Does observer's account ever count as evidence?

[upupa-epops]

@sedgequeen

I disagree with your premise. Perhaps that’s because I come from a birding background, including being on a state committee that evaluated rare bird sightings long before good photos of birds were so common. Written notes are evidence. If a researcher wants to disregard them all on principle, he can, but others will be happy to see the record.

I don’t disagree here. My premise isn’t that notes can’t be reliable evidence; it’s that they typically haven’t been considered so on iNat, at least on their own. An eBird observation with no notes (equivalent to an iNat Casual obs) is equivalent on GBIF to a RG observation from iNat, and I have fairly high trust in eBird observations. It’s always been possible to fake notes, but as AI technology improves photos may well be in a similar category soon.

@spiphany

But here, too, the difference may be that a field sketch made on paper with information such as date/time/location recorded in the journal permanently documents the particular moment of the encounter in a way that digital notes that might have been made or edited at any time do not.

Yeah, and Tony has said recently that they only consider field sketches made at the time of observation to be valid media evidence. A field journal page with drawings and corresponding notes both made while observing the organism is pretty trustworthy.

@spiphany

If observer notes were to be disregarded completely, there would be all sorts of information that could no longer be used to help with ID: smell/taste, texture, DNA sequence results, information about host, links to previous observations of the same individual, etc.

Some of these are easier to fake than others. Usually the host species is visible enough in photos that you can at least tell if the observer is way off base for example. Similarly if there are better photos of the same organism, you may be able to tell that a tree is the same tree even if you can’t identify it from some photos.

I guess the question is how much margin of error is acceptable. If you are more trusting of easier-to-fabricate evidence then you’ll get more inaccurate data. But most examples will still be accurate regardless, or have negligible impact (e.g. in areas where both crow species are abundant).