Four Stages: a Six-Stage Critique
A second edition of the corpus's Four Stages critique, and a markedly less generous one than the 2020 original, written after six years of further practice. The critique is structured, with deliberate irony, as six stages that are not stages and need not be read in order. It opens by granting the model its genuine use, a vocabulary for noticing that psychological safety is not binary, then works through six problems. The empirical one: no peer-reviewed body of research shows teams moving through inclusion, learner, contributor and challenger safety in sequence or at all, and the piece argues the commercial architecture of trademarked vocabulary, proprietary assessment and sold certification creates a structural disincentive ever to test it. The taxonomy it runs on, borrowed with credit from Snowden's reading of Box, is that all models are wrong but not all in the same way: the honest simplification, the incoherent model, and the coherent model dragged outside its domain, and it argues the Four Stages manages the last two while claiming the licence of the first. The dimensionality problem: inclusion, learning, contributing and challenging vary independently, so the ladder is not a simplification of the phenomenon but a distortion of it, and sequential intervention aims at the wrong layer. The complexity mismatch: a stage model applied to a complex phenomenon manufactures a false sense of control, and when reality refuses to match the diagram the conclusion drawn is usually that the application was faulty rather than the model. The politics: the model routes permission to challenge through the person being challenged, which reproduces at its summit the asymmetry it claims to dissolve, and a permission that can be withheld is a revocable licence rather than safety. Follett's power-with against power-over does the analytical work here, as elsewhere in this corpus. The equity problem: respect and permission are things the powerful confer, the model has no concept of substrate, intersectionality or the unequal cost of speaking, and a team is only as safe as its least safe member. And the iatrogenic risk, which the piece treats as the most serious: an organisation that adopts the model, runs the assessment and declares the problem solved can leave its least safe people less safe and officially invisible, harm caused by the treatment itself and worse than doing nothing. The closing position is constructive: psychological safety as a shifting profile rather than a ladder, patterns and distributions rather than assigned stages, substrate alongside leader behaviour, and interventions as probes rather than milestones.
Keep notes on this, flag it to come back to, and build your own map. Create a free account →