Buried in the rules there are a number of numbers that don't make any physical sense. We've already seen one case in the 0.9 factor that ABS inserted in the sloshing force calculations. The idea here is that our design method is too conservative so we need to "calibrate" it to actual experience. In practice, this means that some other class has used a less conservative method, and so far nothing terrible seems to have happened. This is a horrible design philosophy, and an open invitation to commercial pressures, precisely what happened in the design cargo density situation. The change wasn't based on anything other than the other classes were lighter.
Here's a more egregious case. Class claims that it designs the ship on a 20 year basis, that is to the situation which has a probability of 0.5 of being encountered in twenty years. This in itself is a revealingly strange philosophy. It seems to imply that a ship should last only twenty years. But even if you accept this invitation to shoddy construction, do you want to design so that there is 50% chance that the ship won't make it to age 20? Suppose you knew that every airplane was going to be scrapped at age 20. Would you accept a 50% chance that it would break in two in flight before then? A 1% chance maybe, 5% if you really want to be imprudent, but not 50%.
In fact, the ships are not designed to even this dubious criterion. For example, when you go in the rules, both ABS and LR, they say "compute the design sloshing forces on the basis of a pitch that is 0.6 of the 20 year pitch". We've asked a number of ABS and LR people where this 0.6 came from and received either blank stares or a semi-circular argument to the effect that this is the way the rules have been and we haven't seen any real problems yet. Forgetting about the fact that we only recently got rid of the swash bulkheads on big tankers and most cargo tanks are not slack most of the time, so we have no real experience, the point is that the ship is not being designed on the basis of 20 year encounter, but something much less. The rules should make that clear.
Get rid of all calibration factors unless they are strongly supported by experimental evidence from carefully designed and publicly documented tests. And then make sure that these remaining calibration factors are very publicly documented, so owners who would prefer not to use them are alerted to their existence.