Helsinki Birder
Stealth Birder
Yes - deliberately basic as I felt a full explanation of Bayes' theorem would be overkill. That doesn't negate the point though.
With regards to a bird so distinctive that its ID becomes instantaneous, the key issue is whether ID is beyond any doubt, in which case, yes indeed, a priori probability is irrelevant, as the theorem shows. Often that isn't the case though.
With regards to whether a record should not be based wholly on what was seen by the observers, I disagree entirely for the very reasons I outlined. A lot more evidence should be needed to support claims of a yellow-eyed penguin in UK waters than a Sabine's gull. It's why there are "description species" in the first place.
If an odds based approach did eliminate most major records then in my opinion, eliminating them is the correct decision. I don't think it would though (and it certainly wouldn't if ID was beyond any doubt) - its just a question of making an appropriate judgement of the weight of evidence versus the likelihood of it occurring in the first place and to do this, I think it helps to understand how the numbers work.
Doubt is a frame of mind Ilya, and subjective to the personality of the doubter, which in and of itself is subjective to a vast array of factors in its psychological make up. Its a mirror reflected in a mirror and leads no where.
Empiricism means that you stick to evidence and evidence alone. Avoiding the subjective. Whether a verbal description, regardless of its detail is considered evidence is another matter. In reality the only evidence ever obtainable in such a scenario is visual images. And the percentage of rare seabirds which are photographed at all is very low, due to the nature and logistics associated with seawatching.
So how then do you judge a rare seabird record, without such evidence?
Well realistically in such a scenario, the detail of the description only matters to the reviewer if it contains features which are anomalous for the species being described, in which case the reviewer can safely say it was not what it is claimed as.
If the record is a perfect description for the species, it really does not matter, as either (A) it is being judged on numerical probability in your case. or (B) faith in what has been described was accurate/true/ not false/ not string.
In which case....why write a description at all, if the content of it only matters if its wrong?
Owen