Secret Harbour (ALP 11.54%)
George Tree (ALP) vs Luke Herdegen (ON), Ryan Robertson (Lib) and way too many others
Cause of by-election: resignation of Paul Papalia (ALP)
CALLED (9:34) One Nation gain from Labor
-----------------------------------------------------------------------------------------------------------
Commentary (Comments scrolling up with latest first - refresh for latest comments)
Sunday 10:50 am I am not sure we'll ever see a Labor vs Liberal 2PP for this seat but I estimate that the Liberals would have won it, around 52-48 (by normal preference flows it would be around 53 but I have reduced that on account of it being a normally weak seat). This suggests Labor would have lost this seat even without the One Nation surge. Given that the average swing in government by-elections is around 5% an implied swing of 13.5% is high, but also Labor are coming off a still high baseline from the last election, which would account for some of it.
With results like this there is always a lot of talk about high swings, but the 2CP swing is not meaningful as it is not the same combatants. The swing against the Labor 2CP is what I call a pseudo-swing; both the pseudo-swing and my estimate of the 2PP swing are lower than in Rockingham 2023.
Sunday 6:00 am What has happened in the prepolls is that Herdegen has actually got a preference flow more like I was expecting for the by-election as a whole, 52.8% (which is not surprising given how strong the prepolls were on primaries for One Nation), and so the result has gone back to the high 50s, 57-43 at present, that was expected before the bad preference flows came in. Overall now One Nation has 49% of preferences. From the primaries I would have expected them to get about 52.7% without considering the ballot order, which would close at least a little of the gap.
10:30 End of night wrap: The story of this by-election is mostly in the primary votes. It was to be expected that all else equal One Nation could push well over 30 while Labor would drop into the same sort of range, but there was more to see than that: 39% for One Nation is above expectations and sub-25 for Labor isn't acceptable. The Liberal primary is slightly higher than expected and I think they might have taken bites from Labor too, though it appears that for whatever reasons the Liberal flow to One Nation was quite weak. This is a result that will resonate beyond WA and beyond Labor's candidate choice issues here because it does suggest that outer suburban seats with conventionally high One Nation votes are at risk.
10:17 Just as I was thinking the prepoll would surely be many more hours or tomorrow it has dropped! Herdegen is on an extremely strong 42.3% in it and One Nation have obviously now won. They may well end up pushing 56-44.
10:11 A riveting battle for last between Paul Gullan (23 votes) and Peter Dunne (26). Legalise Cannabis have done okay and may or may not get their deposit back depending on prepolls.
10:04 All booths except Secret Harbour EPP (day booth and prepoll) are in and it's still 53.9-46.1. At the present rate I don't think I'll be staying up to see if the EPP reports tonight.
9:50 I momentarily posted that I was calling this but that was based on a massive miscalculation of what Labor needed off the prepolls, so I've removed that - they simply need to about reverse what One Nation has done so far. Nonetheless I think there is pretty much no chance anyway that Labor can do well enough to pull back the One Nation lead.
9:27 Herdegen now out to a live lead 53.9-46.1.
9:05 Another pile of postals has arrived (not yet counted to 2CP) which has done nothing for Labor's position; indeed my live projection is out to 53.3. (Edit: Coastal Lakes College primaries have put it back to 53.1 but I think that's too low because of the size of the postal vote).
8:57 William Bowe says the prepoll will report in two stages, firstly its day booth numbers and then the prepolls.
8:40 Now projecting the two booths with outstanding 2CPs to turn a current 53.2 into 53.1. Some more theories on the weak preference flow to Herdegen - firstly as a youngish candidate with a drug history he might not be playing well on preferences with older Liberal voters, though it seems not to be hurting his primary vote which is super-strong. Secondly his position on the ballot paper is poor for catching preferences from the Liberals (indeed a few percent of the Liberal votes will just be donkeys).
8:14 With the strong postal preference flow added to the model my live count estimate is now 52.6. Closely matching the live total because there are not that many booth votes to be added to the 2CP. Labor would need to comfortably win prepolls although they are probably losing day votes and that seems incredibly unlikely so I've now put One Nation back to expected win status.
8:10 I've now got a variable flow model going and I'm projecting One Nation to 51.6 on the currently counted primaries - excluding the two remaining booths and prepolls. The prepolls will decide if Labor has any chance left.
8:02 Couple more booths in were pretty horrible for Labor on primaries. Off the booths counted to primaries I'm projecting One Nation to 51.4% but that's on the preference flows from the booths counted to 2CP, and I think the flow to them will go up. So One Nation clearly favoured again but I doubt it will be the huge win it earlier looked like.
7:57 I think the flows to One Nation may improve with more counting - both the booths counted have high Green votes. But still, while they may be ahead it's probably not by a lot.
7:46 And now some preference flows. In Comet Bay College Herdegen is getting 43.9% and in Port Kennedy Primary he is getting 35.8%. These are way, way weaker than I expected - maybe there's a Hastie factor here! This aint over yet. If these flows continue it will project to close to the wire - need to see more booths.
7:37 Couple of booths in, Comet Bay College with Herdegen 36.9 Tree 26 Robertson 14.4 Greens 9.6 and Port Kennedy Primary with Herdegen 32.9 Tree 28.2 Greens 14.0 Robertson 9.4. Overall Labor are now 8.6% behind on primaries with the Liberals way ahead of the Greens so this is looking like a big win for One Nation.
7:31 Mobiles have been added, highly unrepresentative with a strong Liberal vote. (PollBludger was showing these as the WAEC prepoll but this has been corrected.) The postals have been augmented with 2597 in now and now Herdegen has 35.5% to Tree 25.6% so a slight improvement for Labor there. Notably the Poll Bludger projection has already called the seat.
7:17 VOTES! We have 1277 postals and they are very nice indeed for Herdegen, he's on 35.9% to Tree 25.0% Robertson 18.8% the Greens have 7.8%, Legalise Cannabis 3.0% and the rest have nothing much (Paul Gullan has literally one vote). Of course postals will favour the conservative side so this will come back but when the postals are projecting at something like 57-43 to One Nation (my estimate) this is going to take some serious coming back from. Especially as postals are likely to favour the Liberals more than One Nation.
7:00 First hour down and nothing to see.
6:35
Casey Briggs notes that 50.1% of enrolment voted before today, turnout will probably only be 80 or so so that doesn't leave much for the booths.
6:05 Game on! Awaiting first numbers, this could be a while unless we get a little drop of mobile votes or something.
William Bowe relates that the WAEC doesn't disclose who the 2CP count is until they're confident they have the right pair, though in this case there's a good chance that won't take long. While we wait you may be amused to note that in the 1990 Maylands by-election one Arthur Galletly (IND) scored a heroic 0.07% (12 votes), which I think even the worst of this lot could have their work cut out to get under. He was one of six indies to get less than 1% in that one.
-----------------------------------------------------------------------------------------------------------
This is another combined prospects/live (with added snark) post for a state by-election of unusual interest. Since One Nation surged into contention in national polling, there have been three by-elections involving the party: Farrer where it was very obvious they were the favourites, Nepean where they were not given any real chance of winning, and this one which is on paper a lot more up for grabs. This is a key test of whether One Nation can deliver in the kinds of outer suburban seats that need to be the second string in their bow alongside rural seats if they are to break into parliaments in a big way. That said it is a by-election and a win will not necessarily guarantee that they can do it at a competitive general election.
Secret Harbour, on Perth's far southern outskirts and overlapping just into Mandurah, is a traditionally very strong Labor seat versus Liberals (it and its predecessors Peel and Warnbro have never been even close to marginal). But is also a seat where One Nation poll way above their average. In 2025 it was One Nation's sixth best seat statewide in the Assembly (with 8.43% of the vote, less than a point off their best, behind only four rural seats and adjacent Mandurah). However that is a tad misleading as they drew the top position on the ballot. It was still their ninth best Legislative Council seat (the more reliable indicator of party strength) with 6.07% of the vote compared to a state total of 3.84%.
In the just-released DemosAU state poll, One Nation is getting 21% statewide. Experience in peak elections like Queensland 1998 and SA 2026 is that One Nation surge more than uniform swing would expect in areas they were already doing well in. If a normal election were to occur right now, then based off that poll One Nation would expect something like 30% in Secret Harbour. And this is a by-election which is an opportunity to kick an incumbent government. So except perhaps for the large field size it is easy to put One Nation north of 30.
Labor for its part is coming off a 46.5% primary in 2025 but is down 8.4 points in the DemosAU poll and this would be without considering the usual by-election factors of a seat being vacant and any extra kicking of the government. On this basis it's easy to put Labor and One Nation in close enough to the same place on primaries (maybe Labor just a little ahead, on paper) and at the mercy of the size and preference flow strengths from mainly the Liberals and Greens. The Liberals shouldn't get a large vote though they should beat their 13.2% from 2021, especially after drawing the top of the ballot. The Greens polled 8.8% in 2025 and are unlikely to match that in a crowded field. Legalise Cannabis tend to do well here though it will be interesting to see how much One Nation tread on their toes, and the rest including some of Australia's most infamous serial candidates will be scrapping for detritus unless any of the indies can get anything. I'll be watching how many deposits come back, but more than five would be a surprise, it might not even be that many.
There has been no real polling for this seat, just the West Australian embarrassing itself again by pretending that 120 is a sample size. As William Bowe notes the 2025 election did not go well for their "exit polls".
Candidate Factors
This by-election is unusual in that although Labor and One Nation are the obvious frontrunners on paper both are running questionable candidates. I doubt the news that Labor's Georgia Tree is on leave from Woodside lobbying will concern more than a handful of voters in a seat infested with tradies. The fact that she lives 60 km outside the seat, and worse still in inner suburbia, despite having grown up in Warnbro might not go down as well.
On the One Nation side, gym owner Luke Herdegen has attracted ripples mainly for news that he spent most of 2016 in a bowl of cocaine, from which he took brief breaks to consort with the Albanian underworld and lose at least one multi-millionaire girlfriend, like you do. He's turned his life around but it does irk me when people support tough on crime policies when they themselves would not have been running for office had they been caught. (Not that One Nation really care about drugs that much anyway, with the exception of puberty blockers.) The media trail has gone quiet since the initial revelations rather than it turning into an embattled candidate saga and we will see if the voters care about this stuff either.
The Liberals have copped flak from Canning federal MP Andrew Hastie, whose seat overlaps most of this one, for not trying to beat One Nation. If One Nation win or at least thrash the Liberals there will be a strong sense they are coming for him (as if an endless stream of MRP projections where he loses the seat to One Nation 40-60 isn't enough foreboding yet; not everyone believes them.)
My view of this one overall is a win is a win is a win, both for Labor and One Nation. It would be nice to win by lots but unlike a lot of by-elections there is no such thing as winning by not enough. A close loss is no disaster for either side but if it's a substantial loss (46-54 or worse) then questions will have to be asked. For Labor - you threw the kitchen sink at this in spending terms, was it your candidate selection or the feds what did it? For One Nation - if you can't get close here and in a by-election (and after making very confident noises about it in the leadup) then precisely where, are you for real or is it just a bubble? And so on.
(One Nation are very confident and a Labor MP stating "We've lost this" on election eve seems to go a little beyond the usual ALP line in pessimistic expectation management. For those following betting, which is amusing but not predictive, One Nation have been plunging and as of election day 8:30 am are at 2.25 vs 1.55, implied 41% chance)
Electoral Act Sooking
One of my pet themes is that for a party that claims to not care about expressing views that often cause offence, One Nation are often a bunch of sooks. No more the case than with this complaint about a Labor attack flier where One Nation are alleging the flier is improperly authorised. There is no requirement in the WA Electoral Act that both sides of a flier be authorised. There is a requirement that both sides of a how-to-vote card be authorised, but only if both sides contain how-to-vote recommendations.
Election night
The day booth lineup is the same as in 2025 except that the Madora Bay booth has moved about 1.5 km to the north. Also the Secret Harbour EPP polling place is open on election day, which I don't think it was in 2025. Absent votes don't exist in by-elections but an odd thing with this one is that Labor did really badly on those on a 2PP basis in 2025 anyway. Voting before the election is reported as being up by thousands on the combined postal and prepoll vote from 2025 so the count for the main prepoll could be a long one. The huge number of pointless candidates will slow down counting too and it will be interesting to see if the WAEC can rebound after a 2025 election performance that ended with the then Commissioner resigning.
Tune in for an interesting count - comments from close of polls on, scrolling to the top.
