39
5. Voters’ judgments will be affected by the stability of different AV outcomes.
Thus, a Condorcet loser may not be a stable AV outcome—much less a strongly stable
one—so this candidate’s viability is less than were he or she a stable outcome.
6. When there is a Condorcet winner, this candidate is always a strongly stable
AV outcome. Thereby AV preserves the majority will, at least if there is some kind of
strategic coordination among voters, through sincere voting.
7. In large-scale elections, this coordination is possible to a limited extent from
information provided by polls (Brams and Fishburn, 1983, ch. 7). But strategizing by
voters may not be perfect, allowing Condorcet winners, even under AV, to be defeated on
occasion.
8. This failure may sometimes be salutary, especially when a BC winner differs
from a Condorcet winner and “majority tyranny” is a concern (Baharad and Nitzan,
2002). In such a situation, the BC winner may be a more acceptable candidate, even if he
or she is not, like a Condorcet winner, a strongly stable AV outcome.
9. While the stability and strong stability of outcomes facilitates their selection,
even unstable AV outcomes should be considered acceptable, especially if there are no
stable outcomes because of a Condorcet paradox (Miller, 1983).
10. Speaking normatively, AV provides a better way of finding consensus choices
than do other voting systems because of the information that it both suppresses
(preference rankings) and expands upon (who is acceptable and who is not in the
rankings):
Convention