Probably true, but that approach has two downsides:
-
if you’re dealing with thousands of observations, it would probably be impractical unless one had a tool that brings up the image from the iNat observation, and then records the researchers annotation input into the datasheet. In theory, one could build this, and I’ve contemplated doing something along these lines. For now, I’m using the iNat ecosystem because it’s handy.
-
those private annotations are not available for anyone else to use. anyone else doing similar work would have to repeat the annotation process. But as you say, if they want it done right, they’ll probably want to do that anyway, unless they know and trust whoever has already annotated the observations.
My current approach is to augment the existing annotations with observation fields. Even where an annotation already exists to hold a particular piece of information about an observation, there can be several problems:
a) as noted many times in this thread, there often aren’t the required options. I’ll use a “twin” observation field to hold additional values that may not be available for the annotation. My software that harvests observation data merges the annotation and its twin observation field. eg. when both sexes are present in an observation, I add the “sex” field with value “mixed”
b) because you cannot correct an annotation that was applied by someone else, I use the twin field to contain my corrected value. My software always gives precedence to the twin field over the annotation. Therefore, if someone enters “adult” as the Life Stage annotation on what is clearly a larva, I add the Insect Life Stage field with value “Larva”, and my harvested data will contain the correct value for Life stage. The observer may eventually correct the annotation, but my data collection isn’t held hostage.
Doing all of this can be a lot of work, especially when you’re late to the game and decide you should have been applying a particular annotation/field for years. The metadata tool is especially helpful when you are trying to play catchup on annotating thousands of observations. Some tips on using the tool appear in this other thread: