It seems to me that the most likely mechanism for digit loss would be a
phalange at a time rather than simultaneous reduction of all phalanges to
nothing. What does the evidence say about this?
Shrinking of the entire digit and loss of phalanges often seems to happen at
the same time. And from what little we know about temnospondyls,
lepospondyls and lissamphibians, the 5th digit (of both hand and foot) can
disappear at once, though this might be due to the incomplete fossil record.
However, phalanx counts are remarkably constant throughout limbed
vertebrates. Phalanges don't appear to be easy to lose.