[hysan]

IMO, the spirit of the idea is to put higher information density fields first, and let that smooth out the UX for the remaining fields as you go downwards. Yes, there will be exceptions but that only matters if you’re trying to absolve the user of all work for 100% of situations. Trying to do that is a fools errand. Invert the order and use the information gathered to make inputting the rest simpler for 80% of the users. Then make it easy for the other 20% to course correct (ex: don’t disable autofilled fields, highlight all text when tabbing to the next field, etc). I think this pattern is a good one to keep in mind, but not blindly follow, when designing the UX of a UI.