I agree with both your opinion (that almost all users in this situation intend to make a “leading disagreement” not a “branch disagreement”) and your surprise that this bug hasn’t been fixed.
When @loarie posted his blog post in 2019, clearly there was an intention to fix this issue. TBH, without his great illustrations I don’t know that I would be able to get my head around the distinction between these styles of ancestor disagreement.
His proposed remedy was to allow identifiers to explicitly choose one or other type of ancestor disagreement. While that seems fine in principle, there was some debate about how to phrase the three choices so that they would be clear to identifiers and I’m guessing that confusion may have stalled the plan to fix the underlying issue.
My feeling is that there are very few cases when identifiers want to say “By adding the ID ‘Insect’ I disagree that this is a Seven-spotted Lady Beetle or any other taxon on the branch that leads from Insect to Seven-spotted Lady Beetle”. There’s a pretty simple reason for this. If I’m knowledgeable enough to be able to confidently exclude all the taxa on the branch between Insect and Seven-spotted Lady Beetle, I’m almost certain to be able to select some order or family that’s a better fit. Conversely, if I can’t be confident of any ID below Insecta, how come I’m so sure that it’s not anything within Pterygota?
The whole “branch disagreement” thing seems like such an edge case that it doesn’t merit any special functionality. Can’t iNat just switch to treat these ancestor disagreements as leading disagreements? I can see there might be a case for genus- and family-level IDs where the identifier believes that the observation cannot be identified more precisely. However, for that purpose we do already have this DQA flag: “Based on the evidence, can the Community Taxon still be confirmed or improved? No, it’s as good as it can be.”
Another possible issue is that iNat staff are reluctant to re-interpret previous branch disagreement IDs. Maybe someone who gave an “Insecta” disagreement in the past really didn’t think any ID on the path through Pterygota was plausible. I think that’s unlikely, but I do accept that’s the logic of how iNat has always treated an ancestor disagreement and perhaps we shouldn’t disregard that. If so, maybe new disagreeing IDs can be treated as “leading disagreements” and old ones as “branch disagreements” (while they still exist).
Whatever the cause, I feel that the lack of resolution of this issue has a definite degrading impact on the efficiency of the identification process. A small, but finite percentage of iNat observations spend a long time stuck in ID limbo because identifiers often have no choice but to add a high level “branch disagreement” ID that they can’t reliably find and fix later.