[EM] election-methods Digest, Vol 37, Issue 8

[Forest W Simmons]

In response to Juho's questions:

There are various possible versions of DYN.  But if the candidates were 
required to give their approvals before learning the partial results, 
then there would be little point in delegating votes; the voters could 
just copy the candidate recommendations onto their own ballots.

The partial results inform the candidate strategies.

1. True, non-strategical zero info approval voting would give results 
superior to strategical voting results, i.e. if everyone just approved 
all of the candidates that they thought would be satisfactory, based on 
their knowledge of the candidate qualifications, as opposed to their 
knowledge of poll results.

2. Next best is informed strategy based on reliable polls (in addition 
to the knowledge of the candidate qualifications).

3. Worst of all is strategy based on disinformation.

It seems to me that we can expect the worst (3) from our corporate 
media, so we have to build something into the system to counteract the 
misleading disinformation as to the levels of candidate support.

That is the purpose of DYN, i.e. to take us from (3) up to (2), which 
is the best we can do, since (1) is generally impossible.

Remark.  Condition (2) will almost surely result in the election of the 
Condorcet winner when there is one.  However, when condition (1) and 
condition (2) lead to different winners, then the condition (1) winner 
is the more democratic, as the following example shows:

(sincere support)
55 A>B>>C
45 C>B>>C

Under condition (1) B wins with 100 percent approval.

Under condition (2) the A supporters strategically drop their support 
for B, so the Condorcet winner A is elected with 55 percent approval.

Forest









Juho wrote ...

>
>Why "after"? (Is this somehow essential? Will they change their  
>opinions based on the "partial results"? Are they supposed to reflect  
>the general opinion more than their own opinion?)
>
>How about announcing the content of the proxy votes already before  
>the votes are counted, or maybe already before the election?
>
>If the votes are counted "after", will each proxy know the number of  
>delegated votes that other proxies have (or the number of his/her own  
>delegated votes) before they cast their proxy votes?
>
>Juho
>
>
>