[EM] eliminate the plurality loser until there is a Condorcet winner

[Kevin Venzke]

Hi James,

I don't have the answer to your question but reviewing the rangevoting
page and Woodall's papers I note that Woodall defined two combination
methods.

In the 1997 paper, "C-AV" "exclude[s] all candidates not in the Condorcet
top tier (closing up the gaps in the preference listings when candidates
are excluded from them), and appl[ies] [AV] to the remaining candidates."
I call this method "Smith//IRV" myself.

In the 2003 draft, "CNTT,AV" "appl[ies] AV, stopping when all but one of
the candidates in [the CNTT] have been excluded, at which point the
remaining candidate is declared elected." There are suggestions for the
case that no one in the CNTT has first preferences.

Curiously the properties of these methods are listed as being the same.
I think the definitions may have been changed more due to the FPP methods.
The 1997 paper seems to me to claim incorrectly that C-FPP is monotonic,
but only the CNTT,FPP method is. The 2003 paper seems very conscious of
this, though it doesn't refer to "C-FPP" or the earlier paper here.

The notion of repeatedly eliminating the Plurality loser (and 
then recalculating Plurality scores) until there is a CW seems familiar 
but I don't know who first came up with it or what it should be called...

Kevin Venzke