[EM] Ballot formats (Re: new simple legal strategy to get IRV)

[Juho Laatu]

> On 09 Oct 2015, at 13:28, Kristofer Munsterhjelm <km_elmet at t-online.de> wrote:
> 
> On 10/09/2015 10:00 AM, Juho Laatu wrote:
>> I just note that there can be also simple ballots like in
>> https://en.wikipedia.org/wiki/Open_list (see the picture). If one
>> wants to expand that to ranked ballots with high number of
>> candidates, one could simply allow the voter to write multiple
>> numbers in the ballot instead of only one. One could thus cast a
>> ranked vote by writing few numbers, e.g. "23 74 74 5 234 321".
>> 
>> The ballot could be as small as in the picture and still allow
>> ranking of e.g. 5 candidates.
> 
> Schulze also uses a similar ballot format in his proposal for STV-MMP,
> e.g. http://m-schulze.9mail.de/schulze5.pdf page 6. The open list
> article you linked to shows that it's possible to read number input
> since e.g. the Finnish ballot has the voter write down the candidate
> numbers of those he votes for, so having the voter fill out numbered
> rankings shouldn't be a problem.

In order to avoid any misunderstandings I note that currently one should write the number of exactly one candidate in the ballot in Finland. One could make especially party internal proportionality better in such open list system by allowing ranked votes. One could achieve good party internal proportionality already with a very simple ballots that would allow ranking only e.g three candidates.

> 
> That gives two possible ways of setting up the ballot, now that I think
> about it. The first would be like Schulze's, where the voter writes down
> an 1 next to first place, 2 next to second, etc. The second would be a
> table which says (using ASCII art here)

I used the candidate number option to demonstrate that very simple ballots may be sufficient.

> 
> 
> Ranking              Candidate number
>                      to be filled in
> +------------------+ +---------------+
> | 1st rank (best)  | |               |
> +------------------+ +---------------+
> | 2nd rank         | |               |
> +------------------+ +---------------+
> | 3rd rank         | |               |
> =======================================
> =========(more rows come here)=========
> =======================================
> | 20th rank        | |               |
> +------------------+ +---------------+
> 
> where one assumes very few voters will rank beyond 20th rank. This could
> be 40th or 80th depending on how many rows fit on the paper.

In an election with say 200 candidates it may be too tedious to rank all the candidates. One party could have tens of candidates, and already that could be too much to the voters. If one tries to improve the existing Finnish system, one could take the approach that votes will by default support one of the parties. It could be natural that ranking would be allowed only within that one party. The voter would thus vote e.g. 123 127 125, where all those candidates belong to the same party. This would mean that this party gets one vote, and within that party the voter may influence which candidates will be elected more efficiently than with a bullet vote. This means something like "STV within each party". From the voter point of view it is important that already a simple vote (bullet vote or three ranked candidates) automatically supports all the candidates of that party (the party inherits the full vote in case none of those three will be elected).

(An alternative would be use of party numbers. Vote "123 127 125 100" would go to party "100" if none of the three candidates (123, 127, 125) will not be elected. This approach would be however quite complex and prone to mistakes when voting. That's why the "STV within each party" approach could work better (in an election with very high number of candidates).)

> 
> The dual approach (where candidate numbers are filled in) could be
> better in the sense that the candidate numbers could have check digits
> or be augmented by some other error-detecting code, and they could also
> avoid ambiguous digits like 1 and 7.

All elections where voters write the numbers (or names) in the ballot have this problem. Some voters will write the numbers in a hurry, and some votes must can not be read because of this. In Finland the governments tries to remind people to write the numbers carefully, and gives advice on what kind of numbers people should use to avoid any problems. I believe the number of votes lost due to this problem is however quite low, and therefore this problem is not serious.

> If the ballots are read by machine,
> the machines could flag ballots they can't read for manual inspection.
> On the other hand, having to look up candidate numbers is another step
> of indirection for the voter and could make voting more of a hassle,
> thus reducing turnout.

In FInland I guess people typically remember the number of their favourite candidate. In addition list of candidates and their numbers is available in the voting booth. Writing more than one number would mean that people generally do not remember all the numbers but would need to check from the candidate list in the voting booth. No big problem, but this would make the voting process somewhat slower. Nowadays people use maybe 10 seconds to vote (inside the booth). Maybe 20 seconds could be more typical for ranking three candidates. Voters would be allowed to write only one number, and that would make voting as fast as it is now. Many voters might do so.

This approach would delay also the vote counting process. Instead of collecting the ballots (with one number each) into piles, one would have to record the few numbers of each ballot into a computer.

> 
> For equal rank, it'd be possible to expand the table like this:
> 
> Ranking             Candidate numbers for this rank
>                           to be filled in
> +-----------------+ +----------------+----------------+----------------+
> | 1st rank (best) | |                |                |                |
> +-----------------+ +----------------+----------------+----------------+
> | 2nd rank        | |                |                |                |
> +-----------------+ +----------------+----------------+----------------+

Yes, but I tend to think that it is not very important to allow equal ranking. People may toss a coin if they can not otherwise decide which one of the candidates is better. Allowing equal ranking would be good since it makes it possible to cast such a vote when the voter feels like having two or more equally good candidates. But on the other hand the complexity of the ballot and voting process might increase.

> 
> .. etc., the idea being that the voter would fill in every candidate
> he'd rank first in the cells of the first row. It is less intuitive,
> though. People who are serious about accessibility might want to have
> "expensive pencil" type single-purpose (not Turing-complete) voting
> machines for printing to this format.
> 
> MJ or discrete rated systems could have a multi-page ballot where each
> page corresponds to a different grade or rating, and the voter writes
> down the candidate IDs for each page. For approval, there's just one
> page. Continuous rated systems would pretty much have to use something
> like Schulze's format, but instead of filling in numbers, would have
> something to the effect of:
> 
> "Place a mark on the line corresponding to the rating you want to give.
> Further to the right is better."
> 
> Candidate 1   worst |-------------------------------------------| best
> 
> Candidate 2   worst |-------------------------------------------| best
> 
> Candidate 3   worst |-------------------------------------------| best
> 
> Candidate 4   worst |-------------------------------------------| best
> ...
> Candidate n   worst |-------------------------------------------| best

I note that this format (and also a format where people are allowed to write the rating of the candidate next to the name) could be used also by ranked methods. Some people might like the idea of quickly drawing a mark at some point on the line (see your ballot format image above) without thinking too much about the exact ordering of the candidates. My point here is that this format is good in the sense that it would allow raking of maybe even 100 candidates in less than 30 seconds.

Juho


> ----
> Election-Methods mailing list - see http://electorama.com/em for list info