[EM] How choice of voting systems depend on amount of participants

[Kristofer Munsterhjelm]

On 10/07/2014 01:50 AM, Richard Fobes wrote:
> On 10/6/2014 4:10 PM, dikov dikov wrote:
>  > I though that Condorcet methods are better when multiple places are
>  > concerned. Then scoring every entry brings relevant ranking. On another
>  > hand if only one winner is considered than simple majority system is
>  > accurate enough.
>  > Is it correct, or I am wrong?
>
> I'm not sure what you mean by "if only one winner is considered," but
> I'd say that your interpretation is not yet correct.
>
> The biggest issue is the number of candidates.  That is true regardless
> of the number of winners.
>
> The only way that simple majority ("plurality") voting works is if there
> are only two candidates.  It also works if one of the candidates gets a
> majority (more than half) the votes, but most people mistakenly think
> that that means the winner can be whichever candidate gets the most
> votes.  (Most votes and more than half the votes are very different.)

I seem to recall someone mentioning a US region that used majority 
voting: there were rounds of voting until someone got an outright 
majority, and the rounds kept on for as long as needed. I don't recall 
the details, though.

> If there are three or more candidates -- which many of us believe should
> be the case in any governmental election -- then single-mark ballots
> (where only one candidate name can be marked) do NOT work.
>
> In these cases (most real elections), there are three kinds of ballots
> to choose from:
>
> *  Approval ballots, where more than one candidate name can be marked.
>
> *  Ranked ballots, also called 1-2-3 ballots, (and sometimes other names
> are used), which you recognize.
>
> *  Score ballots, where each candidate is given a score, similar to
> Amazon's use of 1 to 5 stars.
>
> Condorcet voting methods use ranked ballots.
>
> If approval ballots or score ballots are used, then whichever candidate
> gets the most "votes" (approvals or points) wins.

There's also cardinal pairwise, which tries to combine ratings and 
Condorcet, and median methods that use score or grade ballots.

Median methods are relatively insensitive to how the voters assign 
scores. In a way, they could be seen as ranked methods where voters may 
"skip ranks", and thus they're often imagined as using grades (e. g. A 
to F) instead of numerical scores. They're also intended to be more 
robust to strategy than simply summing up the scores is.