Blogging Senate forecasts and results in the WA Senate re-election until officially declared.

Twitter: @AU_Truth_Seeker

Friday, 20 September 2013

Senate update - 19-9-2013 - 10pm update.

Welcome to my latest update.

I will focus on WA and NSW. I have found a new path to the Democrats winning NSW, and present a new critical point for WA that may save Dropulich / Ludlam.

NSW: (Current count: 87.63%)
I have been raising the prospect of the Democrats winning the final senate seat in NSW. Yesterday, I described the likelihood as "unlikely".

Today I present a couple of alternative paths for the Democrats winning. To be honest, it's still unlikely, just slightly less so... Here are the critical points for DEM to win:

1: Bullet vs FF: For DEM to win in this scenario, Bullet must overturn a 119 vote margin (or 0.0039%) and overtake FF
2: Dem vs Animal Justice: DEM is coming last at this count, they need to make a gain of 260 votes (0.0064%) to survive
3: The very next count is HEMP vs AMEP - with a margin of 65 votes (0.0016%) to overcome - call this 50% likely
4: Dem vs SXP at count 40. The difference here is 0.02% for DEM to make up. This is approx 800 votes, and may be too much, but I would contend DEM does

Alternatively, there is another path to Critical point 4 above:
1: As per step 1 above
2: The very next count requires No Carbon Tax (NCT) to gain 0.0103% (417 votes)
Strangely, this will now position things identically to Critical Point 4 above, where:
3: As per step 4 above.

I will monitor this Bullet vs FF vote in the first count - a back of the envelope calc shows this margin likely to decrease with additional counting.

WA: (84.54% counted)
I have previously outlined the potential vulnerability of Sports Party candidate, Wayne Dropulich, at very early eliminations - count 9 and 10 (or possibly count 9 and 11, depending on the precise early order of other parties' eliminations). However, Sports has polled very well in Declaration votes, giving it a bit of a lead at these early eliminations, and I would find it unlikely if this momentum reverses.

Christians vs Shooters
It has been pointed out there is another moment of vulnerability for Dropulich - the Christians/Shooters&Fishers elimination (Count 21). Currently, the Shooters are looking at elimination, but if this margin of 183 votes can be reversed, including potential BTLs, then elected would be ALP's Pratt and PUP's Wang, instead of Dropulich and the GRN's Ludlam.

Animal Justice vs HEMP
BUT, although the Christian vote is declining and the Shooters vote is increasing with late counting, another critical elimination may come to the rescue of Ludlum/Dropulich. Two counts earlier, count 19, HEMP is leading Animal Justice Party (AJP) by 117 votes. If this is reversed, then strangely this delays the Christian vs the Shooters battle (I love the visual image created by this terminology!) by a count such that HEMP is eliminated, hence making Christians vs Shooters a "dead rubber" and giving victory to Ludlam/Dropulich.

Summary (mega geek alert here!):
IF Sport eliminated in counts 9-11 THEN ALP&PUP
    IF Animal Justice > HEMP in count 19 THEN GRN & SPORT
        IF Shooters > Christians in count 21 THEN ALP & PUP
            ELSE GRN & SPORT
        END IF
    END IF
If I was up to the Kevin Bonham standards I'd do a flow chart, but I'm just not that cool :-)

Probably looking like the most juicily unpredictable state, but I note Kevin Bonham has it covered. His take on the critical PUP-LDP margin is 1569. The difference between the candidates BTL votes is 1501 (according to Antony Green) so I will assume the margin here is 68 votes.So, 3000 BTLs to overturn a potential 68 vote margin...

Does anyone have a coin I can flip?

I would suspect Jacqui Lambie (PUP) remains slight favourite if I had to make a wager on this one.


  1. Slight correction - Antony has the BTL difference between PUP and LDP at 1401.

    Antony is, however, treating just 50% of #1s for a second Senate candidate as locked-ins and the remaining 50% as BTLs. Looking at the 2007 and 2010 Tas distributions there was a very strong corellation between % of preferences retained to the group for a second or subsequent candidate, and the second or subsequent candidate's vote share. (I'm thinking a basis for this is that the fewer BTLs a candidate gets the more likely their BTLs are from friends voting for them but not their party, or else from strategic voters.) If that holds, the second LDP candidate might well leak half their 40 votes but the second PUP candidate shouldn't leak more than 20-ish % of their 276, if that. So I think 1401 is unflattering to PUP's likely BTL position.

    There are also all the complexities concerning what impact is had on the count when a BTL leaks, depending on who it leaks from, when, and where it goes. It's so messy!

    PUP are holding the calculator gap better than I thought. At one stage it looked like blowing to over 2000, now it may not.


    The ungouped One nation candidate could play a major role in the outcome. Without access to the BTL preference data file it is impossible to ascertain with confidence the true extent of likelihood of a Sports party early elimination. If Sports party is defeated then the interplay of situation in the way the Senate vote is counted comes into play.

    Issues related with the calculation of the surplus transfer value and the segmentation distribution of excluded candidates distort the proportionality of the count

    A preferred option is to adopt a reiterative count where weighted surplus only distributions are undertaken in each iteration.

    If the number of vacant positions are not filled in a single iteration then candidates with the lowest total value is excluded from the count and the count is rest and restarted. A single transaction per candidate no segmentation.

    Votes from excluded candidates are redistributed as if the excluded candidates had not stood. A full value vote would also be attributed at the next available candidate continuing in the count as if it was a primary preference.

    In a reiterative count the quota would be determined and adjusted on each iteration taking into account any optional preferential votes that would exhaust without value. A vote if need be would form part of a candidates surplus transfer which in turn would be distributed within the iteration.

    We can see the impact of segmentation distribution distortion by recounting the 2007 Queensland Senate vote by excluding all candidates except the last seven standing, (3 ALP, 3 LNP and 1 Grn) and redistributing all votes as if the excluded candidates had not stood

    The exclusion of all candidates except the last seven is chosen to simulate the reiterative exclusion process which would end up with the last seven candidates remaining in the count in the final iteration. This is just a quick way of getting to the point when the elections will due decided.

    Because it is a reiterative count segmentation distribution of the vote is not required. All excluded candidates votes are redistributed in a single transaction.

    This eliminated and removes the distortions that occurs with the segmentation distribution where a vote jumps candidates and is allocated at a higher value then would normally be the case, the method of redistribution reflects the voters intentions.

    An excluded candidates voter's "second choice" preference is always distributed at full value to the next available continuing candidate and if need be forms part of that candidates surplus transfer in any given iteration

    The system of segmentation and the calculation of the surplus transfer value does impact on the outcome of the election. The system that is currently in place was designed to facilitate a manual count and shorten the manual steps that were required.

    The system of segmentation was designed to limit the flaws in the way the surplus transfer value is calculated

    With the use of computer aided distribution we no longer need to retain the flaws in the way the vote is counted.

    Two wrongs do not make a right

    A reiterative single weighted transfer count accurately reflects the votes intentions and is a linear process that can be readily followed.‎

  3. You forgot to mention the cake!

  4. Bullet Train taken over FF in NSW. DEM now 230 behind Animal Justice at next count.

  5. The Droop quota. (x/(y+1))

    It is a stand alone issue that we should be looking at also.

    Why divide the cake by seven and throw away a slice when we could just divided it by six.

    Pure proportional representation. x/y not x/(y+1)

    The Droop quota in itself distorts the proportionality of the system. It is claimed that it provides majority mandate majority representation but that is a myth also, Yes in a pure proportional system we should not be using Droop

    The Droop quota was another add on that was implemented to assist in the manual counting process as a point could be reached in the count without having to distributed all ballot papers.

    Now we have computer based distribution we really need to review the system and strip back the process of counting to the basics and much it more accurate without distortions.

  6. Sex Party now nearly 500 up on calculator in Tas. Taking their chances a bit more seriously now - there are reasons to think a few hundred more than that is capable of doing it.