observed facts, interpretations and predictions are distinguished.
- The observer's choices and an excluded perspective are named.
Systems laws: test variety, feedback and local optimisation
Fictional case: a help desk receives four mutually exclusive request states: A password reset, B identity discrepancy, C accessibility support and D suspected fraud. Its safe responses are R1 an automated reset for A, R2 a trained identity investigation for B, R3 supported access for C and R4 a fraud investigation for D. Applying any other response to a state leaves an unacceptable outcome in this deliberately simplified exercise. The new portal can recognise only 'ordinary' and 'exceptional', and offers only R1 or 'send an email'. Replies to that email take five working days. Managers propose counting cases closed by the portal as success.
Construct the disturbance-response table.
Produce: Four rows showing the necessary distinction, safe response and current gap.
Design one legitimate attenuation and one amplification.
Produce: Two changes, each explaining what variety is changed.
Test the five-day feedback delay.
Produce: A failure mechanism and an observable warning signal.
Replace the local success claim with a test of the whole service.
Produce: One outcome measure and one counter-measure.
Checks
Under this case's disjoint-response assumption, how many distinguishable effective responses are required?
- Two, because there are two portal labels
- Four
- An unlimited number
Which proposal changes the disturbance rather than merely hiding it?
- Exclude inaccessible requests from the dashboard
- Correct a recurring upstream identity-record defect
- Rename all exceptions 'ordinary'
Can a daily dashboard compensate for an adviser with no authority to act?
- Yes, information alone supplies control
- No, sensing and effective action are both needed
- Yes, provided it contains more indicators
Repair
'The law of requisite variety proves that every customer needs a unique process and standardisation is always wrong.'
Changed case
B and D can now be safely handled by the same authorised investigation team using one triage action. Does the four-response conclusion still follow?
Review criteria
- The essential outcome and unacceptable states are explicit.
- Response capability is not reduced to headcount or number of labels.
- Attenuating a disturbance is distinguished from suppressing its signal.
- A claim is revised when its modelling assumptions change.
Transposing theory: move up to a pattern and back down
Fictional case A: a warehouse orders replacement stock from last month's sales. Delivery takes four weeks. When shelves empty, several managers independently raise urgent orders. Stock later arrives together and exceeds storage capacity. Fictional case B: a council trains additional caseworkers using last quarter's backlog. Training takes six months. Several services independently commission the same small pool of trainers. Trainees arrive after demand has changed. Casework, unlike boxes, also varies in urgency, consent, relationship and complexity.
Write the common structural pattern without using warehouse or council nouns.
Produce: A four-part abstract model.
Map the abstract elements into case B.
Produce: A correspondence table with at least one non-correspondence.
Derive a modest intervention and a falsification test.
Produce: One intervention hypothesis with a measurable challenge.
Checks
Which is the transferable part of the example?
- A warehouse's staffing ratio
- The delayed signal, replenishment pipeline and duplicated decisions
- The claim that people are stock
What should happen when case mix invalidates interchangeability?
- Discard the evidence to protect the analogy
- Revise the abstraction or stop using it
- Assume the model applies because it is systemic
Repair
'Hospitals are factories, so the warehouse rule should be applied unchanged to patients.'
Changed case
Apply the same pattern to recruiting volunteer interpreters. Name one correspondence and one reason a numerical inventory rule might fail.
Review criteria
- The abstract pattern contains relationships, not just similar vocabulary.
- Correspondences and non-correspondences are both recorded.
- A practical prediction could be challenged by evidence.
- The learner returns to and revises the original situation.
Soft Systems Methodology: build a model that follows its purpose
Fictional case: a repair cooperative allocates home-repair visits. Its scheduler wants every technician hour booked. Residents want a usable home and predictable appointments. Technicians want the information and parts needed to finish safely. The funder pays for completed visits, even when another visit is needed. Residents with uncertain work shifts often miss appointments. The cooperative's board can change booking rules but cannot compel a resident to be at home. There is no evidence yet about which change would improve completed repairs.
Make a rich picture of the situation.
Produce: One picture or accessible narrative showing the competing concerns.
Write two relevant root definitions using PQR.
Produce: Two distinct purposeful perspectives, each one sentence.
Check the resident definition with CATWOE.
Produce: Six explicit entries, including the transformation.
Build the minimum activity model for the resident definition.
Produce: Five to nine verb-led activities with dependencies and monitoring.
Compare the model with the case and propose one accommodation.
Produce: A three-row comparison and a feasible next inquiry.
Checks
Which transformation matches the resident root definition?
- Unallocated slots become allocated slots
- Agreed repair needs become completed, checked repairs
- Residents become compliant
What is wrong with activity boxes labelled Board, Finance and Technician?
- They are roles or units, not logically required activities
- They are always outside the system
- Nothing: any organisation chart is an activity model
A model differs from current practice. What follows?
- Current practice must immediately copy the model
- The difference supplies a question for structured comparison
- The model is necessarily false
Repair
Root definition: restore usable homes. Model: maximise visits, close tickets, penalise missed appointments. No completion check.
Changed case
The board now insists on minimising disruption rather than maximising technician utilisation. Add one necessary activity and one comparison question.
Review criteria
- At least two purposes remain distinguishable.
- CATWOE supports rather than replaces a coherent root definition.
- Activities logically follow the selected purpose and include monitoring.
- Comparison produces inquiry and possible accommodation, not a claim that the model is reality.
System Dynamics: build, calculate and challenge a queue model
Fictional case: a service starts week 1 with a backlog of 40 cases. Exactly 12 new cases arrive at the start of each week. Staff can complete at most 10 cases per week in weeks 1 and 2. Additional trained capacity is available from the start of week 3, raising the limit to 16. All cases are identical in this toy model. Completions during a week cannot exceed that week's starting backlog plus arrivals. There are no withdrawals or reopens in the first model.
Draw the stock-flow structure and state its units.
Produce: One stock, two flows and an explicit time step.
Calculate the first four end-of-week backlogs.
Produce: A four-row calculation table.
Describe a balancing feedback that could generate the capacity decision.
Produce: A signed loop and its delay.
Run two extreme-condition checks.
Produce: Results for zero arrivals and for an almost empty queue.
Identify a missing mechanism that could reverse the policy conclusion.
Produce: One boundary extension and data needed to assess it.
Checks
What is the backlog at the end of week 2?
- 20 cases
- 44 cases
- 64 cases
Which unit belongs to the backlog?
- Cases/week
- Cases
- Weeks/case
Does a positive link mean a beneficial effect?
- Yes
- No: it means change in the same direction, other influences held constant
- Only for reinforcing loops
Starting backlog 2, zero arrivals, capacity 16: how many cases complete?
- 16
- 2
- -14
Repair
Backlog = arrivals / completions. It is labelled 'cases', and managers say its fall proves a balancing feedback loop.
Changed case
From week 3, let 25% of each week's 16 completions return as new backlog at the end of that same week. Calculate end backlogs for weeks 3 and 4.
Review criteria
- Stocks, rates, units, initial conditions and timing are explicit.
- The four baseline calculations and empty-queue test are correct.
- Loop polarity is causal rather than moral.
- The numerical model is distinguished from an unimplemented feedback hypothesis.
- A missing mechanism and a way to test it are named.
Viable System Model: diagnose functions and channels
Fictional case: Neighbourhood Advice Cooperative runs North, Central and South advice centres. Each serves local residents, organises its own advisers and decides ordinary cases. They share specialist interpreters and sometimes double-book them. A cooperative operations group negotiates budgets and capacity with centre leads. A quality reviewer periodically observes cases directly, with an agreed learning remit. A small planning group studies changing resident needs and technology, but its reports rarely reach operations. The elected board sets identity and purpose and settles major policy disputes. An urgent safeguarding concern currently travels through three monthly reporting meetings.
State the focal system and one level above and below it.
Produce: A recursion statement and three candidate operational units.
Map the case evidence to Systems 1-5 and 3*.
Produce: A function-evidence table, leaving unknowns visible.
Diagnose two channel failures.
Produce: Two hypotheses, each with a verification question.
Design a correction that preserves local autonomy.
Produce: One bounded change and a check for over-centralisation.
Test the diagnosis with operational evidence.
Produce: A short sampling plan.
Checks
Which is the strongest System 2 example in this case?
- A shared interpreter scheduling protocol
- The elected board's identity statement
- A future-demand study
What distinguishes System 3* from routine System 3 reports?
- It is a more senior manager
- It provides a direct, agreed check on operational reality
- It is every financial audit, regardless of purpose
Must the three centres have identical local arrangements to be System 1 units?
- Yes
- No, they need viable local responses within the cooperative's identity and constraints
- Only if the chart is symmetrical
Repair
The chief executive is labelled S5, all middle managers S2, every worker S1, and the outside world is omitted.
Changed case
An autonomous helpline is added with its own users, case decisions and local management. Is it necessarily System 2 because it connects centres?
Review criteria
- The recursion and system of interest are explicit.
- Operational units include operation, management and environment.
- Systems 2, 3, 3*, 4 and 5 are differentiated by function.
- The diagnosis tests channels and evidence, not just missing boxes.
- Corrections consider autonomy, legitimacy and timely response.
Critical Systems Heuristics: inspect all twelve boundary categories
Fictional case: a council proposes replacing its assisted benefits-application desk with a digital service. The finance director sponsors the proposal and counts annual administrative cost as success. A vendor controls the software design. The council controls staffing and opening hours. An analyst advises using completion rates from current online users. People unable to start online are absent from that dataset. A tenants' group has asked to be heard but has no place in the decision meetings. The project document says 'digital by default is modern and fair'. No replacement assistance or appeal route has been agreed.
Complete the four groups of boundary categories in the descriptive 'is' mode.
Produce: Twelve labelled entries, separating case evidence from unknowns.
Write an 'ought' counterpart for each category.
Produce: Twelve proposals with the person or group whose judgement they express.
Trace one boundary change into the evaluation.
Produce: A before-and-after judgement with a changed evidence requirement.
Design a challenge that an affected person can actually use.
Produce: One route to contest the proposal without accepting its framing.
Checks
Does the analyst's involvement resolve the legitimacy questions?
- Yes: expertise represents everyone
- No: expertise and the voice of affected people are different categories
- Yes: the data are numerical
What does 'ought' mean in this exercise?
- A hidden fact waiting to be discovered
- An explicit normative boundary proposal open to challenge
- The tutor's mandatory policy preference
Who is missing from the stated online completion measure?
- Only the vendor
- People unable to start the online process
- No one, because all applications are digital
Repair
A stakeholder list has twelve names and is labelled 'the twelve CSH questions answered'.
Changed case
The tenants' group gets one seat but cannot alter the agenda or challenge the savings target. Which part of the critique remains open?
Review criteria
- All twelve categories are addressed in both is and ought modes.
- Evidence, absence and normative proposals are distinguishable.
- An affected voice is not silently replaced by an expert or the practitioner.
- A boundary change alters a concrete evaluation or action.
- The answer includes a practicable route for challenge.
Confrontation Analysis: separate positions, options and credibility
Fictional case: a council and a charity are negotiating an evening advice service. The council alone can authorise funding (F). The charity alone can commit a trained team (T). Both publicly propose F=yes and T=yes. The council says it will withhold funding if the charity does not commit. The charity says it will not commit without funding. Both prefer a funded staffed service to no service. The charity doubts that the council's promised funding will survive its next budget meeting. A council manager's personal assurance is not a budget authorisation. The charity could offer an unpaid four-week pilot, but has not offered or agreed to do so.
List actors, controlled options and the four possible combinations.
Produce: An options table with controllers clearly marked.
Separate preference, intention and belief.
Produce: Three statements drawn from the case, each correctly typed.
Propose a change that addresses credibility without coercion.
Produce: One revised option and evidence that it is genuinely available.
Rebuild the table after the change.
Produce: A revised commitment record and one remaining uncertainty.
Checks
Who controls the charity's staffing commitment?
- The council, because it funds the service
- The charity
- The analyst drawing the table
Does a shared preferred outcome prove cooperation will occur?
- Yes
- No, beliefs about performance and credible commitments still matter
- Only if an analyst assigns utility scores
How should the possible unpaid pilot appear?
- As an agreed charity commitment
- As an uncommitted potential new option
- As proof that funding is unnecessary
Repair
The options table assigns 'charity supplies staff' to the council and treats a manager's assurance as secured funding.
Changed case
The council obtains authorised funding, but the charity now doubts it can recruit enough qualified staff. What changes in the analysis?
Review criteria
- Each option has its actual controller.
- Positions, fallbacks, preferences and beliefs remain distinct.
- The proposed change affects a specific credibility or feasibility issue.
- New options and revised preferences are allowed without inventing consent.
INFORMED Group Dynamics: investigate the distinctions behind a decision
Fictional case: three managers are choosing how to improve an advice service. Aisha supports a shared portal because residents should not have to repeat their story. Ben supports named advisers because difficult cases need continuity. Cara supports walk-in access because a portal-only service could exclude people. A meeting note labels Aisha 'efficient', Ben 'resistant' and Cara 'emotional'. No manager has agreed with those labels. Options are P: portal only; N: named adviser with telephone access; H: a hybrid service with named continuity and assisted digital access.
Investigate narratives before interpreting motives.
Produce: One neutral follow-up question for each manager.
Frame the orientations as provisional contrasting constructs.
Produce: Three distinctions, with a note that the speaker must validate them.
Make a small reflective option map.
Produce: A matrix of P, N and H against the three contrasts, retaining unknowns.
Explore distinctions and revise your first story.
Produce: A revised account and a question that could expose a real disagreement.
Checks
What is the status of 'Ben is resistant'?
- An observed fact
- An interpretation not validated by Ben
- A required INFORMED category
Who should validate a proposed personal construct?
- Only the analyst
- The person whose meaning is being represented
- The highest-ranking manager
Can the unscored map establish that H is optimal?
- Yes, because it sounds inclusive
- No: feasibility, evidence and people's judgements remain to be examined
- Yes, if it is drawn neatly
Repair
The facilitator invents three personality scores and uses them to explain why one manager must be excluded from the decision.
Changed case
Ben now says he supports the portal provided a resident can retain a named adviser. Revise the account.
Review criteria
- Narratives are sought before motives are assigned.
- Contrasting constructs remain provisional until checked with their owners.
- Unknowns are not converted into invented numeric scores.
- The map changes at least one inquiry or interpretation.
Interactive Management: construct a small interpretive structural model
Fictional case: a partnership is opening a shared referral service. Five conditions are proposed: A agreed decision authority; B permission to access required records; C a workable triage protocol; D an allocation process; E a follow-up review. For this exercise the group has explicitly agreed these immediate prerequisites: A before C, B before C, C before D, and D before E. There are no other immediate prerequisite links. The relation being modelled is 'must be in place before', not 'causes better outcomes'.
Create the immediate-link matrix in order A, B, C, D, E.
Produce: Five rows of five binary values, with a legend.
Add implied reachability and the identity diagonal.
Produce: A reachability matrix and one explained inferred link.
Draw a layered model and identify the foundations.
Produce: A dependency diagram with a stated direction.
Return the model to the people who supplied the relation.
Produce: Two questions that could change it.
Checks
Under the agreed prerequisite relation, does A reach E?
- Yes, through C and D
- No, because no direct A-to-E link was supplied
- Only if A and B are merged
Does the matrix establish that A causes a successful referral outcome?
- Yes
- No: it represents agreed prerequisites among these conditions
- Yes, because a binary model is objective
Which pair has no reachability in either direction in this model?
- A and B
- C and E
- D and E
Repair
A matrix mixes 'must precede', 'is preferred to' and 'is funded by' without labelling the difference, then applies transitive closure.
Changed case
The group adds E before C: learning from review is needed to revise triage. What should you do rather than pretending the original acyclic sequence still holds?
Review criteria
- The relation, direction and element meanings are explicit.
- The immediate and reachability matrices are distinguished.
- Inferred links and the independent A/B pair are correct.
- The learner returns mathematical structure to participant judgement.
Interactive planning: design the service you would choose now
Fictional case: residents must give the same household information separately to a repairs service, an advice service and a community transport service. Staff spend time chasing missing details. Residents avoid contact after repeated frustration. Each team meets its own response target by referring complicated cases elsewhere. The organisation is considering a faster referral form. For the design exercise, the existing arrangement has disappeared overnight, but current technology, legal responsibilities and today's environment remain. No unproven technology or compulsory data-sharing may be assumed.
Formulate the mess rather than selecting one isolated symptom.
Produce: Three mutually reinforcing problems and a consequence of continuing unchanged.
Design an arrangement you would choose now if the old one no longer existed.
Produce: A purposeful design with three capabilities and explicit constraints.
Specify ends, means and resource implications.
Produce: One desired result, two means and the resources or authority they require.
Work back to a feasible first move and its control process.
Produce: A small implementation step, prediction and review rule.
Checks
Which proposal fits an idealised design under the stated constraints?
- A service using currently available means that can learn and adapt
- A perfect future AI requiring no consent
- A forecast that present trends will continue
Why might a faster referral form leave the mess intact?
- All forms are harmful
- It may speed a process that still transfers rather than resolves need
- Technology cannot improve services
Repair
The group writes a perfect future vision, assumes unlimited money and full consent, and labels implementation 'roll out'.
Changed case
Residents reject a shared database but welcome a named coordinator. Does the design fail completely?
Review criteria
- Interdependent problems are represented as a mess.
- The design is desirable now, technologically feasible and adaptable.
- Ends, means, resources and authority are distinguished.
- Implementation includes participation and feedback, not just rollout.
Mosaic transformation and bubble strategy: sequence changes you can sustain
Fictional case: a service has two staff-days each week available for change, separate from its essential delivery work. A: remove duplicate reporting, needs two change-days in one week and releases two additional change-days per week from the following week. B: redesign triage, needs three change-days in one week and cannot start before A is complete. C: test a resident-led appointment process, needs four change-days in one week and cannot start before B. B and C release no additional capacity in this toy calculation. The released time is available only if managers stop demanding the old duplicate report. The C experiment needs a bounded group of cases and explicit interfaces with ordinary delivery.
Build the smallest feasible sequence.
Produce: A week-by-week capacity and dependency table.
Check the assumption that makes the sequence possible.
Produce: A release condition and an observable test.
Define the C bubble without cutting it off from reality.
Produce: Boundary, permissions, interfaces and stop/exit conditions.
Compare a benefit-first ranking with a feasible change sequence.
Produce: A short explanation of why the rankings differ.
Checks
When can B first start under the supplied assumptions?
- Week 1
- Week 2
- Only after C
If managers continue the duplicate report, what is the week-2 capacity?
- Four days regardless
- Two days
- Six days
What must a protected experiment retain?
- No connection to ordinary delivery
- Explicit interfaces and a route to learn, stop or integrate
- A guarantee that its results will be scaled
Repair
The programme launches A, B and C together because each has a strong business case, using the same two staff-days three times.
Changed case
A releases only one usable change-day each week. Can the original sequence continue?
Review criteria
- Capacity is counted once and release timing is explicit.
- Dependencies and contribution are distinguished.
- Released capacity is tested in actual work.
- The bubble has boundaries, interfaces and exit conditions.
Multi-methodology: justify the move from one method to another
Fictional case: a regional advice partnership faces rising waits. The sponsor calls this a capacity problem. Residents say they are referred repeatedly without help. Charities argue that eligibility rules exclude the people most in need. Service leads disagree about who can change the rules. Reliable arrival and completion counts exist for two services, but not for the whole partnership. The sponsor wants a recommendation in four weeks.
Separate the questions that require different kinds of inquiry.
Produce: Four questions, each with a needed output.
Choose a justified three-method sequence.
Produce: Method, question, output and limitation for each stage.
Specify the transition into the numerical model.
Produce: One translation rule and one thing that must not be translated automatically.
Set a trigger to change your method design.
Produce: A revision trigger and a stopping rule.
Checks
Which is a sound reason to add a method?
- It makes the report look comprehensive
- It answers an important question the current approach cannot adequately address
- It is the facilitator's favourite
Can a purposeful activity from SSM be treated immediately as a calibrated SD variable?
- Yes, both are systems methods
- No, its meaning, measurement and modelling role must be established
- Yes, provided the label is unchanged
Two learners choose different method sequences. Must one be wrong?
- Yes
- No, compare their purposes, assumptions, omissions and revision rules
- No, every sequence is equally useful
Repair
'We will do SSM, VSM, SD and CSH, then combine the diagrams into one master model.'
Changed case
The sponsor withdraws access to residents but still asks for a whole-system recommendation. What changes?
Review criteria
- Every selected method has a distinct inquiry purpose.
- Limitations and omitted questions are explicit.
- Transitions preserve differences between normative, conceptual and empirical claims.
- New evidence can change the sequence and scope.
Patterns of Strategy: model dependence, fit and timing
Fictional case: a local repair network needs a specialist supplier to complete 60% of its jobs. The network accounts for only 5% of that supplier's sales. Finding and qualifying an alternative supplier would take six months. The network has promised residents completion within one month. The supplier can change its delivery terms with two months' notice. Neighbouring networks use the same supplier but have not agreed to cooperate. A proposal says 'negotiate harder because we are a large organisation'.
Map the strategic relationship from both sides.
Produce: A two-direction dependence account with the purpose of each actor.
Test temporal fit.
Produce: A comparison of notice, response time and the resident promise.
Generate two moves that change the relationship rather than merely state an ambition.
Produce: Two moves, their preconditions and likely counterpart responses.
Specify the evidence that would change your strategic choice.
Produce: A signal and a conditional decision.
Checks
What is the stated gap between notice and alternative qualification?
- Four months
- One month
- Eight months
Do neighbouring networks' purchases automatically increase this network's power?
- Yes
- No: cooperation and decision rights are not established
- Only if added to a spreadsheet
Which is a strategic move rather than an aspiration?
- Become the best repair network
- Qualify a second supplier before the current contract becomes vulnerable
- Demand excellence
Repair
A SWOT list declares 'strength: we are big' and concludes the supplier must accept any terms.
Changed case
The supplier now depends on this network for 40% of its sales. What follows, and what does not?
Review criteria
- The relationship is represented from both actors' positions.
- Dependence and timing are tested with the supplied facts.
- Moves alter relationships or options rather than restating goals.
- Uncommitted partners and uncertain responses are not treated as controlled resources.
Socio-Technical Systems: redesign the work and technology together
Fictional case: a repair service introduces handheld job software. The software permits only one fault code per visit. Workers often find several linked faults, but cannot alter the code or order additional parts. A supervisor can approve changes the next morning. Residents then receive extra appointments. Management proposes training workers to comply more consistently. Workers propose discretion to record linked faults, while the safety lead insists that electrical work remains restricted to qualified staff.
Define the primary task and the social and technical arrangements.
Produce: A short work-system account rather than a technology inventory.
Locate a significant variance and where it is currently controlled.
Produce: A variance-control trace.
Design a joint change with minimum necessary constraints.
Produce: A paired social/technical proposal with a safety boundary.
Test the whole work system.
Produce: Three measures and a participative review question.
Checks
Which proposal addresses both social and technical design?
- Add fault fields but prohibit workers from using them
- Enable linked-fault recording and give qualified workers bounded authority to act
- Tell workers to be more positive
Where is the variance first encountered?
- At the home by the worker
- Only at the board meeting
- In the annual budget
Does local discretion require removing electrical qualification rules?
- Yes
- No: discretion can operate within necessary competence and safety constraints
- Only if productivity rises
Repair
The project buys the software first and calls a final staff briefing 'participative socio-technical design'.
Changed case
Workers can now edit codes, but the parts store still accepts orders only from supervisors. What remains?
Review criteria
- The primary task and social/technical interdependence are explicit.
- Variation is traced through actual work and control locations.
- Authority, competence, information and technology change coherently.
- Safety and effects on residents and other workers remain visible.
Syntegration: test the communication structure before the event
Fictional case: a regional partnership needs to consider twelve interdependent topics about access to advice. For the classic structural model, use twelve topic groups and thirty participants, with each person a member of two groups and each group having five members. This is a membership calculation only; critic, observer and facilitator roles are additional and must not be erased from a real design. For a small original rehearsal, focus on three topics: A access, B workforce and C funding. Lia is in A and B; Mo in B and C; Ren in C and A. In round 1, A proposes longer opening hours, B identifies an unfilled staffing requirement and C states a fixed short-term budget.
Check the classic membership arithmetic.
Produce: Two independent counts of membership places.
Trace how the three-topic rehearsal can change the next discussion.
Produce: A round-2 note for each topic.
Identify what is lost in three isolated breakout groups followed by one report-back.
Produce: One structural difference and one possible consequence.
Test a proposed event design for exclusion and premature closure.
Produce: Two design questions and a record of unresolved disagreement.
Checks
How many membership places are in the stated classic model?
- 30
- 60
- 120
Does matching the place count prove a full Syntegrity design?
- Yes
- No: topology, roles, timing, participation and protocol still matter
- Yes, if everyone attends
Is the original three-topic rehearsal a certified smaller Syntegration?
- Yes
- No, it isolates one communication idea for practice
- Only when the answers are correct
Repair
A conference has twelve tables and thirty delegates. The organiser declares the process to be Syntegrity and promises consensus.
Changed case
Lia attends A only and Mo attends B only, while Ren carries every message. What structural risk appears?
Review criteria
- Membership arithmetic is correct and distinguished from total roles.
- Overlapping participation and iteration have an explained purpose.
- The reduced exercise is not passed off as the complete protocol.
- Legitimacy and disagreement are not assumed to follow from symmetry.
SODA: build a means-ends map without erasing disagreement
Fictional case: a service leader says 'Give advisers more discretion so people receive help without repeated approval.' A finance lead says 'Keep approval for unusual spending so the service remains within its budget.' An adviser says 'Waiting for approval means people call again, which leaves less time to resolve cases.' A resident says 'I need a clear answer, not another referral.' None has agreed that eliminating all approval is desirable.
Construct four short concepts with meaningful contrasts.
Produce: Four bipolar or contrast-explicit concepts, retaining the speaker's meaning.
Connect means to ends and retain the tension.
Produce: A directed map plus a statement explaining every arrow.
Find a candidate option that connects the concerns.
Produce: One option and a question for negotiation.
Checks
Which statement is supported by the case?
- Everyone wants all approvals abolished
- Some approvals may contribute to repeat demand while others protect a valued constraint
- Finance is opposed to residents receiving help
What is the status of an arrow derived from an interview statement?
- A causal belief to explore and test
- A proven physical law
- A vote
Repair
The facilitator merges 'financial control' and 'bureaucracy' into one negative concept and reports consensus.
Changed case
The adviser says most delays occur after approval, while waiting for a specialist. What changes?
Review criteria
- Concepts preserve meaning through useful contrasts.
- Arrows form explainable means-ends reasoning.
- Disagreement and inferred claims remain visible.
- An option is distinguished from negotiated agreement.
Fractal Enterprise Model: alternate processes and the assets they depend on
Fictional case: an advice charity delivers appointments. This process uses trained advisers, an appointment system and suitable rooms. Advisers are maintained through supervision and training; new advisers come through recruitment. The appointment system is maintained by a supplier-management process. Rooms depend on facilities maintenance. Every supporting process also needs people, information and time. Recruitment was suspended because managers counted only time spent on appointments as productive.
Separate processes from assets.
Produce: Two lists and a notation legend.
Construct one complete alternating chain.
Produce: A process-asset-process-asset chain with typed relations.
Use the model to challenge the productivity decision.
Produce: A dependency-based prediction and an observation plan.
Checks
Which is an asset in this case?
- Recruit advisers
- Available trained adviser capability
- Conduct supervision
Does drawing the dependency predict the exact staffing shortage next year?
- Yes
- No, the dependency model needs further quantitative assumptions for that prediction
- Yes, if every box is connected
Repair
A chronological flowchart of appointment steps is labelled FEM, with every item drawn as the same type of box.
Changed case
The charity outsources the appointment system. Does that remove it from the model?
Review criteria
- Processes and assets are distinct.
- Use and maintenance/acquisition relations are labelled.
- At least one supporting process's own dependencies are exposed.
- Structural insight is not passed off as a quantitative forecast.
Critical social learning: build a learning relationship across boundaries
Fictional case: a council, a charity and a residents' group meet monthly about access to advice. Each presents success figures, but nobody changes practice. The council values prompt closure, the charity values sustained relationships, and residents value getting an answer they can use. One joint case worked unusually well because a resident and adviser could contact a named decision-maker directly. That person has since left. Residents say meetings use unfamiliar language and are held during their working hours.
Map the appreciative differences.
Produce: Three judgements about what counts as a good result, without forcing one metric.
Investigate the successful case without ignoring failure.
Produce: Evidence of enabling conditions and a question about transfer.
Design an action-research cycle.
Produce: A jointly owned question, small action, observation and reflection.
Create a double-loop learning test.
Produce: One change to action and one change to a governing assumption.
Checks
Does a positive case remove the need to investigate exclusions?
- Yes
- No, enabling conditions and excluded experiences both matter
- Only during Appreciative Inquiry
Which is the clearest double-loop move?
- Produce the same closure report faster
- Question whether closure represents the purpose people value
- Increase meeting frequency without changing anything else
Repair
The group records attendance as proof that a social learning system exists and that residents share the official purpose.
Changed case
The named decision route improves speed but residents feel pressured to accept an answer. What should the learning group do?
Review criteria
- Different appreciations and named traditions are not collapsed.
- Participation changes who can influence the inquiry.
- The action cycle includes evidence and revision.
- A governing assumption can be questioned.
Service systems: follow demand and distinguish activity from resolution
Fictional case: ten consecutive calls to a repair desk are recorded. Calls 1, 2, 4 and 8 request new repairs. Calls 3, 5 and 9 ask why an agreed visit did not happen. Calls 6 and 10 report that a supposedly completed repair still leaks. Call 7 asks for an unrelated general-information leaflet. Staff are measured on average handling time. One repair trace shows three minutes on the first call, two days waiting for approval, a failed visit because the wrong part was sent, and two more calls before a completed repair eight days later.
Classify the demand using the reason for contact.
Produce: A ten-row classification with a stated definition.
Trace value, waiting and rework end to end.
Produce: A resident-centred work trace.
Design a PDSA test rather than a rollout.
Produce: Prediction, bounded action, evidence and decision.
Checks
How many calls in the sample arose from missed visits or ineffective repairs?
- Four
- Five
- Nine
Does the sample prove 50% of all future demand is failure-related?
- Yes
- No, sample size, timing and representativeness matter
- Yes, because ten is a round number
What belongs in the Study step of PDSA?
- Compare observations with the prediction and revise the theory
- Declare the change a success because it was implemented
- Check only that staff complied
Repair
'Calls are waste. Block repeat callers and the demand problem will disappear.'
Changed case
Handling time rises by one minute while failed visits fall. Can the change still be worthwhile?
Review criteria
- Demand is classified from purpose and cause, not blame.
- A sample is not treated as a population certainty.
- End-to-end resolution remains visible.
- The test includes a prediction and a learning decision.
Participative inquiry: test who your evidence leaves out
Fictional case: a transport partnership surveys people through its smartphone ticketing app. Managers report that 92% of respondents are satisfied. The response count and invitation count are not given. Drivers say some passengers pay cash and need help. A disability group cannot attend the partnership's daytime meetings. The operator controls timetables, the council controls a subsidy, and passengers have informal influence but no direct timetable authority.
Map interest, influence, control and affectedness separately.
Produce: A four-column actor table.
Audit the 92% statement.
Produce: Three missing facts and one excluded group.
Design a proportionate mixed inquiry.
Produce: Two evidence routes and an accessible participation option.
Plan a group comparison without forced consensus.
Produce: An opening question, a participation safeguard and a record format.
Checks
Can 92% app-user satisfaction establish satisfaction among all potential passengers?
- Yes
- No: the sampling boundary excludes some relevant people
- Only if the chart is clear
Which interview opening best avoids importing a diagnosis?
- Why do you resist digital ticketing?
- Tell me about the last journey you tried to make
- Do you agree the app is convenient?
Repair
Only high-power stakeholders are invited because low-power passengers cannot affect delivery.
Changed case
The disability group declines the proposed workshop. What should not be inferred, and what is a better next move?
Review criteria
- Influence and affectedness are not conflated.
- Sampling limits are explicit.
- Questions seek accounts before explanations.
- Participation has a practical route to affect the inquiry.
Futures and probes: test robustness without pretending to predict
Fictional case: an advice service is choosing capacity arrangements. Scenarios are low demand at 80 cases/week and high demand at 140. Option A has fixed capacity 100; option B has fixed capacity 120; option C has baseline 90 and can add 60, but only after two weeks. Costs and case complexity are not supplied. Separately, the service does not know whether residents avoid appointments because of timing, lack of trust or confusing letters. No one can safely be denied essential support for an experiment.
Calculate the initial capacity gap under each demand scenario.
Produce: A two-by-three table, treating C's delay explicitly.
State what is needed before choosing an option.
Produce: Three missing assumptions and a robustness judgement.
Design two parallel probes for different explanations of avoidance.
Produce: Two small interventions with distinct hypotheses.
Define sensing, dampening and amplification before acting.
Produce: A named observer, review point and stop/adapt rules.
Checks
At high demand, what is C's initial weekly shortfall?
- 50 cases
- 10 cases
- No shortfall
Which is a safe-to-fail design feature?
- A large mandatory rollout called a pilot
- A bounded, monitored action with an available stop or adaptation route
- A promise that nothing unexpected will happen
Repair
The plan chooses the highest eventual capacity, ignores mobilisation delay and calls a compulsory service withdrawal an experiment.
Changed case
Extra capacity now takes six weeks, while high demand is expected to last four. How does that affect C?
Review criteria
- Scenarios and forecasts are distinguished.
- Capacity arithmetic and response delays are correct.
- Options are not ranked without criteria and missing assumptions.
- Probes test distinct possibilities and have real safeguards.
Ethics and power: make an intervention's authority and effects explicit
Fictional case: a sponsor asks an analyst to rank families for extra support using attendance records. The sponsor controls the project budget but does not control the schools' data. Families have not been told about the proposed use. The analyst suspects that absence may reflect transport barriers, caring responsibilities or illness, but no inquiry has tested those explanations. The sponsor says consultation will slow delivery and asks for a hidden pilot.
Map authority and affectedness.
Produce: A distinction between budget authority, data rights and effects on people.
Separate observations, labels and causal claims.
Produce: Three typed statements and alternative explanations.
Design an ethically defensible next inquiry.
Produce: A narrower action with permission, safeguards and a stop condition.
Prepare a constructive challenge to the sponsor.
Produce: A short statement that names the risk and a workable alternative.
Checks
Does budget ownership automatically authorise the proposed data use?
- Yes
- No
- Only during a pilot
What is the status of 'absence means low motivation'?
- An established fact
- An untested causal interpretation
- A harmless administrative label
Repair
The ethical review says 'all stakeholders consulted' because the sponsor approved the model.
Changed case
Families agree to discuss barriers but decline automated ranking. What follows?
Review criteria
- Authority, influence, affectedness and permission are distinct.
- Proxies and causal claims are not treated as established truth.
- The intervention has transparency, safeguards and a challenge route.
- Sector-specific compliance is explicitly left for qualified checking.
Evaluation and reflexivity: test the claim you are tempted to make
Fictional case: after a new triage process, recorded average waiting time falls from 20 to 12 days. During the same period, 30 difficult cases are moved to another team and excluded from the report. Staff report less duplicated work. Three residents describe improved clarity; two describe feeling dismissed. The analyst designed the change and chose the waiting-time measure. There is no comparison group and no information about seasonal demand.
Calculate the reported change, then bound the claim.
Produce: An absolute change, a percentage change and a qualification.
Build a small evaluation matrix.
Produce: Outcome, process, distribution and unintended-effect evidence.
Name your own influence on the evaluation.
Produce: A reflexive note and one practical safeguard.
Decide what can responsibly be reported now.
Produce: A short finding and a next learning action.
Checks
What is the reported percentage reduction from 20 to 12?
- 8%
- 40%
- 60%
Which finding is currently defensible?
- Triage caused a 40% improvement for everyone
- The recorded average fell, but the causal and population claim remains uncertain
- Nothing can be learned
What should happen to the two negative resident accounts?
- Remove them as outliers
- Investigate and retain them as possible evidence of unintended effects
- Treat them as proof all results are false
Repair
The evaluator calls the pilot successful because the selected indicator improved, while excluding transferred cases and dissenting accounts.
Changed case
Transferred cases now show waits of 50 days. What does that tell you, and what else is needed?
Review criteria
- The arithmetic is correct.
- Description, causation and generalisation are distinguished.
- Displaced and negative effects remain in view.
- The evaluator's role changes the safeguards and inquiry.
Systemic constellations: use a spatial representation as a hypothesis
Fictional case: a project coordinator receives requests from a sponsor and three service teams. The sponsor's messages are frequent; the teams' concerns often arrive late. Residents are discussed but do not attend project meetings. Use five labelled objects: coordinator, sponsor, teams, residents and project purpose. You may instead draw the arrangement or describe it in text; no movement or personal disclosure is required.
Arrange the objects to represent one interpretation of attention and contact.
Produce: A diagram or text account with a stated meaning for distance and direction.
Describe before interpreting.
Produce: Two observations about the arrangement and two hypotheses about work.
Try a different arrangement and identify a real-world test.
Produce: One changed relationship and one observable inquiry.
Checks
What can the tabletop arrangement directly establish?
- The sponsor's hidden motives
- The modeller's current representation and questions it raises
- The feelings of residents who are absent
May a learner decline embodied or personal participation?
- Yes, use drawing or text and respect the boundary
- No, refusal proves resistance
- Only if the facilitator approves the reason
Repair
The facilitator says 'the objects prove the sponsor resents residents' and instructs the coordinator to confront them.
Changed case
A second learner places the teams closest to the coordinator. How should you compare the models?
Review criteria
- The representation's conventions and author are explicit.
- Impressions are not treated as facts about absent people.
- Participation is optional and bounded.
- The exercise generates a testable inquiry rather than a diagnosis.
Systemic conversations: expose inference and repair a broken commitment
Fictional transcript: Sam says 'Could you get the referral picture sorted soon?' Lee replies 'I'll see what I can do.' Three days later Sam says 'You never follow through.' Lee says 'You keep changing what you want.' Sam privately thinks Lee is avoiding responsibility. Lee privately thinks Sam wants a complete service redesign, which Lee cannot authorise. No deadline, output, quality criterion or decision right was agreed.
Separate what was said from what was inferred.
Produce: A two-column observation/inference record.
Map the commitment sequence.
Produce: Request, clarification, promise, completion and satisfaction, marking absent stages.
Draft a repair that combines advocacy and inquiry.
Produce: A short replacement exchange.
Name an undiscussable safely.
Produce: One observation and an invitation, without a forced confession.
Checks
Is 'I'll see what I can do' an unambiguous promise of a complete redesign by Friday?
- Yes
- No
- Yes, because Sam is senior
Which is an inference rather than a supplied observation?
- Lee said 'I'll see what I can do'
- Lee is avoiding responsibility
- Sam asked for the referral picture to be sorted
Repair
The coach instructs Lee to admit avoidance and labels any disagreement a defensive routine.
Changed case
Lee now promises the one-page map, but a service lead refuses access. What should happen before Friday?
Review criteria
- Observation and inference are distinguished.
- Missing commitment stages are identified.
- The repair includes scope, timing, conditions and authority.
- The helper's own interpretation remains challengeable.