[E-voting] Some back of the envelope calculations on e-voting (and worrying conclusions)

Cosgrave, Mike M.Cosgrave at ucc.ie
Sat Apr 17 08:25:08 IST 2004


I don't think it is quite as bad as Dermot says.

Accepting his 3m voters, or 441 per machine as a good figure; the real
problem is the 30% of those who'll vote between 6pm and 10pm - say between
140 and 150 voters per machine in 240 minutes. During that evening rush,
voters need to get through in under 2 minutes each. 

In a paper vote, 2 mins per voter is no problem - I can usually vote much
faster than they can find me on the register, cross me out, punch my paper
and hand it to me. Writing down 1,2,3,4,5,6,7,8,9 isn't hard, and doing it
on the machine shouldn't be either. The machine setup for each voter tho'
could be a big slowdown.

A useful comparision might be an ATM - most people can perform an ATM
transaction in under a minute, I think (anyone in Banking got the numbers on
this?). Mutiple transactions (check balance, print mini-statement, get
cash), comparable to voting in 3-4 polls, could easily run to 2 mins tho.

I think Dermots estimate of 5 mins per voter is on the high side. It is a
potential problem, and if the turnout goes above 50% it could become
serious. 

Obviously, the best way to test this is to encourage as many people as
possible to turn out and vote. It is legitmate to suggest that if the
turnout goes to 60&-70%, there may not be time for everyone to cast their
vote, and therefore encourage people to turn out and demand their right to
vote. While we don't want to take sides, anti-government voters who might
not bother to come out might find the opportunity to screw up the system
appealing.

Mike

-----Original Message-----
From: Casey, Dermot (GE Consumer Finance)
To: E-Voting
Sent: 4/16/2004 12:37 PM
Subject: [E-voting] Some back of the envelope calculations on e-voting (and
worrying conclusions)

I did the following rough calculations and came to some worrying
conclusions..... Am I making mistaken assumptions. I'd love to be proven
wrong on this. 
At a minimum we will be facing long queues in June. In the worst case
people will not be able to vote before polls close

Regards

Dermot

6,500 voting machines (now 6,800 apparantly)

Election timetable
Polls Open 8am
Polls Close 10pm

1. How many voting machines required
Assumptions
A.1 Assume 3 million voters (total electorate in 1999 was 2.86 million)
A.2 Assume 6,800 voting machines (based on figures given)

There will be votes for 
1. European Elections
2. Local Elections in some cases two 
   2.1 County Council
   2.2 Town Council
3. Referendum 

Allow 5 minutes to do all the voting for all the options and review you
votes
This doesn't seem unreasonable for setting up the machine, doing your up
to 4 votes, checking your votes and pressing the big red button.

Scenario 1 Everyone Votes
Average distribution of 441 people per machine
Requires 2205 minutes per machine
Number of minutes in 14 hours  - 840. There is no way everyone could
vote.

Scenario 2 50% Turnout
Average distribution of 220 people per machine
Requires 1103 minutes per machine
Number of minutes in 14 hours  - 840

Conclusion even with a 50% turnout it is unlikely that everyone could
vote, unless people were constantly voting and it took them less than 3
minutes each!
This makes no allowances for machines failures or other problems during
the day in question.

The big thing to remember here is that voting is not evenly distributed
during the day.*
There is a burst in the morning with people going to work. 
There is a trickle to lunchtime.
There is a burst at lunchtime
There is a drip in the afternoon 
And a large number of people vote after work.
* does anyone have accurate figures for the pattern of when people vote,
and I could model this properly

_______________________________________________
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