"I chose Th because unlike what someone else said, its intake isn't grammar school. GS are academically selective, so if you have a less able DC you cannot get them in; whereas I can and did exercise what 'power' I do have in buying into the catchment of a good comp, one where they seem to do well by all the DC. Had DS2 been more academic, I would have taken the risk of buying into the catchments of the Salisbury grammar/s."
The fact is however that there are almost no 'less able DC' at Thornden.
That's the whole point of these schools - people think they are turning shit into sugar, but actually if your children are genuinely less able, which is statistically very unlikely given that 95% of Thornden's intake then they probably will not get 5 good GCSEs at Thornden - 4 out of 14 did last year. It's only by excluding nearly all of the least able that the school looks good.
Now 4 out of 14 (29%) is actually a pretty good result, compared with the national picture, but given the tiny number of less able children at the school the true chance of success is going to have a wide range, because the school's admissions policies excludes most less able children; however if we had more data then it would be a pretty good result if the school actually did become more comprehensive, if that 29% figure is accurate over a larger sample.
Statistically from what you say, your son is probably a 'middle attainer', and not really 'less able' at all, by national standards. Under the grammar school system in Kent, there are lots of 'middle attainers' going to grammar school. Dover Grammar gets 32% middle attainers, Thornden gets 42%.
The grammar schools in Salisbury are super-selective, and a different kettle of fish, but just because Thornden is selecting 'average and above', rather than 'above average', doesn't mean it's not selective. It's selective because its intake does not look anything like the intake of other local schools.
That it selects on income/house prices isn't really the point - selecting by wealth is incredibly effective, and is across the globe, as pointed out in the recent UN study of educational outcomes globally.
Thornden does appear to be very good, from its results across the spectrum, but it would be much better if it were to take a much larger proportion of less able pupils, and do with them what it appears to do (but we can't be completely sure, statistically) with its current tiny cohort of the less able.
That's far more socially useful (which is the purpose of education, ultimately) than creating a middle class ghetto. A fair banding system would serve the needs of the community as a whole much better than the current system which purely serves the needs of middle class parents.