[E-voting] The best VVAT is good for

Catherine Ansbro cansbro at eircom.net
Thu Oct 27 03:01:28 IST 2005


I'd add that in most cases, elections are to elect /several/ winners in 
a given area.  (e.g. a "3-seater" constituency would elect 3 
representatives to our national parliament).  On the first vote count, 
only the "first preference" votes are counted (on all ballots, just who 
got all the #1 preferences).  Typically, there might be 10-15 
candidates!  So it's common that after the first count, there might be 
no candidate who has enough votes to be elected for sure (that's called 
the quota--based on a formula of the number of voters and the number of 
"seats" to be won).  In this case, the person with the least number of 
"first preference" votes gets dropped out.  All the ballots that had 
given that person a #1 are then re-examined, and are distributed 
according to the #2 preference indicated on them.  (One can rank ALL the 
possible candidates if one wants.  Some choose to only indicate their 
first preference, or they might rank 3 or 5 or however many they want.)

This is all done at a vote counting centre in each county.  The ballots 
of the candidate who just got knocked out are literally put onto 
different piles according to the #2 preference.  (And if there is no 2nd 
preference indicated, they are put to one side and no longer have any 
bearing on the election.)  After this distribution, all the candidate's 
piles of votes are re-added, to take into account the additional votes 
some may have gotten from the last-place candidate.  Once again, if no 
one yet has won one of the seats, the last place person is kicked out, 
and their votes are distributed (to their #2 preference, or to the #3 
preference if the vote had originally come from the #1's of the 
first-eliminated candidate).   

As soon as any one candidate has reached the quota, the number of votes 
above the quota amount are distributed (votes are taken off the top of 
their last-counted pile I think--so it's essentially a random choice) to 
whoever their next preference is.  (So if the quota was 4211 votes, and 
after a couple of counts one of the candidate has 4250 votes, their 
"surplus" 39 votes are examined and are put onto the vote piles of the 
next preference indicated).

There might typically be anywhere from say 5 to 16 counts until all 3 
winning candidates are known.  Each count involves physically 
redistributing ballots from the pile of one candidate to another.  
That's why this is easiest to do in a centralised location.

It would be better described to give you a full description of a simple 
election--but I don't have the time right now.  I should be in bed!

Catherine

Andrew Ó Baoill wrote:

> Our proportional representation system, particularly the 'vote- 
> transfer' process as currently implemented, makes counting in  
> individual voting places problematic.
>
> Each voter places a '1' beside the candidate of their choice, a '2'  
> beside their second choice, etc. We're using a 'single transferable  
> vote' system - it's only if your first choice is elected or  
> eliminated (see below) that your second choice comes into play.
>
> The number of votes needed in order to be elected depends on the  
> total number of valid votes in a constituency (district). So if you  
> were counting in individual polling places you would need ongoing  
> communication between polling places and a central point.
>
> If no candidate gets a 'quota' their votes are redistributed to other  
> candidates (in order of preference). This would be feasible, if  
> somewhat complex in polling places (you would need additional piles  
> for these votes, as you should keep them separate from votes that  
> were originally for a candidate).
>
> Where it gets really complex is when someone is elected. Then their  
> 'surplus' - the number of votes in excess of the quota - is  
> distributed to other candidates. There are a number of ways this can  
> be done, including:
> (1) Distribute all votes, giving each distributed vote a fractional  
> value of (surplus)/(total_votes_for_that_candidate). This is a long  
> and complex process, and the figures would have to, again, be co- 
> ordinated across polling places.
> (2) Distribute some (pseudo)-random sample of (surplus) votes. In  
> Ireland we use a variant based on the last votes to be added to a  
> candidate's total.
>
> It is the variant of surplus distribution that causes the major  
> difficulty. If you have 50 polling places, and want to distribute 30  
> votes, or 75, it becomes difficult to figure a fair way to pick the  
> 'random' votes for the surplus.
>
> In addition, the multiple separate piles - you can theoretically end  
> up with several dozen for each candidate - make it a complex  
> operation that is easier done in a single space, rather than  
> duplicated in each polling place.
>
> I think this covers the main points (and is reasonably accurate).  
> Anyone?
>
> Andrew
>
> On 26 DFómh 2005, at 18:06, Patrick Kobly wrote:
>
>> Catherine Ansbro wrote:
>>
>>
>>> I wish we could get some kind of hand counting done at the voting  
>>> places.  I don't yet see any practical way of doing this.
>>>
>>
>> Explain to me like I'm a little baby why this doesn't work.  Being  
>> in Canada, where the paper ballots are counted in the voting  places, 
>> this comment is a bit surprising.  I understand the rules  and 
>> counting procedures are different.  Perhaps a pointer that  describes 
>> the process in detail would be useful.  I'm interested in  which 
>> aspects of the process limit the paralellizability of counting.
>>
>> PK
>>
>>
>> _______________________________________________
>> E-voting mailing list
>> E-voting at lists.stdlib.net
>> http://lists.stdlib.net/mailman/listinfo/e-voting
>> http://evoting.cs.may.ie/
>>
>>
>
> -- 
> Andrew Ó Baoill
> - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -  
> - -
> Inst. of Communications Research, U. of Illinois
> Communications / Participatory media / Political action
> - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -  
> - -
> andrew at funferal.org / baoillo at uiuc.edu
> aim: funferal at mac.com
> +353-87-241-7003 / +1-(217) 384-3142
> http://funferal.org / Galway, Ireland / Urbana, IL, USA
>
>
>
> _______________________________________________
> E-voting mailing list
> E-voting at lists.stdlib.net
> http://lists.stdlib.net/mailman/listinfo/e-voting
> http://evoting.cs.may.ie/
>
>




More information about the E-voting mailing list