A big gripe about the ACT Hare-Clark system is that you do not get a clear result on the night. Indeed, in the part two elections it has taken more than a week to get the result of the last seat confirmed.
This Saturday, it is very likely we will know all 17 MLAs mid-evening on Saturday. This is because of electronic voting. Electronic voting will be available at Melba, Richardson, Gungahlin and Weston, and the four pre-poll voting centres. At the very least 10,000 voters will vote this way. More likely 20,000 and may be double that.
The electronic vote can be counted right down to the last preference in a matter of minutes. Manual voting takes much longer.
Electoral commission staff count the first preferences by put the ballot papers in piles under each candidate’s name. When preferences are distributed, the ballot papers are physically carried across to the pile of the next preferred candidate. It takes a long time.
Further, in Hare-Clark mullet-member electorates, a candidate gets elected on a quota – 12.5 per cent in the seven-seat Molonglo electorate and 16.6 per cent in the five-seat Ginninderra and Brindabella electorates. At the count, the preferences of candidates who get more than a quota in their own right also have to be distributed. For example, take a candidate with one and half quotas, say 15,000. One quota (10,000 votes) is used to elect the candidate and the remaining votes are distributed according to the preferences. But the electoral commission does not just take 5000 votes out at random. Nor does it count the preferences of the ballot papers that happen to be left over once the quota is reached in the first count. Rather it counts the preferences of every vote and assigns them to the next available preferenced candidate at a reduced value to take account of the votes “”used up” in getting the quota. In our example, at 50 per cent value.
It adds to the time taken to do the count. Incidentally, the last seat in the House of Representatives always takes at least a week, often, two weeks, to decided, but because the House of Representatives has 148 seats, government rarely depends on the last seat. Whereas, in the ACT, government usually comes down to the last seat or two, so the wait delays the formation of government and is more serious.
But that will be a thing of the past because of the electronic voting.
But how will just 20,000 electronic votes speed up the process so dramatically?
Well, in the long run the 20,000 votes will form part of the main count and be treated just like manual votes. However, once they are in the computer they can be counted separately as if they were the only votes cast in an election and a result worked out as if they were the only votes in the election. It takes just a few minutes. The electronic result giving all 17 members of the Legislative Assembly is like a giant deadly accurate opinion poll with a sample of 20,000 instead of the usual 600 and going to the last preference.
Ah, but what if the sample is skewed? Surely, it is not a random sample. What if the sort of people who would be attracted to the option of electronic voting (and it will be optional) are more likely to be young, forward-thinking, better-educated and therefore more likely to vote for the Yuppie Education Party than the general mass of voters?
It is possible. This is the damnation of opinion polling. Opinion pollsters have a way of verifying the quality of their sample. They ask demographic questions (age, sex and postcode) in addition to voting intention. They then check the demographics of the sample with demographics of the whole population as provided by the Australian Bureau of Statistics. If the age-group spread and postcode spread in the poll sample are similar to those in the whole population, you can assume that the voting intention will be the same.
But we don’t know the demographics of the electronic voters because we have a secret ballot (in fact the electronic vote is more secret than the paper one for the visually impaired who would otherwise need help).
However, we have one checking mechanism. We will know the first-preference voting trend of the whole population and the first-preference voting trend of the electronic-voting sample. If they correlate, you can assume the preference voting will correlate, too. Moreover, unlike opinion polling, our electronic sample will be take right on election day, so there will be no error factor due to people changing their mind between poll and election.
There may be some hidden factor. It may be that electronic voters will tend to number more (or fewer) preferences. We will know later on. My guess is, though, that we will be able to confidentially calculate the identity of all 17 successful candidates on the night from the electronic vote.