A couple of weeks ago, Ofsted’s Sean Harford said this at a CurriculumEd conference in Lichfield: “Personally, I don’t believe that 90 per cent of schools in the country have good or better behaviour across the school.”

The complaints from the unions’ Easter conferences about the threats and violence suffered by many teachers suggest that Mr Harford’s scepticism is well-founded. But he speaks as the inspectorate’s national director of education. What should we make of his statement? If a very senior member of staff at Ofsted lacks confidence in its judgements, we should worry - as we might about his proposed solution to such under-reporting.

Knowing that children and teachers will “hold things together for a day” when inspection comes, he reckons inspectors should talk individually to lunchtime/breaktime supervisors and NQTs/trainees. So will they then make their judgement on the basis of what those people (who also might “hold it together”) say? Such information may offer useful triangulation, but it’s surely not first-hand, observed evidence, and thus no reliable basis for possibly critical judgement.

How does Ofsted make reliable judgements? Indeed, does it? Mary Bousted, joint-general secretary of the NEU teaching union, stressed this week in Tes the size of the challenge Ofsted has set itself in its new framework, “not only to inspect the breadth and balance of a school’s curriculum, but also its quality”. She queried how this can be adequately achieved in a two-day inspection.

‘Ofsted is as welcome as a plague of boils’

Ofsted’s boss, Amanda Spielman, protests that, if the swathes of data now commonplace in schools are still produced, she’ll ask who did it: was it worthwhile use of their time? An implication, perhaps, that the teacher who did so should have been in the classroom? That particular salary is likely nowadays to go to a non-teaching data-manager.

Yet it will be a rare and brave school that backs off on data collection, let alone avoids a new paperchase, as it accumulates evidence for inspectors that teachers’ “resources… clearly support the intent of a coherently planned curriculum, sequenced towards cumulatively sufficient knowledge and skills for future learning and employment”. They can’t take the chance.

Mary Bousted explains. It’s all about the “high stakes for heads and teachers whose careers are… ended by a poor Ofsted judgement. And high stakes for those pupils who… too often see their school spinning into a cycle of decline”.

That’s why Ofsted remains government’s Rottweiler, frankly as welcome a guest in school as a plague of boils. I’ve met Amanda Spielman, and debated with Sean Harford on a conference platform: both are humane, civilised and deeply committed to education. I’ve encountered impressively experienced HMIs, not to mention heads leading or forming inspection teams in the Independent Schools Inspectorate (ISI). How good it would have been simply to welcome such experts into school and use their wisdom and judgement to help my school improve.

That remains impossible when the complexity of what happens in school every day (effectively acknowledged by Sean Harford on behaviour) is ultimately reduced to a single Ofsted adjective, always with the possibility of professional or institutional damnation.

Despite the efforts of many good people in the system, inspection has long been causing more harm than good. Mary Bousted demands both “an inspection system that is fair, valid [and] reliable” and “a new deal for education professionals.”

I disagree - but only slightly. Tinkering with Ofsted won’t help. It needs to go.

Dr Bernard Trafford is a writer, educationalist, musician and former independent school headteacher. He tweets @bernardtrafford