Voice Referendum Ticks And Crosses Beatup

I thought I should make some quick comments on the matter breaking yesterday (following a Sky News interview) regarding the use of ticks and crosses in the Voice referendum.  There was sudden outrage on various right-leaning outlets and from politicians including the Leader and Deputy Leader of the Liberal Party when it was noticed that the AEC's ballot formality guidelines advise that ticks in referendums are treated as formal votes for Yes while crosses are treated as informal.  This has led to a fairly large outbreak of clueless wheel-reinvention and simple-minded outrage on social media.  It is doubtless worse on talkback.

I should note clearly near the top that the ballot paper will instruct voters to vote "Yes" or "No", and they will also be instructed thusly verbally if they are voting in a booth.  This debate only concerns the tiny minority who fail to follow the instructions.

People are entitled to their views on how Parliament should legislate in this area.  However the attacks on the Australian Electoral Commission for interpreting the legislation to the best of its ability are disgraceful, and in many cases are coming from lazy cowards who had every chance to pursue legislative change themselves but did nothing. Of course, had they tried to amend the legislation they may well have failed, but having not tried the honest thing would be to own their failure to even try, or I suspect to even be aware of the issue.  Sussan Ley, for example, tried to dodge questions about why she had not pursued the issue in Parliament and say it was a matter for the AEC.  No it is not.  It is the Parliament that sets the rules and that has the authority to render ticks informal for referendums (as they are for House of Representatives elections but not automatically for Senate where they are read as 1s) should it choose to do so.

The basis for counting ticks as formal and not crosses is Section 93 (8) of the Referendum (Machinery Provisions) Act 1984 which states "Effect shall be given to a ballot paper of a voter according to the voter’s intention, so far as that intention is clear."  This is what is known as a savings provision - it recognises that while not all voters will follow the verbatim instructions on the ballot paper exactly, some of those who do not follow the instructions may express a clear intention.  Counting such votes helps all voices be heard in a multicultural society with a range of levels of education, English language skills and electoral understanding, but can create controversy about what votes are clear or are not clear.  

Examples of votes that could be saved under such provisions include but are not limited to:

* cases where the voter writes Yes or No but disobeys the instruction to write the word inside the square

* cases where the voter clearly attempts to write Yes or No but fails to do so, eg they use a pen which runs out of ink partway through writing the S in Yes

* cases where the voter instead of writing Yes or No writes a variant, either in English or in another well known language

* cases where the voter instead of writing their intention uses a sticker or a stamp to communicate it

* cases where the voter starts writing one answer, changes their mind, crosses it out and writes another

* and cases where the voter uses an unambiguous symbol.

The AEC has long-standing legal advice to count ticks as Yes and crosses as informal - the latter except for cases involving multiple questions on the same page of ballot paper, where ticks for some and crosses for others would mean the crosses could be counted as No.  A tick appears to clearly indicate approval even if the voter has failed to follow the verbatim instructions on how to display that approval - no one so far in the debate has demonstrated any reason why a tick should be considered differently to "Yep!", "Absolutely", "Ja" etc.  However a cross is ambiguous - firstly because crosses are often used to check-mark applicable boxes on government forms (something that is in fact becoming more relevant not less in the electronic age where it is often the default form of check-marking in software) and secondly because a cross is used to signify voting for a candidate in many overseas elections with multiple check-boxes.

What we do know about the rate of ballots being marked with a cross is that it's very small.  The 1999 Republic and preamble referendums recorded only 0.86% and 0.95% informal, and that is a total including blank ballots, "scribble", illegible and otherwise ambiguous responses as well as crosses; I would expect the cross rate would have been below 0.1%.  (There is no indication I'm aware of re the tick rate, but if I find one I will add it).  In the 2016 House of Reps election - sadly the last for which the AEC has thus far published an informal ballot survey - the combined rate of ballots ruled informal because they contained at least one tick or cross was 0.39%, but that would have included voters who did not use a tick or cross as their primary response.  Far from steadily rising with increased ethnic diversity as might have been expected, that rate had fluctuated in the range 0.39% to 0.65% without a clear pattern in elections between 2001 and 2016.

History of this matter

Section 93(8) was created in 1984.  A thread by André Brett has noted that referendums used Yes and No boxes from 1906 to 1926, then used preferential-voting style 1s and 2s in Yes and No boxes from 1928 to 1951, and then the current format of writing Yes and No came in from 1965.  Equivalent intention-based savings provisions for referendums go way back to 1919 - see comment from Michael Maley below - and the AEC had legal advice from at least the 1988 referendum cycles that ticks should be counted as Yes while crosses did not display clear intention and should not be counted.

In 1999 Phillip Benwell of the Australian Monarchist League sought an injunction that would require the AEC to segregate a range of votes for both sides, including ticks for the Yes side, that would be saved under 93 (8) and record them under the heading of "Formality checks".    The application was rejected primarily because it was unnecessarily disruptive: the AEC was required to preserve all the ballot papers anyway.  In the event that there were enough relevant ballot papers to affect the result, a State or Territory could challenge the outcome in the event that the AEC's rulings were wrong.  But the judge also found that the AEC interpretation of a tick as a valid Yes vote appeared to be correct: " It is difficult to see, for example, what other intention could lie behind a voter’s use of a tick (with or without the word “NO” crossed out) [..]" [29]  Although that ruling was therefore non-binding, it is nonetheless an opinion from a judge to go with the AEC's advice that a tick should be treated as Yes.  The case did not cover crosses as there was no application to segregate them.  

In the leadup to the current referendum, the Parliament passed the Referendum (Machinery Provisions) Amendment Act 2023.  This made a clarifying change to formality by adding Section 93 (9) which made it clear that if the voter just voted Y that was to be counted as a Yes, and if the voter just voted N that was to be counted as a No (in each case that's provided that the letter is legible enough to be confident that is what it is).  (I am curious as to how the AEC would rule concerning a vote that said simply YE, a word that has other possible meanings not carried by other presumed Yes-equivalents like YEP).

The process of debate about the amendment Bill included a referral to the Joint Standing Committee on Electoral Matters.  This was an excellent chance for anyone who wanted to to submit that the Parliament should also clarify or amend views about other votes still covered by Section 93(8) - not one of the 78 submissions received by JSCEM did so.  No MP present at the JSCEM hearings asked the AEC any question about the formality of ticks and crosses.  No discussion of the formality of ticks and crosses was present in the Coalition's dissenting report on the JSCEM inquiry.  No Coalition or any other member raised the matter of ticks and crosses in debate on the Bill in either House.  No Coalition or any other member sought to amend the Bill to address the matter.   

Is the No side being diddled?

There has been much outrage from the No side based on the idea that the invalidated crosses are their votes that are being stolen from them.  Given the evidence that the number of invalidated crosses is so negligible, this seems more than a little bit precious in a referendum format where historically the double majority requirement has given No an average leg-up of about 3%.  Peter Dutton's claim that "It gives a very, very strong advantage to the ‘Yes’ case,” is very, very, very, very silly.

No reliable evidence has been presented that No is being disadvantaged at all.  Given that a cross is an ambiguous and apparently very rare response, it is possible that more voters who use a cross in fact intended to vote Yes than No, and that Yes is being disadvantaged by the counting of crosses as informal compared to if they could be counted based on some kind of Dr Mind Reader assessment of what each individual cross-marker meant.  And furthermore, cross-users who intended No would have to be much commoner than cross-users who intended Yes for the exclusion of crosses to unfairly advantage Yes by more than including crosses as No would unfairly advantage No.  (Assuming all cross-users intended one or the other, a 75-25 split would be required).  

What the No side seems unwilling to admit here is that there is a genuine asymmetry in how reliably a tick or a cross might be taken as saying something.  A potential solution would be scrapping the savings provision or amending it to disallow symbols, but (i) that would increase the informal voting rate (ii) the No side has had ample time to propose that and hasn't done so.

Social media responses

I've seen a lot of misguided responses on social media, and without singling out individuals I shall give some common types.

The first category is those who miss the point of savings provisions.  They try to argue that a tick should not be counted because the voter's intention is unclear given that the voter hasn't followed the instructions.  But if the Parliament accepted this argument then there would be no reason for it to have savings provisions in the first place - they are obviously there to do something that accepts that some votes that do not obey the instructions should be counted.  The Parliament has also had 39 years of chances and three federal referendum cycles of experience based on which it could refine the rule (for instance it could say that only words can be counted and not symbols) and has deliberately left it open, thus obliging the AEC to interpret it broadly as to the form of expression included.  The AEC could in theory decide that a tick is not clear voter intention, but that would be at risk of being overturned by a challenge to a sufficiently close result in the absence of any actual evidence that it wasn't.

The second category is those who reinvent the wheel and ask why the AEC does not design the ballot paper differently (again ignoring that ballot paper design is a matter specified by the Parliament).  Examples have included reintroducing separate Yes/No boxes, printing the words Yes and No and having the voter circle one, having empty circles that the voter can colour in, and so on.  

The problem with Yes and No boxes is that referendum provisions must be suitable for referendums conducted alongside or close to elections.  Antony Green has noted the 1991 NSW debacle where the use of Yes and No boxes caused the informal vote rate in the simultaneous Lower House election to spike from 3.28% to a ludicrous 9.32% (it was back to below 3% two elections later).  

For any other solutions it is easy to devise formality problems.  If the voter is required to circle a Yes or No vote, what do we do with a circle that is nearly closed but not quite closed?  A semi-circle? A square? A square with little bars in the middle of it perhaps indicating that the respondent intends to represent that option as being in prison?  (Voters are very inventive!)  What about if instead of circling an option, the voter chooses to tick an option?  Now what if the tick is actually between the two options but three times as close to one than the other? Two times?  Or what if the voter draws a giant circle that includes all of one option but also part of the other? At what point have they included enough of the other option to render their intention unclear?

Suppose a voter is filling a circle in.  What if they fill two-thirds of a circle only?  One-third?  All of one and a third of the other?  One only but with a horizontal line partway through the circle that might or might not be a crossout? Both circles but one in green and one in red?  What if they punch a hole in one of the circles instead of filling it in?  (This particular proposal reminded me of the hanging chad nonsense from the punch-cards in the notorious 2000 Bush vs Gore Florida case).

Many people on social media need to realise that electoral authorities have the benefit of a lot of experience in what works and what doesn't in electoral design.  In the case of the current referendum method, Australia has long experience of an informal voting rate of below 2% - and in the case of 1999 below 1%.  It is miraculous that the informal voting rate even for the obscure preamble proposal in 1999 was so low.  No-one's silly idea out there is going to beat that performance in a compulsory attendance booth ballot, unless it involves electronic voting which contains what I expect one day will be proven to be formally fatal security/anonymity paradoxes.

-------------------------------------------------------------------------------

Addendum 1:

I wish to add a few pieces of historical research.  Firstly I have determined that the current Ballot Paper Formality Guidelines has existed at the same URL and contained advice on the handling of ticks and crosses (ticks prospectively formal, crosses not) since at least August 2013.  This is a document that is widely read by party scrutineers and Coalition MPs should have been aware of its contents.  

Secondly I have found that during the 1988 referendum cycle Labor attempted to follow a 1986 JSCEM recommendation, reported as being to count only Yes and No as formal.  Labor's proposed amendment "For the purpose of ascertaining a voter's intention, regard shall only be had to the word `Yes' or `No' written alone or without words of qualification." was abandoned on 2 June 1988 because the Coalition and the Democrats opposed it, the Coalition saying that it didn't capture JSCEM's precise intention, and a consensus was reached to leave the savings provision alone and come back to the matter after the election.  Then on 24 August 1988 (with the referendum on 3 September) there was a sudden blow-up with the Coalition unsuccessfully attempting (with voting underway) to revisit the treatment of ticks and crosses (based on the AEC having advice that crosses on all four questions of the 1988 ballot were informal, but four ticks or any combination of ticks and crosses was clear voter intention).  

Labor tried again on 21 December 1989, moving an amendment to add "For the purpose of ascertaining a voter's intention, regard shall only be had to the word "Yes" or "No" written alone or without words of qualification."  The Democrats moved an amendment to change that to "For the purpose of ascertaining a voter's intention, regard shall only be had to the word "Yes" or "No" written alone within the square provided, irrespective of words of qualification elsewhere on the ballot paper". Labor said they would not accept the Democrats' amendment as it would mean that a voter who wrote Yes inside the square, crossed it out and then wrote No outside the square, would not be counted.  As a result both Labor's amendment and the Democrats' amendment to the amendment were withdrawn, and the 1984 savings provision again survived.  This is the last time the parliament seems to have addressed the issue.

Addendum 2: 

I have found two times between 1990 and 2023 that the matter appears in Hansard.  The first is 24 Oct 2009 where the matter of the asymmetric treatment of ticks and crosses was explicitly discussed by Michael Maley and current (but then not yet) MP Julian Leeser said "I would like to see, for what my submission is worth, the tick provisions removed."  This shows that at least one serving Coalition MP was aware of the issue which the Coalition did nothing about when in government, as well as probably every second Coalition election scrutineer of any quality.

The second is not so flattering for the AEC unfortunately - on 10 Sep 2015 with current Senator Linda Reynolds present the Legal and Constitutional Affairs Committee, heard discussion of the Benwell v Gray 1999 case with Paul Pirani of the AEC saying:

"To add to that, people think that yes and no is all you can do in a referendum. Can I bring to the attention of the committee that there was a case in the Federal Court in 1999, Benwell and Gray; its citation is 1999 Federal Court of Australia 1532. In that case Mr Benwell tried to argue before the court that our guidelines for scrutineers, which said that a tick and a cross could also be used, were inconsistent with the requirements of the referendum act. In that case the court held that a tick or a cross could be used instead of a yes or no, provided that the intention of the elector was clear from what is on the ballot paper. So there are complexities—it is not just as simple a thing as saying, 'that is yes, that is no', and that is the end. There are also issues about ticks and crosses and other variations."

Unfortunately this is not the most accurate of summaries - the guideline for scrutineers being discussed in that case (for the 1999 republic referendum) explicitly stated that crosses were informal.  

Addendum 3: 

And more!  On 8 Aug 2005 Sir David Smith appeared at JSCEM in relation to a submission where he proposed to restrict answers to Yes and No.  There was a lengthy discussion in which, among other things, then Government Senator and later Attorney-General George Brandis said:

"I was wondering about the issue of the cross. I understand why a tick might be regarded as acceptable, because I think most people regard a tick as a universal symbol of assent. But a cross is ambiguous, it seems to me. Some people regard it as a mark which may be a symbol of assent, and other people regard it as a mark which may connote disapproval."

Despite nearly the whole of Smith's appearance as a witness being devoted to discussing referendum formality, there is not even a mention of the issue in the 2005 JSCEM report on the 2004 election.  (Smith then made a similar submission to the 2009 Legal and Constitutional Affairs Committee's 2009 inquiry into the machinery of referendums, which was ignored.)

I have also found a discussion at the AEC's appearance in Finance Estimates 30 May 2013 where the Benwell v Gray case was also discussed, albeit again in less than accurate terms.  

I have also found that there was a similar ticks and crosses media storm in the 2009 WA Daylight Saving state referendum.

Addendum 4: 

Australians for a Constitutional Monarchy also raised the matter in their 2020 submission to the Standing Committee on Social Policy and Legal Affairs, albeit citing Smith in the context of pushing the old chestnut that informal votes should be in effect counted as No (see comments).  Several current Coalition MPs were involved but the matter went no further than the ACM submission.  

UPDATE Sep 10:  A Ralph Babet/Clive Palmer challenge on this matter will be heard in the Federal Court on September 20.

UPDATE Sep 20: The challenge has been rejected with Justice Rares finding that the AEC's interpretation as concerns ticks and crosses is correct under law, and declining to rule on whether Babet/Palmer had standing.  Written reasons tomorrow, Babet/Palmer have one week to appeal.

UPDATE Oct 10: Babet/Palmer appealed but the appeal has also been rejected.