All major assets like locks and bridges are monitored, and an attempt made to classify them in order not only of condition, but also rate of deterioration. I have been to presentations by CRT where they make it clear that what gets priority is not determined solely by what is in worst condition, though.
If something is being monitored, is in poor state, but judged to not be dangerous or in danger of imminent failure, it can lose out on the priority stakes, either to something in less poor state, but where safety is at issue, or something in less poor state, but where the implications of not attending to it promptly may massively increase the final bill. In other words, sometimes a concious decision is made not to give something priority, because although it is known not to be "100%" it is judged it will remain OK for a while, and the ultimate remedial work will not be significantly higher if repairs are deferred, but the situation monitored. Vince Moran pulls no punches that they are constantly having to juggle to make best use of available resources, and that it is inevitable sometimes that they will get it wrong. Loads of Internet pundits and NBW contributors seem to know after the event that there was a problem brewing at a location where a failure has occurred, but I'm sure people also know of worrying signs at high numbers of locations where no failure has occurred, and none ever will. Does anybody think they can accurately predict (say) the next three major failures? No, of course none of us can if we are honest, and that is the dilemma for CRT - some of the issues at these structures can probably only be found by pulling them apart.
As I have said in response to the Aylesbury Arm failure, it has now been discovered that the failed lock is far less substantially built than others on the arm, but nobody knows why. It had not stopped it standing for 200 years, but that bit of information may well be a contributory fact in its failure. I'm not enough of an engineer to know if you can start a program that non destructively tests every lock to see if it actually contains less brick or stonework than people have hitherto assumed it did, but I kind of assume such a program might be far from trivial?