# [EM] I think Bishop's deconstruction algorithm fails

Dave Ketchum davek at clarityconnect.com
Sun Nov 27 20:48:10 PST 2005

```Who is claiming Condorcet does not start with ballots?

There has been discussion recently that seemed to start with an array,
trying to reconstruct a possible set of ballots.
If the array was truly part of counting Condorcet votes, then one or
more possible sets of ballots are findable, though the effort could be
painful and I fail to understand why bother - can even get here with an
array constructed some other way that happened to conform to Condorcet
expectations.
If there are no possible collections of Condorcet ballots that could
have produced the array, then it had to have been produced some other way.

Below Rob talks of stacking up the actual ballots.  Another effort that
seems not worth the pain - think how tall a stack you get if we elect a
governor here in NY and have FIVE MILLION ballots.

What we do have with Condorcet is the arrays - likely one per precinct -
summed for the complete district, and likely summed for subdistricts such
as towns, cities, or counties.  These are manageable amounts of data that
can be made available to whoever may be interested.

DWK

On Mon, 28 Nov 2005 02:45:23 +0000 (UTC) Rob Brown wrote:

> Paul Kislanko <kislanko <at> airmail.net> writes:
>
>>This is my philosophical objection to vote-counting methods that use the
>>pairwise-matrix as input. You cannot map the pairwise matrix to the voters'
>>ballots unambiguously, so any such method is by definition "not
>>transparent".
>>
>>I believe any valid Condorcet-method should be able to be defined without
>>reference to the pairwise matrix. Begin with the array of ballots = voters x
>>alternatives, and work from there. If you don't begin with ballots, then
>>it's not an election-method.
>>
>
> As I mentioned before, condorcet methods *do* begin with ballots.  The pairwise
> matrix is nothing but an intermediate stage between input (all the ballots, lots
> of data) and final results (declaration of winner, tiny amount of data).
>
> Anything that processes lots of data into a little data is going to lose some
> data along the way (pardon me for stating the obvious).  Condorcet methods just
> happen to have an intermediate stage that is viewable and interesting.  I fail
> to see how that is a bad thing.
>
> BTW, you say such methods are not transparent, but you seem to be assuming that
> the pairwise matrix is all that the public will have access to.  Why can't the
> ballots be made accessible as well, just like any of the methods you prefer?
>
> -rob

--
davek at clarityconnect.com    people.clarityconnect.com/webpages3/davek
Dave Ketchum   108 Halstead Ave, Owego, NY  13827-1708   607-687-5026
Do to no one what you would not want done to you.
If you want peace, work for justice.

```