Confidence and Uncertainty

[ecurrano]

During our community workshops, paleobotanists brought up the importance of considering uncertainty in data and interpretations. At our Senior Personnel Meeting in June 2022, we brought up the idea of adding a confidence property to certain nodes. Here are my thoughts on which nodes should have confidence properties. Additionally, we might want to distinguish why there is uncertainty.

Nodes for which we need a confidence property:

1. Age
2. Latitude & Longitude
3. Character state described for a particular specimen or OTU
4. Taxonomic assignment of an OTU
5. OTU (i.e., how confident are you that this is a real OTU? Are there enough diagnostic characters?)

Uncertainty scale:

• High, medium, low
• Numeric scale from 1-5
• Note: We do not want people to have to mark this for every character state, so maybe you only mark this if there is uncertainty associated with it.

It is also important to note that there are different reasons for uncertainty, and it may be important to distinguish among these. These include:

• Quality of Preservation - applies to 3-5 above
• Particular Features preserved - applies to 4 and 5 above. We need a better way to say this, but what I'm trying to get across here is that in some cases, you will have a diagnostic feature preserved that explicitly links to a particular taxonomic assignment. Or you have a taxonomically important reproductive structure, as opposed to a not particularly informative leaf.
• Variability within a specimen- applies to 3
• Variability within an OTU - applies to 3 & 5
• User experience level - applies to all the above.

[ecurrano]

This also brings up a discussion of how things are synonymized and how users can note their confidence in synonymizing two OTUs or linking isolated organs into a whole-plant reconstruction. This comment may be more relevant to another issue, but writing it here to get it down.

[aazaff]

@ecurrano It seems that this was dropped quite some time ago. paleo team please revisit this because the window to add anything more about this is passing (if not having passed already)

[ecurrano]

@aazaff : OK. We will revisit this week. Age, GPS info, and OTU have fields to capture this information, so we can cross those three off. But, the other two are still potentially important and were requested by the community.

[aazaff]

@ecurrano Look at this issue and figure out specifically what we want to happen and by when.

[ecurrano]

We have adequately handled: Age, Latitude & Longitude, and OTU uncertainty through the use of fields.

I think that character states are also somewhat handled through the allowance of multiple character states and ranking. People can add notes about the uncertainty in notes and comments.

The community might like some additional features for handling uncertainty in the taxonomic assignment of an OTU. We should get feedback on this at the Hackathon.

[ecurrano]

Further discussion of this issue may need to wait until v 2.0. I do still want us to do more about uncertainty in character instances and taxonomic assignment.

[aazaff]

I concur that this is not only V2.0, but is also something that would have to be done in concert with PBDB.