[EM] No. Condorcet and Hare do not share the same problem with computational complexity and process transparency.

[robert bristow-johnson]

> On 03/18/2024 9:51 AM EDT Toby Pereira <tdp201b at yahoo.co.uk> wrote:
> 
> 
> I don't think Approval strategy is overly burdensome. As an honest voter,

This is the same (bad) assumption that Borda made two centuries ago: "My system is only intended for honest men."

Bad assumption.  We need the system to be more bullet-proof than that.  Because we're not talking about impartial judges, these are *partisan* voters that want to promote their political interests by voting.  They want their favorite candidate elected.  At the same time, they want to prevent their least favorite candidate from getting elected.

When we test systems, we must use the "acid test".  We must test them hard, with worst-case scenarios.  To assume "honest voters" is throwing a soft ball at the system.  We gotta throw hard balls and see how well the system behaves.

> ... you just pick your least favourite candidate to approve, so one decision. 

Toby, do you mean "just pick your least favourite candidate to DISapprove, so one decision"?

A common syntactical mistake.  Like the "Wicked Bible".  Because if you meant "just pick your favourite candidate to approve, so one decision", that's of course, no different than FPTP.

> Approval has advantages outside the usual criterion compliance as well. The results of an approval election are easily published and digested, as with a FPTP election. It's simply a list of votes (approvals) for each candidate. I think this is worth more than it gets credit for. In a Condorcet election, how would the result be published? Every head-to-head separately?

Yes.  If C is the number of candidates (including Combined Write-In), then the number of pairs of candidates is C(C-1)/2.  So C(C-1) tallies.  If the method is Condorcet-Plurality (or for the legitimate interest of election observers), if we include the tallies of 1st-choice votes, then it's simply C² tallies.  5 candidates, that's 25 tallies.  10 or 12 centimeters of paper tape.  Quite doable.  For IRV, to make it summable it's 205 tallies, that would take about 3 meters of paper tape to print out.  Not doable.

> Also in a debate versus FPTP, approval is the only method that has no real disadvantages relative to FPTP. FPTP proponents can always fall back on saying their method is simple, much simpler than the proposed alternatives. Well, approval is essentially the same level of simplicity. FPTP passes the rarely passed participation criterion, which approval also does. FPTP proponents don't really have anywhere to go, except to make the spurious claim that approval violates one person, one vote. But this is easily countered.
> 

No it's not.  One reason I began my paper with an old (1911) ruling from the North Dakota Supreme Court striking down Bucklin Voting is because one of the concurring opinions had an important simple insight about government elections: We're counting people.  Not marks on a ballot.  The marks must simply correspond to people.

Just because the same rules apply to each voter equally does *not* mean that it satisfies One-Person-One-Vote, to count people equally.  If, using whatever method you choose or scrutinize, if at the end of the day, M voters preferred Candidate A while N voters preferred Candidate B and Candidate B is elected even though M>N, then those fewer voters for B had cast votes that were somehow more effective than the greater number of voters that preferred A.  A method that does not make that simple body count is not guaranteed to satisfy One-Person-One-Vote.

(And, of course, when there is a cycle, **no** method can satisfy it.  But that does not excuse a method for failing to elect the Condorcet Winner when such exists.)

I am not a FPTP proponent.  But this is not a spurious claim.

If I read you correctly above, then that voter has not been able to express a vote that differentiates their favorite candidate from their 2nd-choice candidate.  Now for some voters, they might like both and they don't care which of those two candidates gets elected (just as long as one of them does).

But I find it very hard to believe that such a voter is the most common.  Nearly every voter in every election that I had been aware of *did* actually have a favorite candidate.  And if the Approval election turned out to be competitive between their favorite candidate and their 2nd-choice these voters that Approved *both* candidates are going to regret that they Approved their 2nd-choice if that candidate ends up beating their favorite candidate.

Essentially, for the voter that Approved both A and B (but preferred A to B) and finds out that B wins in a close election with A, they're going to think that their vote didn't count.  And it didn't.

They *have* to be thinking about that as they wonder if they should fill in the oval on their 2nd choice.

But, of course, if they *don't* Approve their 2nd choice and the race turns out to be between their 2nd choice and the candidate they hate, and the latter candidate wins, they're going to regret not Approving their 2nd choice.

But in both cases these are tactical considerations that the voter has to make.  With a ranked ballot we know what to do with our 2nd choice.  We mark them #2.

--

r b-j . _ . _ . _ . _ rbj at audioimagination.com

"Imagination is more important than knowledge."

.
.
.