npbuilds Ackoff Mess — Systems of Interrelated Problems
Russell Ackoff's diagnostic move: most real problems are not problems, they are messes. A mess is a system of interrelated problems whose interactions matter as much as the individual problems. Treating a mess as a problem — picking one element and solving it — typically makes the mess worse, because the solution propagates through the system in ways the problem-framing doesn't capture.
The mess concept extends Rittel & Webber's wicked classification. Where wicked problems resist clean formulation, messes resist clean decomposition — they cannot be split into atomic sub-problems without losing the structure that makes them what they are.
The Three Levels
| Level | Definition | Example |
|---|---|---|
| Puzzle | Has a unique solution; constraints fully specified | Sudoku; debugging a unit test with a known invariant |
| Problem | Has multiple possible solutions; trade-offs exist; objectives can be specified | Hiring decision with criteria; route planning |
| Mess | A system of interrelated problems; the interactions matter; no single objective dominates | Healthcare reform; organizational dysfunction; climate policy |
A mess is not just a "big problem." It is a different category. The distinguishing feature: solving any individual sub-problem leaves the others worse off, because the sub-problems are causally entangled.
Four Modes of Treating a Mess
Ackoff identified four ways to deal with a mess. Most practitioners default to mode #3 (solve), which is the worst choice for true messes.
| Mode | Definition | When to use |
|---|---|---|
| Absolve | Ignore the mess; hope it goes away or someone else handles it | Rarely correct, but sometimes the only option for problems outside your sphere |
| Resolve | Use past experience and judgment to find an outcome that's "good enough" | When stakes are low and the mess is familiar |
| Solve | Apply formal methods (optimization, decision analysis) to find the best outcome | Only valid for problems, not messes — using this on a mess produces locally optimal, globally bad results |
| Dissolve | Redesign the system or the framing so the mess no longer exists | The Ackoff move. The hardest, most powerful, and most often correct choice for messes |
Dissolution doesn't solve the mess; it makes the mess unnecessary. Example: if "employees keep gaming the bonus system" is the apparent problem, solving means tightening rules; dissolving means removing the bonus system or restructuring the work so the incentive misalignment doesn't exist.
Diagnostic: Mess vs Problem
You are looking at a mess (not a problem) when:
- Every solution you propose creates two new problems.
- Stakeholders disagree on what the problem is, not just what to do about it.
- The "obvious" answer has been tried before and failed; trying it harder won't help.
- Multiple departments / disciplines / stakeholders each hold one piece of the truth, and none of their pieces fit together.
- The problem keeps "moving" — it shows up in different forms, but there's a felt sense it's the same underlying thing.
- Improving any single metric makes another important metric worse.
If three or more of these apply, the unit of analysis is wrong. You are not looking at a problem; you are looking at a mess in problem clothing.
How binding-vow uses Ackoff
When problem-typology (Phase 2) returns mess, six-eyes's response is structurally different from its response to wicked or ill-defined:
- Don't decompose with
claim-decomposer— atomic decomposition destroys the mess's structure. - Don't pick a single audience — messes typically have multiple stakeholders whose framings differ; route to multiple compression skills.
- Default to deep mode — quick mode will produce a locally good statement that worsens the mess.
- Apply
stakeholder-rotatorbefore any reframing — callsvalues-excavator(philosophy/ethics) for whose interests are at stake. - Consider dissolution before solution — ask: what change to the system would make this question moot?
The most common output for a mess-typed problem is not a single problem statement. It is a map of the mess (sub-problems and their interactions) plus a recommended dissolution candidate, with the original framing flagged as misleading.
Common Mistakes
- Treating a mess as a problem and applying optimization. Local optimum, global pessimum.
- Decomposing a mess. The interactions ARE the structure; decomposition discards them.
- Picking the loudest stakeholder's framing as canonical. The mess exists because multiple framings are simultaneously partially true.
- Believing dissolution is always available. Sometimes the system can't be redesigned in your sphere of action; in those cases, resolve (mode #2) is the honest choice.
- Confusing a wicked problem with a mess. Wicked = formulation-resistant single problem; mess = decomposition-resistant problem-system. Heuristic: if you can write one clear problem statement that everyone agrees is the right one (even without a known solution), it's wicked, not a mess.
Connections
problem-typology(binding-vow, future) — invokes this skill when classifying as "mess"wicked-vs-tame(binding-vow) — the upstream typology that distinguishes mess from wickedstakeholder-rotator(binding-vow, future) — primary tool for navigating a messvalues-excavator(philosophy/ethics) — surfaces the competing values that constitute the messclaim-decomposer(research) — explicitly not called for messes (use only on individual problems extracted from the mess map)
Sources
- Ackoff, R. L. (1974). Redesigning the Future: A Systems Approach to Societal Problems. Wiley.
- Ackoff, R. L. (1979). "The Future of Operational Research is Past." Journal of the Operational Research Society, 30(2), 93–104. (The "puzzle / problem / mess" distinction in concise form.)
- Ackoff, R. L. (1981). Creating the Corporate Future. Wiley. (Dissolution as a primary intervention mode.)
- Pidd, M. (2003). Tools for Thinking: Modelling in Management Science (2nd ed.). Wiley. (Modern OR treatment of the mess concept.)