Post #3528718

2026-05-08 07:53 UTC

Organisational Dysfunction of the Day The error factory Context: Something goes wrong. A post-mortem is held, a root cause is identified, and a fix is put in place. A few months later, the same thing goes wrong again, in a slightly different form. The organisation responds with more process, more checklists, more training. The error rate stays stubbornly high. Leadership concludes that people are not following the processes correctly, so more oversight is added. The errors continue. Nobody questions whether the structure itself might be amplifying the mistakes rather than catching them. OST explains: The research is mathematically precise on this. In a DP1 (bureaucratic) structure, if five people each make sound judgements eight times out of ten, the probability they give you correct unanimous advice is only one in three. The more you control through hierarchy, the deeper you move into error. In a DP2 structure with the same five people and the same fallibility, wrong unanimous advice occurs only three times in ten thousand. In DP1, errors get amplified because asymmetry and competition mean people filter information to serve their own position rather than the truth. In DP2, the same errors become learning opportunities because everyone has shared responsibility, and it is in nobody's interest to hide a mistake. The diagnosis of inadequate training and the prescription of more training will not reliably reduce error rates; that requires attention to the underlying structural cause. #OpenSystemsTheory #SocioTechnical #OrgDesign #management

Replies (1)

@trondhjort@hachyderm.io 2026-05-11 06:37

Organisational Dysfunction of the Day DORA, the wrong way round Context: The DORA metrics have become the gold standard for measuring engineering performance. Deployment frequency, lead time for changes, change failure rate, and time to restore service. The four key metrics. Teams build dashboards around them, set quarterly targets, and run improvement initiatives to move the numbers in the right direction. Some teams genuinely improve, while others find the numbers are stubborn or that improvements one quarter quietly reverse the next. Leadership concludes that the teams need more discipline, better tooling, or another round of training. What gets called cargo culting (for lack of a better term) in the industry, copying the visible practices without the underlying conditions, is exactly this pattern. OST explains: DORA was designed as a research instrument, not as a target system. The metrics are downstream signals of healthy delivery, not the drivers of it. Healthy delivery is when self-managing teams own the whole product, make decisions without escalation, and have tight feedback loops with the people they serve. Take those structural conditions away, and the numbers regress, no matter how many dashboards you build. Treating DORA as a goal in DP1 (bureaucratic) is exactly the goal displacement Goodhart warned about: the moment a measure becomes a target, it stops being a good measure. In DP2, the self-managing-group structure, the same numbers emerge naturally as side effects of work well done. You do not need to chase them. You need to build the conditions that produce them. #OpenSystemsTheory #SocioTechnical #OrgDesign #DevOps

Open ##3528717