The library
Delivery dynamics & feedback
How does the project change as work happens?
Explore dependency loops, learning, rework, reviews and site occupancy through deliberately small models.

Critical-path calculation
Find the delay that moves the finish
Change durations and watch total float, critical routes and project completion respond.
What to try & what it leaves out
Try. Add one unit to A, reset, then add one to B. One uses float; the other moves completion.
Limits. A deterministic finish-to-start network, without calendars, resources, lags, uncertainty or costs.
Earlier example, repaired and reviewed 2 October 2026. This review checks the stated model and interactions; it does not establish usefulness in practice.
GitHub source (may require access)
Bounded scheduling model
See what a plan rewrite changes
Try graph rewrites and compare dependencies, resource use, duration and uncertainty.
What to try & what it leaves out
Try. Collapse two serial steps and expand them again. Inspect the resource schedule, then test an explicitly assumed dependency change.
Limits. Heuristic resource scheduling; a graph rewrite is not proof that real work is safely interchangeable.
Earlier example, repaired and reviewed 2 October 2026. This review checks the stated model and interactions; it does not establish usefulness in practice.
GitHub source (may require access)
Scripted decision simulation
A project review changes the next move
Walk through a scripted project narrative using prediction error and active-inference language as an analogy.
What to try & what it leaves out
Try. Advance through the reviews and compare the beliefs, progress, recorded decisions and later outcomes.
Limits. Scripted teaching values, not variational or expected free-energy calculations, inferred policies or a validated Active Inference model.
Earlier example, repaired and reviewed 2 October 2026. This review checks the stated model and interactions; it does not establish usefulness in practice.
GitHub source (may require access)
Schedule-risk calculation · supplied assumptions
Explore the assumptions behind a delivery date
Build a small schedule, sample duration uncertainty and inspect completion percentiles and activity criticality.
What to try & what it leaves out
Try. Change an activity’s range or distribution assumption, rerun the simulation and compare the distribution with the baseline.
Limits. Probabilities describe the supplied model, not calibrated delivery confidence. Inputs and omitted risks still require judgement.
Earlier example, repaired and reviewed 2 October 2026. This review checks the stated model and interactions; it does not establish usefulness in practice.
GitHub source (may require access)
Decision records
Trace decisions beside overlapping work
Compare task dates and decision response times in a knowledge-graph project.
What to try & what it leaves out
Try. Select Security review, then select a task bar. Trace the associations across both charts; clear the selection to restore the whole picture.
Limits. No dated decisions, causal gating rules or dependency network are supplied, so the chart cannot attribute schedule delay or calculate a critical path.
Earlier example, repaired and reviewed 2 October 2026. This review checks the stated model and interactions; it does not establish usefulness in practice.
GitHub source (may require access)
Bounded cost–time calculation
Time Exchange
Explore how discrete task-acceleration choices change project duration, additional cost and the critical path.
What to try & what it leaves out
Try. Set a budget, compare the time–cost frontier and inspect the selected schedule alongside the procurement proposal.
Limits. Exact only within the bounded supplied crash menus and precedence model; excludes shared resource constraints and validated incentive payments.
Earlier example, repaired and reviewed 3 October 2026. This review checks the stated model and interactions; it does not establish usefulness in practice.
GitHub source (may require access)Dependency analysis · worked calculation
Find the feedback groups.
Which activities depend on one another in a loop? Reorder a six-activity matrix to expose the coupled groups while keeping every dependency.
What to try, assumptions & source
Inputs. Six named activities and editable, binary information dependencies. A mark means that the row activity receives information from the column activity.
Try this. Compare the one-feedback-group and one-way-flow examples. Change a dependency and inspect which activities belong together. The second tab separately illustrates a change of coordinates.
Limits. Dependency groups do not supply a feasible schedule. This example has no durations, resource capacities, coupling strengths or iteration counts; its three rational-matrix examples are not a general solver.
The app explicitly records a revision on 20 September 2026, correcting an earlier comparison from June 2025. This is the app's revision date, not a claim about later use.
Maintained source; original revision recorded · Current model checks and their scope

Response model
How changes spread between project interfaces
Try explicit feedback and spillover rules across five organisational interfaces.
What to try & what it leaves out
Try. Clarify scope, compare local and neighbouring changes, then alter a coupling assumption and undo the intervention.
Limits. The coefficients and equal-weight index express assumptions about organisational performance.
Earlier example, repaired and reviewed 2 October 2026. This review checks the stated model and interactions; it does not establish usefulness in practice.
GitHub source (may require access)
Stochastic model
A software team learns and reworks
Explore assumed feedback between delivery, learning, debt, reviews, scope and morale.
What to try & what it leaves out
Try. Advance a week, switch effects off and on, then run and reset. Read the event log alongside the trajectory.
Limits. Supplied stochastic rules and coefficients drive the simulation. Reset keeps your effect choices.
Earlier example, repaired and reviewed 2 October 2026. This review checks the stated model and interactions; it does not establish usefulness in practice.
GitHub source (may require access)
Flow model
Waste-route capacity
Follow material through treatment, assay, storage and transport; compare capacity, outages, rework and queues.
What to try & what it leaves out
Try. Change storage or transport, then compare a seeded run with a stress ensemble and inspect unfinished inventory.
Limits. Fractional-flow rules with stated timing and accounting assumptions; no physical hazard model.
Earlier example, repaired and reviewed 3 October 2026. This review checks the stated model and interactions; it does not establish usefulness in practice.
GitHub source (may require access)
Occupancy model
A day on the pool-and-gym site
Read and run a downloadable NetLogo model of workers and deliveries over one day.
What to try & what it leaves out
Try. Download the model, load it into NetLogo Web, set worker count, then run setup and go.
Limits. A small occupancy illustration with fixed schedules. It cannot establish safe staffing, site capacity or construction logistics.
Earlier example, repaired and reviewed 2 October 2026. This review checks the stated model and interactions; it does not establish usefulness in practice.
GitHub source (may require access)Linear feedback calculation
See when feedback settles or grows
Step through a three-state signed feedback model and inspect how contributions and eigenvalues explain its behaviour.
What to try & what it leaves out
Try. Change a gain, follow its contribution through a feedback loop, and compare the trajectory with the stability calculation.
Limits. A discrete linear teaching model with supplied gains; it does not simulate nuclear governance, regulatory approval or a calibrated delivery system.
Earlier example, repaired and reviewed 3 October 2026. This review checks the stated model and interactions; it does not establish usefulness in practice.
GitHub source (may require access)Dependencies & dates
Trace dependencies across phases and time
Compare bundled phase and gate relationships with faceted timelines in four supplied programme views.
What to try & what it leaves out
Try. Select a node to follow its incident dependencies, then inspect dated activities and calendar overlaps in the timeline.
Limits. Calendar overlap alone does not prove resource conflict, criticality or delay.
Earlier example, repaired and reviewed 3 October 2026. This review checks the stated model and interactions; it does not establish usefulness in practice.
GitHub source (may require access)Coordination model
What makes concurrent change costly?
Change the assumed cost curve, split the same work between teams, and count the interfaces that remain.
What to try & what it leaves out
Try. Use three areas and three teams; increase the cost per cross-team pair until splitting costs more.
Limits. Supplied cost functions and equal-weight areas; no empirical capacity threshold or calibrated organisational model.
Earlier example, repaired and reviewed 3 October 2026. This review checks the stated model and interactions; it does not establish usefulness in practice.
GitHub source (may require access)