Crikey’s received several emails recently about the NSW Local
government elections claiming the electoral commission does not count
all the votes. Unsure what this was about, we contacted the guru Antony
Green to find out the story, and as usual he has responded with an
It is true that at the recent NSW local government elections, not
every vote would have had its preferences counted in full. All ballot
papers are counted for primary votes. However, random sampling
procedures are used for parts of the distribution of preferences. This
in effect means the system has no definitive result, merely the result
produced by a given random sample of preferences. Re-counts are not
possible because you may get a different result.

Yes, the system is that silly.

Most NSW local government bodies use a variant of quota-preferential
voting, similar in theory to the Senate and Tasmania’s Hare-Clark.
However, for reasons that make no sense, NSW is the only state that
does not examine every vote for preferences during the count.

If a candidate is elected with more than a full quota of votes under
all quota-preferential systems, some votes are set aside as the quota,
and the surplus to quota preferences then distributed as preferences to
other candidates.

So if a quota was 25,000 votes, and a candidate received 40,000 votes,
15,000 are surplus to quota. You calculate a transfer value (15,000 /
40,000), which is 0.375.

Now, everywhere else in Australia, every vote would be transferred, but
at this reduced value of 0.375. However, in NSW, votes are split out
into valid next preference bundles, and then randomly sampled using the
transfer value. Then the selected ballot papers distributed.

So if 1,000 of the 40,000 votes for candidate A had 2nd preference for
B, the transfer value is applied to these votes. So a sample of 375
votes is extracted and distributed, while 625 remain with A as part of
the quota for election. Interstate, all 1,000 votes would be
transferred at 0.375.

The problem is, how do you ensure your sample is random? How do you
know that the 375 votes you sample do not have significantly different
3rd, 4th, 5th etc preferences than the 625 that remained with candidate
A? The answer is you do not know. It is a random sample. For that
reason, in NSW once the primary votes have been settled and agreed to,
and the quota set, you cannot do a re-count of preferences. Using
random sampling, you could never guarantee the same result.

Yes, let me repeat that. Under the NSW counting system, you have no guarantee of getting the same result twice.

In fact, the NSW system is even madder thanks to recent ballot paper
changes. There is no longer a registered ticket of preferences for each
party. Instead, voters may preference order parties above the line in
the same way they can preference order candidates below the line. This
has made the whole process of ensuring a random sample more difficult.

So, what has been the solution in NSW? Well, they have typed all the
votes into a computer to ensure the random sampling is done correctly.
Not to ensure that the vote is counted correctly, but to ensure the
sampling is done correctly.

Now the computer could count every vote. The only thing that stops this
occurring is that the act specifies random sampling. So the NSW local
government act changes have resulted in all the votes having to be
entered into a computer system, but then counted using a random
sampling process that cannot be repeated because it might produce a
different result.

Without doubt, one of the silliest cart before horse computing solutions I have seen.

And don’t forget the other mad changes for these elections. Parties had
to be registered twelve months ahead of the election. This has been a
conspiracy of the established major and minor parties against local
independents and resulted in a fall in the vote for Independents.

Secondly, the mad rules for registering how to vote cards forced
independents to direct preferences to other groups even if they did not
want to, and even if their vote met the required formality rules.

I’m surprised there has not been more fuss about all these changes.
Certainly there should be more questioning of the procedures used to
conduct the local government elections.

But I wouldn’t hold your breathe. For the last century, the rules
governing elections in NSW have been fiddled with by governments of the
day, and no one has ever publicly examined the way the changes worked.
In NSW, on electoral matters, self-interest always wins out over the
public interest.

Maintain the indignation.