Posts by user "arewenotmen" [Posts: 7 Total up-votes: 0 Pages: 1]

Doesn't work like that, and significant logical errors have been made on that basis in the past. The chances of simultaneous independent engine failure might be a trillion to one, being a million million.
​​​​​
But failure is often not independent. Hit a flock of birds with both engines, run out of good fuel, etc etc - then the probability is primarily that of the root event.
​​​​
Edit: I make no comment on whether they were running or not in this case, only the statistics

​​​You might be technically correct here but this line of thinking doesn't fit with the rest of the events. If they weren't physically and observably set to CUTOFF, setting aside the spoken comment, then how were they subsequently reverted to RUN? You would be saying that some invisible systems fault produced a temporary 10 second condition, starting to apply to both switches 1 second apart and ceasing to apply 4 seconds apart. Even within a very low probability scenario, it doesn't seem plausible - this was all physical, for reasons unknown.

Speculation is naturally going in a particular direction towards deliberate action, but it gives me some pause as to whether this actually makes sense as a pre-meditated scheme, and I don't know either way.

Now, perhaps this is a question that shouldn't actually be answered for various good reasons, so more a prompt to consider, but if you were going to set about doing this intentionally, is the probability of the desired outcome actually high enough to commit to this specific course of action, out of all the possibilities available? Perhaps not in this specific environment, but modern survivability in relatively low energy accidents is remarkably high.

In this incident, prior to the release of the report, we have previously remarked that the crew seemed to have done the best they could with it and got unlucky with the circumstances - the buildings, essentially. For me that also raises a question of how carefully planned & rehearsed such a plan would have to be - the breadth of timing parameters where this would 'work' or not.​​​
​​​​

Well, not really, because separated wiring alone doesn't require independent events to produce a common outcome. I mean, pour a drink into the panel, or something more exotic like panel-localised EM interference (I have absolutely no idea how realistic that is, just illustrative).

For other reasons, I think it very unlikely that the switches were anything other than physically moved, so this is kind of pedantry rather than useful. But the probability analyses that folk (including experts) come up with often loses sight of the above. I posted much the same in one of the earlier threads about a month ago when all we knew was twin engine failure.

​​​​​

I don't think this is a good assumption.

I think you should assume data records either have millisecond or better accuracy, or have no item-level timestamps and time is determined from the stream in which they're contained.

Sampling certainly introduces ambiguity, e.g. for the purposes of illustration:

t=0.0, SAMPLE, no switch active
t=0.1, L switch activated
t=1.0, SAMPLE, L switch active
t=1.9, R switch activated
t=2.0, SAMPLE, L & R switches active

Sampled time between activations = 1.0, actual time 1.8

t=0.0, SAMPLE, no switch active
t=0.9, L switch activated
t=1.0, SAMPLE, L switch active
t=1.1, R switch activated
t=2.0, SAMPLE, L & R switches active

Sampled time between activations = 1.0, actual time 0.2

If you then introduce sampling of L & R at independent times, and skew one to induce the maximum delay, you still can't produce a more misleading representation than the above.

You are talking about some further weakening of accuracy beyond the sampling rate, either by the data recorder recording rounded timestamps or in post-retrieval processing/reporting by the AAIB, and I don't think either of those things is justifiable.

I fully expect that the investigators will have reported the times as the separation observed from the sampling, thus \xb11 sec accuracy (for this particular input). There doesn't need to be any further inaccuracy. We are free to turn a quoted four seconds into three or five, but not two or six.

Edit: sorry, I see what you mean now. I was misunderstanding, because their reporting on the switches to CUTOFF was a duration - 'time gap of 01 sec' - whereas restoration to RUN is described in real timestamps, e.g. '08:08:56 UTC'. I'd sort of forgotten the latter and assumed four seconds was quoted. Mea culpa.

Mrshed - yep, you are right - see my edit.