Like other education ministers before him, Ed Balls has been badly briefed about tests. Every teacher knows that it's necessary to test pupils. The problems start when they have the wrong kinds of tests, and the results are used for too many purposes – at least 18 in England according to one senior figure. When tests are used for high-stakes targets, things start to go badly wrong.

Take today's A-level results. Why are grades so much higher these days? Because teachers, under pressure, are getting better and better at squeezing every last mark out of every pupil regardless of whether the students really understand what they are writing about or how long the information is retained. Spoon-feeding is endemic. No wonder lecturers in higher education complain about a lack of independent learning skills and results which do not seem to be reflected in performance at university.

This year's SATs marking debacle has laid bare the case that the Association of Teachers and Lecturers, and others, have been making for years. This huge testing exercise is just not value for money. The results are insufficiently reliable at school level, and we could get the same national indicators of how well pupils and schools are performing with a small national sample, the way other countries do.

The same problems – of inflated results due to teaching to the test, which infect A-levels – also make SATs results unreliable. We know this because other research on the performance of English children, as well as the increasingly useful international testing, show a much lower rate of improvement than the SATs suggest.

But the response of national test enthusiasts has been to introduce yet more tests. The single level tests, now being piloted, risk making things worse as long as they remain connected to national reporting and target setting. The current plan is for twice-a-year testing slots, just ratcheting up the stress on staff, which is unfortunately but inevitably transmitted to most children.

The pity is that just a bit of tweaking of the proposals for single level tests would result in a much better system. The excellent aspirations for them voiced by Mr Balls could become a reality, if they really were under the control of teachers without the tyranny of targets. The idea of a bank of tests ,which could be used as and when to confirm the teacher's own assessment, could transform the atmosphere in our classrooms.

It would be better if politicians were a bit more open about the purposes of the tests. The damage is caused by using them as the most significant measure of accountability of schools, headteachers, and every classroom teacher. If tests were no longer used this way, there would be many other forms of accountability which could do the same job more effectively.

Quite apart from the very expensive tests, we have the very expensive inspection regime of Ofsted, the duties on local authorities to monitor quality and intervene when problems arise, and a host of advisers, supporters, and national agencies telling schools what to do. In total, this amounts to hugely excessive, overlapping accountability systems, so getting rid of nationally reported tests would scarcely leave schools rudderless.

And the damage that would be prevented? A generation of young people who know how to pass tests, but do not know how to learn, who think education is just about certificates, but lack intrinsic commitment to lifelong learning. And the capacity for lifelong learning is what will mark out successful societies in this new century.

The writer is general secretary of the Association of Teachers and Lecturers