The move by the exam regulator, Ofqual, to freeze the proportion of particular grades awarded in A-levels and GCSEs looks like a retrograde step. In broad outline, it returns the grading system to something more like the one that existed until 1987. It will also be condemned by many interested parties. They include school-leavers who need their grades for university, the teachers who prepared them, and the schools trying to improve their place in the league tables. The question they will all ask is why hard-working pupils could receive a lower grade for attaining the same numerical score than they would have received, had they taken the exam in previous years.

There is undoubtedly an element of unfairness here. And it would have been better if Ofqual had announced such a substantial change in the principles of grading well in advance. But it is a one-off unfairness that was badly needed to remedy one of the most malign distortions of recent years. And it will cut both ways: a lower score might also achieve a higher grade. When the proportion to be allocated to each grade is fixed, the actual score required depends to an extent on the performance of others. If it is a strong year, or the exam is marginally easier, a higher score will be needed for an A – and vice versa.

The change is designed to end the increasingly bitter argument about so-called grade inflation and whether exams are becoming easier. Schools tended to argue that more good grades reflected harder work and better teaching, but this view was not shared by most universities or employers, or some public school teachers, who insisted that an A grade was no longer what it once was.

The downside, of course, is that real, across the board, improvement may be harder to track. But if it is a choice, as it is, between two evils, the preference should go to the one that better preserves the credibility of the exam system. It may not be a popular choice, but it is one that had to be made.