Of course ...

My point is that the calculations made are wrong (make that 99.999...%). It is not a case of waiting for enough to die, it is simply that the formality of the prediction attempted cannot be done with the data we have (or could realistically have), assumptions notwithstanding. What we have is an association between certain kinds of event and the equipment involved. This is entirely enough to trigger questions about equipment design, its usage, diving practices, and the people who dive that kit, including their purposes, quality of training, health and so on.

Consider CO-poisoning from gas-fired water heaters. Whilst a probabilistic approach could be taken, it is entirely inappropriate to do so. The association of death (with a clearly identifiable proximate cause, in this case) with those devices led to queries about installation, safety mechanisms, education of the user, and so on. Changes occurred. If I understand it correctly, the burst of problems with the Meg is a parallel situation. Calculating risk there is pointless.

One preventable death is enough to trigger those questions, and the onus is on the producer, the trainer, and the user to honestly appraise - and fix - problems. There is no other threshold for action. Denial is counterproductive. However, the deletion (or at least the tagging) of deaths that had nothing material to do with the use of a rebreather in any way except that it was used for the dive is a legitimate filtering.

The facts of the list may need some emendation from time to time, but the motive for its existence does not change, nor can the responsibility of anyone concerned be measured against some risk value when the deaths recorded are not stochastic.

BWD