Successive votes for tie breaker

[DEMOREP1 at aol.com]

Assume four candidates are in a cycle. The head to head matrix produces (with
the T being total votes for on each row) ---

    A    B    C     D
A  --  AB  AC  AD  AT1
B  BA  --  BC  BD  BT1
C  CA  CB  --  CD  CT1
D  DA  DB  DC  --  DT1

The 2 letter votes (X left over Y top) are the actual votes-- 10, 24, or
whatever. The row totals are 35, 67 or whatever. Assume that the CT1 total is
the highest. 

Should C be first using the total votes-for tie breaker  ? 

(Note to Mr. Ossipoff- the highest total votes-for is the *criteria* and the
*property* and whatever else are your tests for a tie breaker)

Should C be dropped to determine any additional choices (such as elections
for multiple executive or judicial offices- 2 or more sheriffs or judges in a
given electoral area) (assuming that a cycle still exists among such
remaining candidates) ?
A  --  AB   AD  AT2
B  BA  --   BD  BT2
D  DA  DB  --   DT2
Assume that the BT2 total is highest.

A  --  AD  AT3
D  DA  --  DT3
Assume that the DT3 total is highest.

Final result--- C>B>D>A

The matrix has the solution to the candidates in a cycle situation. The
question is what part(s) of the matrix should be used for a tie breaker. 

Another semi-obvious tie breaker is the total votes for (row totals) minus
the total votes against (column totals). Either tie breaker obviously is a
simple solution which can be understood even by the media, judges and
politicians.

It also becomes critical whether tie votes between two candidates should be
permitted (and the value of any such tie votes in the matrix- namely, zero or
1/2). If tie votes are 1/2 in the matrix, then the row totals will add up to
the number of votes cast. If tie votes are zero in the matrix, then the row
totals will not add up to the number of votes cast. 

Note- A candidate cycle results from having a divided majority. Having a
votes-for tie breaker will encourage voting for additional choices (i.e.
discourage truncation voting strategies). That is, if a voter's first choice
is not the Condorcet winner, then a voter will be encouraged to vote for
his/her second, third, etc. choices.

Presumably the voters who like similar *good* candidates (more than similar
*evil* candidates) will vote for such *good* candidates with low ranking
numbers -- 1, 2, etc.

Whichever group of the voters has the most votes will presumably result in
the candidates of such group getting the most votes.