The South Australian election finally concludes today with the resolution of the Legislative Council count, just one day before the resumption of parliament. It does not appear to be in doubt that this will result in the election of five Labor, three One Nation, two Liberal and one Greens member, to add to the five Labor, four Liberal, one Greens and one One Nation-turned-independent members elected in 2022.
The lower house outcome of 34 seats for Labor, five for Liberal, four for One Nation and four independents has resulted in contention over which party should be granted official opposition status. Normally the Liberals’ superior seat count would decide the matter, but One Nation has claimed a moral right based on its higher vote share and the number of seats in which it reached the final preference count. More concretely, it has sought backing from independents, while reportedly also targeting potential Liberal defectors. The legalities of the matter are rather fuzzy, at least from my perspective – when a party defection in Western Australia put the Nationals and Liberals at parity on three seats each in 2023, it was a relevant consideration that the Labor Speaker favoured the status quo. Barring any firm commitments from independents, of which there are have been none that I’m aware of, it is presumably relevant that Peter Malinauskas says he has been “working under the assumption that (Liberal leader) Ashton Hurn is Leader of the Opposition”.
The tardiness of the upper house count is one of a number of concerns that inspired the government to appoint the former federal Electoral Commissioner, Tom Rogers, to head an independent review. Most notable was the failure of 642 out-of-district votes cast in the regional electorate of Stuart to be sent on as required to be counted in the seat for which they were cast, which was not discovered until after the results had been declared. The 81 votes cast for Narungga were examined to check if they provided a basis for One Nation’s 58-vote win to be voided, and it was found they would have increased the margin by 14 votes. A week later, the Electoral Commission revealed further uncounted votes had been found for Enfield and Newland without offering further details, though Attorney-General Kyam Maher said their number was “in the order of dozens”.
The close contest for Narungga was again in the spotlight after it emerged a One Nation volunteer had been allowed to count votes at an understaffed booth in Port Clinton, and that a community member enlisted ad hoc to tick people off the roll had allegedly failed to initial ballots as required, resulting in them being counted as informal. However, the Liberal Party has provided no indication it might challenge the result.
Can we please keep this thread for discussion of South Australian state politics. The open thread for general discussion is here.
The prospects aside from a fifth Labor for the Legislative Council are a second Green and first Legalise Cannabis, correct?
Has there been any public explanation of why the vote count took so long, or how ballots were lost? I have not seen one.
How do you declare a result with ballots missing? With all ballot boxes numbered and tagged, and the number of votes issued recorded, as a minimum EC staff should have known votes were missing.
There was discussion on ABC talkback radio of the EC not contacting many past election workers. Regardless of which party won, this is odd. The increased number of pre-poll votes should have made the full count easier.
Socrates @ 10.10am
The transfer of sensitive electoral materials like ballots from one place to another always has the potential to cause problems if it isn’t meticulously planned and executed. There should always be a parallel paper trail which documents what is coming.
Another issue worthy of investigation by Mr Rogers is whether the ECSA sought to draw on the extensive work done by the AEC over several years to ensure that the problem with lost ballots at the 2013 WA Senate election would never be repeated. If it didn’t, that suggests that the mechanisms by which Commonwealth, State and Territory electoral authorities learn from each others’ experiences need to be revisited.
That last part with an ad hoc community member drafted in to do a critical job and – understandably – failing to perform the technical requirements and rendering an unknown number of votes informal might be the worst of all of it.
What an absolute shitshow.
I doubt the SA Liberals have the money to fund a challenge in Narungga. Nor do I think they would fancy their chances in a re-run.
It’s far better them to bide their time and wait for the inevitable One Nation implosions to occur. If the Libs are really smart they might even pick up a defector or two.
Result out. Labor wins last seat as expected.
While plenty of people have complained about the mistakes, irregularities and mismanagement, it seems to be mostly One Nation supporters who have expressed some suspicion that the system may be rigged against them. From my reading of comments on many Facebook posts.
That being the case, I think it’s lucky that:
(a) One Nation won Narrunga, the closest seat and the one with the potential to be affected by losing or ‘finding’ votes.
(b) It was a One Nation volunteer who helped count votes and not someone from another party.
Why on earth is it going to take ECSA another few days to provide more detailed reports. As of a couple of minutes ago, on the Legislative Council results page, they have the 2022 results appearing as the 2026 results.
Currently ECSA has the main results up on website correctly but inexplicably they also have results up from the 2022 Legislative Council election. Apprentice looking after website.
The election highlights why it’s important to properly resource elections
Website corrected but still waiting for print up. Could be a while as Labor 5 elected on count 5948. Labor 5 ended on 54 160 with quota 94 800. So only added about 15 000 from preferences. A lot of minor party votes exhausting.
I won’t be trawling through 5948 counts! That’s best left for others. However, we already know from 2022 that there is a high exhaustion rate for SA LegCo elections.
This election has had some problems with how it was run. I think it has been exacerbated by the sheer lack of transparency and a sense of arrogance and resistance to accountability from the higher ups in ECSA. I certainly hope the pending review identifies the problems and proposes some substantive solutions (that are taken up.) While I think there has been some operational problems with how this election was run, I certainly think none of it has been malicious or anything like that – and certainly any conspiracies from certain corners implying that ECSA were rigging the contest or whatever shouldn’t be given any oxygen. Although, considering the state of modern discourse, and how widespread conspiracy theories and persecution complexes are, I’m not holding my breath.
As for the LC results, in the end it turned out to be the most predictable outcome. How boring. Just kidding. Boring is good nowadays. Anyway, with Labor being just two short of a majority and three different parties having at least two MLCs, Labor have different paths to get legislation passed. Of course, they’re most likely to get things passed with the Greens but, if there’s legislation they want passed that the Greens refuse to support, they can try and get it passed with Liberal and/or One Nation support.
Right now, Sarah Game’s vote is superfluous (unless it’s a private member vote.) However, acknowledging the history of defections from One Nation (a point made on this forum repeatedly), if one of the One Nation MLCs decides to stay true to that stereotype and sits as an independent (or sole member of some “next big force in SA politics” party that ends up amounting to nothing, then their vote in combination with Game’s could get the legislation over the line, if the other parties say no. But that seems a bit convoluted.
Greens 2 ended up in 12th position which explains the small increase in Labor 5 total from exclusions.
The Greens did pretty well on preferences to get ahead of Legalise Cannabis and then Family First to get into 12th position. Bit of a lesson for some purists who rather take delight in not giving prefences beyond those they closely agree with.
(Federal) JOINT STANDING COMMITTEE ON ELECTORAL MATTERS
Inquiry into the 2025 federal election – Public Hearing
Tuesday 5 May 2026 (tomorrow)
Electoral Commission South Australia, 9am-11am.
https://www.aph.gov.au/Parliamentary_Business/Committees/Joint/Electoral_Matters/2025federalelection/Public_Hearings
I am feeling rather more sympathetic to ECSA today. The final report on distribution of preferences has been released – all 5948 pages!
Scrolling to page 5948 reveals some interesting information about the last spot filled:
Quota was 94,829 votes.
89,438 votes were exhausted and a further 1,307 votes were “lost by fraction”.
Labor’s Scriven was the last to be elected with 54,070 votes – 0.57 of a quota.
The Greens (McCusker) ended up with 44,019 – 0.464 of a quota.
After first preferences were counted, Labor had 4.405 quotas; the Greens had 1.224 quotas.
The Greens overtook both Legalise Cannabis and Family First in the contest for the last spot.
All in all, the Greens did quite well in the preference contest, picking up 0.24 of a quota compared to Labor’s gain of 0.17 of a quota.
However, in the end, it wasn’t even close – Labor “wining” at the last count by 10,051 votes.
It is an enormously complicated counting process, and I have more insight why it took so long for ECSA to finalise the results. Let’s hope that in 2030 ECSA will be better funded and better resourced, so that a repeat of some of this year’s glitches can be avoided.
How on earth was there 5948 counts? There were 48 candidates – that’s more than 100 counts per candidate! The first nine won on quotas, so that’s the first 10 counts; then winnowing out the lower order candidates until there’s one for each party would be another couple of dozen. One Nation got the tenth seat partway through, whatever surplus they had when they hit the quota was distributed, then just exclude candidates until there’s two left. How is that ~5900 counts? I get that BTL votes complicate things, but to that extent?
@Outsider the calculation of the final result takes place in the blink of an eye.
It’s the data entry that takes the time. Not helped by the delayed submission of a small number of ballots.
No doubt the inquiry will look at this and suggest improvements to the process.
The report from the election inquiry will make for interesting reading.
SA needs to simplify its upper house preference distribution system. A 5948 page distribution is farcical. Probably a result of failure to use amalgamation of vote parcels of the same value.
I think the main root cause of this problem was the predetermination of what the 2CP counts were going to be before the counting of ballots even began, and having to change 30 different preference counts just because they got the top 2 wrong. Had they actually took the time to start counting the votes before doing indicative counts like how the AEC does its counting there wouldn’t be such a huge problem with the counting of votes in the first place, and they wouldn’t have to redo as many preference counts which adds time to counting the Legeslative Council as well.
TBM – There’s a few things going on here.
I don’t have knowledge of the exact work sequence that ECSA runs, so can’t specifically say that changes of 2CP counts directly contributed to the processing of the LC ballot papers. If these two tasks were assigned to the same group of people, then perhaps it’s a thing.
Regarding 2CP candidates. They are predicted by all EC’s ahead of counting. Polling place managers have (theoretical) access to the determined candidates selected before the start of any counting and reporting systems are designed to take in information based on said determined candidates.
Starting a count before deciding 2CP candidates is not a terrible idea but it brings a whole new series of logistical issues. Basically you have to have enough of the overall count done to know who the 2 candidates are and then you have to inform and have the count conducted across your widely dispersed election night workforce in a timely manner. The former is perhaps achievable; the latter raises all sort of new things.
“Having to change 30 different preference counts just because they got the top 2 wrong.” is a genuine (and somewhat new) issue for EC’s which is only solved by any number of workarounds; it’s actually a system design issue.
2CP candidate choice is going to be a huge issue for VEC later this year (heck, the single by-election just now had the wrong selection) and a large enough issue for NSWEC early next year for their respective state elections. The AEC should thanking all the gods there’s no federal election soon.
There’s an interesting TCP perspective from “Simon” over at https://antonygreen.com.au/2026-nepean-by-election-results/#comment-3108
“Why haven’t they changed the 2CP to ON and Lib*. Seems incredibly bias and arrogant to assume the left or independent automatically gets 2CP consideration. Happened throughout SA too. This alone is enough for me to switch right.”
“Yea, sure it was clear**, but it’s the principle around how ABC and other pundits broadcast and advertise the race on forums, dashboards and TV portraying this misguided perception that ‘outsiders’ aren’t part of the equation or contention. But that will change I expect.”
I found it insightful and unexpectedly informative that the 2CP candidate choice and a perceived bias in the 2CP candidate choice was enough that Simon has considered changing his vote. I don’t know how widespread such a perspective is but should be something that all EC’s are atune to – that is, an incorrect 2CP candidate selection may call into question the independence of the EC and any other organisation using this information.
* My note. Simon is referencing the incorrectedly selected Nepean (VIC) by-election 2CP candidates.
** My note. In reference to Antony noting the leading / likely to win candidate.
There might be some seats where you don’t know who the final two will be but you do know the final 3. In those cases the optimal process would be to do a 3CP. There will be some scenarios where once you know who came 3rd, you know who came 1st, so you know who won the seat despite not having a 2CP count.
There will be cases where you know pretty well who will win even though you don’t know who the final two will be, in the same way that you often know the final two regardless of the order of exclusion of the others.
With an increasing number of contests the 2nd v 3rd margin is smaller than the 1st v 2nd, and had 2nd & 3rd been the other way a different candidate would win. In such places merely listing the 2CP margin doesn’t show the closer margin that mattered. It’s therefore trickier to describe how close a seat is. Even a 2D pendulum can’t convey it all.
Whether ECSA are to blame or not, releasing the results the day before newly elected members take their seats is unacceptable. New members lacked the proper “ed-u- macation”.
Surely the expected time frame and required resources could have been foreseen and extra resources provided – or the commencement of the LC delayed.
The actual overall vote was roughly in line with polls, so how was this not better co-ordinated.
By all media accounts the only seat in doubt (almost from election day +1) was the final seat. If the LC must start it’s work on that date then could there there not be a provisional determination of those elected that are reasonably beyond doubt so that “indoctrination” into the system can occur and suitable training happen. The final member finally elected may then be placed in a “buddy” system for a short while whilst orientation occurs.
@PaulTu
The date for the return of the writ was “on or before 4th May 2026”
The EC met that legal requirement.
There can’t be a provisional determination because the law doesn’t make provision for one.
“Media accounts” and “reports” aren’t actual results and bodies like Parliament deal in the actual results via the return of writs.
@Bird of Paradox
On his site Antony Green write this in response to a similar question
“COMMENT: SA and WA carry out a single count for every bundle of votes received by a candidate. So if when excluded a candidate has votes sourced from 10 other candidates, then this corresponds to 10 counts, one for each receipt from another candidate, distributed as bundles in the order they were received. The Senate system excludes all votes received at the same transfer value no matter when received. This is why SA and WA have huge count numbers. SA should accumulate its reports by stop points, say accumulating all distributed votes into a report that only has a count when an exclusion is completed. Or adopt the Senate counting method.”
https://antonygreen.com.au/sa2026-legislative-council-result-declared/#comments
That still doesn’t quite explain it. There were 39 candidates left after the ones with quota were elected, the last two of which were never excluded (Labor #5 and Green #2); denote them as C1 to C37, in order of when excluded (or elected, for the One Nation #3 who won the 10th seat – distributing their surplus would’ve worked similarly, just with a transfer value that isn’t 1), and count all candidates they could’ve got preferences from:
C1 excluded, prefs from 1st 9 only = 9
C2 excluded, prefs from 1st 9 + C1 = 10
C3 excluded, prefs from 1st 9 + C1,C2 = 11
…
C37 excluded, prefs from 1st 9 + C1…C36 = 45
9+10+11+…+45 = 37*27 = 999 counts. Still an outrageously huge number, but not 5984.
5984 is just a smidge less than 6*999 = 5994, which sets off my mathematical spidey-senses. Might be a reason why, might just be a coincidence.