Exactly! There's got to be an underlying selective rationale behind the
loss of flight, irrespective of how few (or how many) anatomical changes it
involves.
The thing to do would be to find a
clades where reversals *do* happen, and then determine if the
reversals tend to happen more often closer to the root.
We can use the Theropoda as a test case. But here we're back to square one.
At face value, the earliest examples of secondary flightlessness are
_Patagopteryx_ (cursors) and hesperornithiforms (divers). But advocates of
the 'neoflightless' hypothesis argue that we are missing the reversals that
occur at or near the root of the Avialae because the body-plans of the first
'neoflightless' taxa are so similar to those of the primitively flightless
taxa that lie just a few nodes down.
Furthermore, because the flight-related characters are incipient or nascent
in expression and function in early fliers they can be converted back to
their ancestral state at little expense, genetically or ecomorphologically,
to the animal. Thus, the fact that we are overlooking 'neoflightlessness'
is the fault of cladistic methodology, which is accused of being insensitive
to these one-step-forward/one-step-back reversals. If this is true, and
reversals do indeed "tend to happen more often closer to the root", we would
miss these reversals because current cladistic methods cannot pick up the
'signal'. I'm not saying I buy any of this, but it is something to keep in
mind.
This same problem (if it is a problem at all) may also occur in the lineage
that led to tetrapods - did certain branches make a U-turn and go back to
using the limbs solely for swimming, and are we misinterpreting these forms
as more primitive taxa? I don't think we are; but the same question lies
behind the 'neoflightless' isuue.