"The fact that so many have let children down in the past and far too many are still doing so really makes me question if it is possible to make all work."
To sow that a comp - as in the school, by itself, simply by being comprehensive - lets children down, then you have to do a controlled match in which you compare schools with the same intake, in communities with the same socio-economic intake, but of different types e.g. comp, secondary modern, grammar.
In the times I have been on this thread, when people have posted details of 'dire' comprehensives, they have been schools in communities 9either immediately around the school or wider) with high unemployment or collapse of traditional employment (post-industrial or seaside); above the national average FSM; about to close / have already closed; in areas with surplus school places where other schools (religious, selective, private) have siphoned off children with engaged parents; or actually been secondary moderns but 'called' comps.
Of course there are comprehensives that don't do as well as they should with more average intakes. There are definitely grammars that do less well than they should with their highly able intakes (statistically, given its level of selectivity, any grade less than an A, or even in some cases an A* ,from a superselective probably represents less than expected progress).
However, in saying that comprehensive have 'failed', we do need to consider whether a school of a different type, given the same intake - or the average of the 2 schools that split the intake between them (weighted to represent the proportion of the pupils each take), if in a selective system - would produce better results. The information we have from comparing existing selective vs non-selective areas is that the cohort results for comprehensives and selective systems are pretty much the same, with a small gain at the top end in selective areas being offset by a larger loss for the less able.