More or Less on Radio 4 yesterday explained in some detail exactly how the algorithm worked. I strongly recommend anyone who didn't, to listen to it. My jaw hit the floor. Apparently the students results weren't moderated at all - their mock results and teacher predictions were totally ignored. Instead, it was the schools that were graded, based on past performance. Each school was given a range of results based on past performance, and these were handed out to the students based on teacher ranking of the pupils - the top-ranked pupil got the highest result allocated to that school, and the bottom-ranked pupil got the lowest grade allocated to that school. So if you were an absolutely brilliant pupil in a particular subject, but your school hadn't had a good result in that subject recently, you stood no chance of getting a good grade, no matter how well you had done in your mocks or what your teacher predicted.
Schools with small class sizes were excluded from the algorithm, and pupils given their teacher-predicted grades instead. So pupils at private schools, who usually have small class sizes, got higher results than pupils at state schools with large class sizes. You couldn't make it up!
The Times is reporting this morning that Williamson was warned back in July of the problem, but ignored it as he was more concerned about grade inflation. Even if he wasn't warned, I would have expected him to have asked some detailed questions. Either Williamson was misled, in which case he should sack those who misled him, or he knew, in which case he should fall on his own sword.