Battle for the Beehive(A New Zealand Election Thread)

[Nobel Hobos 2]

Some commentator said "Adern rides another disaster to victory" which is plausible. Even when she didn't manage well (Christchurch earthquake?) she got the left equivalent of 'war leader' support, now she managed Covid-19 to a world standard and was doubly rewarded. Some people got behind her because she was the leader when (to be honest) the whole country did a great job. Plus a few who don't feel that but grudgingly admit it feels good to be the first in the world again.

Adern is nice, but more than that. She seems bright and sincere at the same time: that's a hard combination for a politician. Our PM Turnbull was like that at first, but he couldn't sustain it when forced to announce policies he plainly didn't agree with. I liked him so much I nearly voted Liberal, but I could tell he wouldn't last being so far left of his party. Anyway, this is about you lot not us lot, so I'll put this on the table:

We get Adern and you get ScoMo ... wait, I'm not finished ... any 3 Liberals and 1 Labor MP or Senator. Except Penny Wong, you can't have her. We'll also take Winston Peters off your hands. How about it?

One day you'll get yourselves a leader who won't drag your country farther to the far right. Hopefully it's just the Great Barrier Reef that's gone wrecked by Aussie mismanagement.

Horrible to say, ScoMo is pretty average for a Liberal. But this is the NZ thread.

[Forsher]

Congrats to Labour for winning an outright majority. In proportional representation electoral system too.

Now if only British Labour could accomplish the same feat, that would be hella nice.

You did. He was called Tony Blair. Ardern is a Blairite. Literally. She used to work for him.

The problem you have is that MMP means failed candidates like Ardern can build experience and remain waiting in the wings, within the public eye but out of public scrutiny.

Then there are the specific problems of Auckland Central. The electorate is quite unusual in the sense that it consists of the CBD, student accommodation and lifestyle blocks on an island. This is a potent combination for the media because it's not necessarily clear which major party it favours. I'm sure the UK has plenty of "exciting" electorates like this but, again, losing didn't mean Ardern was out on her arse. It just meant she didn't really have to do anything but could also be in parliament and gain experience.

Now there's the personal factors. First, here's some... interesting... coverage decisions from the second Kaye vs Ardern battle that put Ardern on the map. Note, also, the way the coverage plays up her relative youth. And, sure, that's fair... candidates in their late twenties and early 30s are young for politicians. But the brilliant thing is that the media are slow to update their perceptions. And thus you get pushing 40, career politician Jacinda Ardern with a youth framing in 2017. Also, Ardern is very, very good at making people ignore inconvenient facts about her... like the immigration comparison with Trump or, of course, [ur]=https://www.stuff.co.nz/national/politics/96123508/jacinda-ardern-i-didnt-want-to-work-for-tony-blairworking with Tony Blair[/url]... well, under his auspices, really.

Add in some extremely negative campaigning, John Key's "smile and wave" style of premiership and the weird personal profiles of Phil Goff, the Davids and Andrew Little, what you'll get is the main question running into the election being "Why isn't Jacinda Ardern the leader?" No joke, this was a question put to a fairly senior Labour MP by a the MC of a multi-party debate in early 2017.

So... now comes the penultimate trick: the Illusion of Change. Ardern changed absolutely nothing about Labour's campaign in 2017 in terms of promises but its fortunes altered enormously. And part of that is probably that people weren't sure if Ardern would get rid of controversial policies like Labour's extremely anti-immigration stance because Ardern was a political blank slate. She'd been in parliament for years but because she was a List MP and a member of the opposition, she didn't have a clear political record. Of course, Ardern's pre-parliamentary political career allowed her to be a Blairite to one audience and also point at her socialist credentials for a different audience. And she can always come back to that, no matter how Blairite Labour's been.

But the final trick is the best one... get lucky. Fail to win enough seats to form a coalition with the Greens, so you have to take on board NZ First. Who you can then blame for the failure to deliver pretty much anything from the transformational vision from the election. Then, have a terrorist attack... that's straight out of the Bush playlist... where all you have to do is just not fuck it up... which is where Bush went wrong. Then, at the point where people are maybe getting tired of your shit... have a nice global pandemic. I don't think it's a coincidence that we shut up shop so early when you remember that pesky little comparison to Trump on immigration...

Ardern is, in short, Hugh Grant from Love, Actually.

[Shrillland]

Congrats to Labour for winning an outright majority. In proportional representation electoral system too.

Now if only British Labour could accomplish the same feat, that would be hella nice.

You did. He was called Tony Blair. Ardern is a Blairite. Literally. She used to work for him.

The problem you have is that MMP means failed candidates like Ardern can build experience and remain waiting in the wings, within the public eye but out of public scrutiny.

Then there are the specific problems of Auckland Central. The electorate is quite unusual in the sense that it consists of the CBD, student accommodation and lifestyle blocks on an island. This is a potent combination for the media because it's not necessarily clear which major party it favours. I'm sure the UK has plenty of "exciting" electorates like this but, again, losing didn't mean Ardern was out on her arse. It just meant she didn't really have to do anything but could also be in parliament and gain experience.

Now there's the personal factors. First, here's some... interesting... coverage decisions from the second Kaye vs Ardern battle that put Ardern on the map. Note, also, the way the coverage plays up her relative youth. And, sure, that's fair... candidates in their late twenties and early 30s are young for politicians. But the brilliant thing is that the media are slow to update their perceptions. And thus you get pushing 40, career politician Jacinda Ardern with a youth framing in 2017. Also, Ardern is very, very good at making people ignore inconvenient facts about her... like the immigration comparison with Trump or, of course, [ur]=https://www.stuff.co.nz/national/politics/96123508/jacinda-ardern-i-didnt-want-to-work-for-tony-blairworking with Tony Blair[/url]... well, under his auspices, really.

Add in some extremely negative campaigning, John Key's "smile and wave" style of premiership and the weird personal profiles of Phil Goff, the Davids and Andrew Little, what you'll get is the main question running into the election being "Why isn't Jacinda Ardern the leader?" No joke, this was a question put to a fairly senior Labour MP by a the MC of a multi-party debate in early 2017.

So... now comes the penultimate trick: the Illusion of Change. Ardern changed absolutely nothing about Labour's campaign in 2017 in terms of promises but its fortunes altered enormously. And part of that is probably that people weren't sure if Ardern would get rid of controversial policies like Labour's extremely anti-immigration stance because Ardern was a political blank slate. She'd been in parliament for years but because she was a List MP and a member of the opposition, she didn't have a clear political record. Of course, Ardern's pre-parliamentary political career allowed her to be a Blairite to one audience and also point at her socialist credentials for a different audience. And she can always come back to that, no matter how Blairite Labour's been.

But the final trick is the best one... get lucky. Fail to win enough seats to form a coalition with the Greens, so you have to take on board NZ First. Who you can then blame for the failure to deliver pretty much anything from the transformational vision from the election. Then, have a terrorist attack... that's straight out of the Bush playlist... where all you have to do is just not fuck it up... which is where Bush went wrong. Then, at the point where people are maybe getting tired of your shit... have a nice global pandemic. I don't think it's a coincidence that we shut up shop so early when you remember that pesky little comparison to Trump on immigration...

Ardern is, in short, Hugh Grant from Love, Actually.

That last one's true enough. Ardern's been perhaps the luckiest head of government in the entire developed world even as-and partially because-NZ itself has seen disaster after disaster from the earthquake to the mosque attack to the volcano to Covid. When she didn't have a massive crisis on her hands, her actual governing has so far been somewhat meh, to be honest.

[Forsher]

Congrats to Labour for winning an outright majority. In proportional representation electoral system too.

Only sort of. They still have single-member constituencies (like UK parliament) but there are extra members elected to make it more proportional.

That Labour seems set to take the majority of seats despite not having an outright majority of the vote, shows that it isn't perfectly proportional.

I don't think the result is final yet, btw.

That's down almost entirely to the historically high proportion of wasted vote created by our absurd threshold.

In 2002 when National got hammered even harder than this year, the wasted vote was only 4.89% but (at the moment) it's 7.7%.

In 2002 to get a majority in parliament you needed something in the region of: 47.56%, e.g. here

In 2020 to get a majority in parliament you need something like 46.15%, e.g. here.

The obvious thing to do seems to be to rank party votes and just keep reallocating until everyone's vote counts in some fashion...

Of course, you might say "lower the threshold" is a better solution... 1/120 = 0.83, so even 2.5% (half the current setting, half!) means a party will have around 3 MPs, and the approximate majority percentage might* be more like 48.75%, but the reality is people appear to whine about tiny parties too much to get any movement on this. A preference based reallocation of votes (with the current 5% threshold) would ensure parties have to have at least 5% of the post reallocation vote (barring electorates) to be in parliament, which is 6 MPs... and surely above whatever number people say makes a party "too small".

The threshold should be 1%.

*On average, the elections have returned wasted vote proportions (including the preliminary 7.7% of this one) of about 5.3% so I imagine a lower threshold would again cause a tendency for the wasted vote to bounce around the size of the threshold. There is, of course, no reason why we cannot have an election where no party manages to clear the threshold (meaning wasted vote % is 100%), whatever we set the threshold to... provided there are enough parties... it's just never going to happen in practice.

[Forsher]

That last one's true enough. Ardern's been perhaps the luckiest head of government in the entire developed world even as-and partially because-NZ itself has seen disaster after disaster from the earthquake to the mosque attack to the volcano to Covid. When she didn't have a massive crisis on her hands, her actual governing has so far been somewhat meh, to be honest.

This was a Covid election because if it was on any other issue, this Labour's record is against it.

[New Rogernomics]

After seeing Ardern's speech, I think a few Greens will be disappointed with how she's going to stay closer to the centre, especially now that she's all set to govern alone without any help.

The Greens tend to do confidence and supply with Labour, and as much as Labour might want to govern alone, I very much doubt Ardern will ignore the Greens due to the fact that it is a majority (not heavy super majority), and for controversial votes that annoy the neo-liberals in Labour she'll need Green votes.

[Angermanland]

That last one's true enough. Ardern's been perhaps the luckiest head of government in the entire developed world even as-and partially because-NZ itself has seen disaster after disaster from the earthquake to the mosque attack to the volcano to Covid. When she didn't have a massive crisis on her hands, her actual governing has so far been somewhat meh, to be honest.

This was a Covid election because if it was on any other issue, this Labour's record is against it.

Well, that and the whole 'people like not dying' thing.

Keep in mind, National really didn't have a coherent position on anything beyond "not what Labour's doing" and "Judith Colins is pretty cool, you should vote for us so she gets to be PM". (seriously, their own electoral candidates had no idea what positions they were supposed to be presenting.)

And ACT keeps banging on about the debt with an argument that only makes a scrap of sense if you subscribe to economic theories that were essentially discredited in those circles that actually care about reality Decades ago. They were pretty much trying to sell us on Austerity and (if I recall correctly) fully opening the border as the solution to an utter non-problem. Nevermind that Austerity consistently achieves nothing but the destruction of the nation's ability to actually pay its debts at all <_<

[Forsher]

That's down almost entirely to the historically high proportion of wasted vote created by our absurd threshold.

I have some better* numbers for the preliminary results now. And some really lazy tables.

So, three main things to pay attention to here. On the far left we've got the hypothetical threshold, in "wasted_v" we have the wasted vote (i.e. the votes that are made and quite literally discarded) and in "disproportionality" we have a sum of squares measure of the difference between the % of votes and the % of seats won. Actually... I'll show you how that works with another table:

Parties which have a "negative difference" are OVER REPRESENTED in the preliminary parliament (i.e. that parliament calculated from the preliminary results). We square the differences for two reasons. One, to make everything positive so we can more easily summarise (literally by using sums) all the information (remember, the direction of failed representation doesn't really matter**). Two, to reflect that bigger numbers are worse than others (highlighted)... technically we should be thinking about loss functions in here but I haven't done that. And then, of course, we sum everything up... I've used a cumulative sum so the final number we care about is actually in the bottom right corner... which is why that's highlighted.

It's important to note that over-hangs (which might happen if there's a big swing in one of the close Maori seats and the Maori Party also doesn't gain many party votes based on the specials) are probably actually worse for representation than the threshold effects. I don't want to understate that the badness of the threshold effects, however (so, yeah, loss functions... not a thing to just gloss over, really).

Anyway, so here's the preliminary results in one of my cheapo tables:

You might wonder how the Population Share is so low... it seems disturbing that we can have a party with more than 53% of the seats in Parliament that represents barely a third of the enrolled voters, right? Well, that's because the preliminary results have the same denominator as the final results when it comes to turnout (about 3.4 million) but about 480,000 fewer votes (this year), meaning turnout is effectively 68.35% at the moment (not 82.5%).

But here's the interesting question... what would parliament look like without a threshold? Now, I tried to find a calculator to do this (the Electoral Commission's is great but it doesn't allow one to change the threshold... maybe because it'd show what a crock of shit their recommendation of 4% is!!) so I had to "make" one. And, well, I've had about six different versions of it because I keep finding BIG problems with my work. Even the one I'm about to show you tries to give the Greens an extra seat relative to the Commission's Calculator in this insane hypothetical (133 seat parliament!) and I'm not sure why. I think it comes down to how ties are handled but I couldn't find a better explanation of how we actually implement the Sainte-Laguë Method than Wikipedia . So, I fudged it and just have the calculator give the final iterative seat to the largest tied party in the case of a tied quotient (rather than the smallest, which is its behaviour... assuming no more screw ups... elsewhere). It's not great R code, but you can have a look at my calculator for yourselves:

Code: Select all


mmp.ties = function(parties, votes, electorates, threshold, parliament, enrolled = NA, detail.req = FALSE, turnout = 1){
  #I should really allow this to take a data argument, but that's not something I remember off the top of my head and I've been doing this for practically 24 hours now
 
  alice = hatter = votes / sum(votes)
  wonderland = electorates

  alice[alice < threshold & wonderland == 0] = 0
  #attenuates parties with NEITHER an electorate or enough votes to clear the threshold
  hatter[hatter < threshold] = 0
  #attenuates the parties below the threshold only
 
  counted = sum(alice)
  #proportion of votes actually counted
  wasted = 1 - counted
 
  #Setting up the Iterative Process
 
  columns = (length(parties))
 
  current_seats = quotients = matrix(0, nrow = parliament, ncol = columns) #matrix of zeroes
    colnames(current_seats) = parties   
    colnames(quotients) = parties   
    #I needed the names to be here because I use the names to resolve a duplicate maximum problem and if they're not before the actual iterations start it screws up
 
  i = 1
    qs = alice / (2 * current_seats[i, ] + 1) #the all important quotient
    quotients[i, ] = qs #which I will now be saving for investigation at the end
   
    random.initial = sample(which(quotients[i, ] == max(quotients[i, ], na.rm = TRUE)), 1)
    #in theory, if the initial seat is a tie, this allow it to be resolved... maybe I should've (23:56 18/10/20) tested this one
   
    index = ifelse(sum(quotients[i, ] - quotients[i, which.max(quotients[i, ])] == 0) > 1, random.initial, which.max(qs))
    #choosing the index (i.e. the party) that gets the seat, with an ifelse() to handle ties
 
    current_seats[i, index] = current_seats[i, index] + 1
    #this is why I have i=1 out of the loop
   
 
  for(i in 2:parliament){
    qs = alice / (2 * current_seats[i - 1, ] + 1) #current is possibly misleading, it's more "immediately prior" seats but it IS current from the perspective of "how many do they have right now" (and now is defined as after i=1 but before i=2, which hasn't happened yet)
    quotients[i, ] = qs
   
    drawn.loser = ifelse(i < parliament,
                         names(which.min(current_seats[i - 1, which(quotients[i, ] == max(quotients[i, ], na.rm = TRUE))])),
                         names(which.max(current_seats[i - 1, which(quotients[i, ] == max(quotients[i, ], na.rm = TRUE))])))
    tied.loser = which(parties == (drawn.loser))
    #handling BOTH the which.max() duplicate problem AND a discrepancy between my numbers and the Electoral Commission's
    #the discrepancy arose with the final seat... my minimum seats rule would have given it over to the Greens whereas they give it to National and I can't see why
    #giving the larger party the final seat probably makes it more likely that a party with more than 50% actually clears that in terms of seats won, so maybe it's reasonable
    #ad hoc rationalisations ftw! (!!!)
    #I think what the Commission's done is generate the quotients for every party over the threshold or with an electorate seat under the scenarios of everything between 0 and 122 previous seats; I'll have a look at that in a minute
   
    index = ifelse(sum(quotients[i, ] - quotients[i, which.max(quotients[i, ])] == 0) > 1, tied.loser, which.max(qs))
    #subtly different to the i=1 case because i>1 is different
   
    current_seats[i, index] = current_seats[i, index] + 1
    #the seat the party that just won, wins
    current_seats[i, ] = current_seats[i,] + current_seats[i - 1,]
    #I guess I could sum these all at the end, but this way you see the whole count each session... I think that's easier to understand/read
    quotients[i, ] = qs
    #probably superfluous but I've been doing this too long to want to confirm that
  }
  #Loops are inefficient but easy
   
  seats = current_seats[120, ] #this would be where the colSums happen
  seats[current_seats[120,] < electorates] = electorates[current_seats[120,] < electorates]
  #parties with more electorate seats than iterative seats cause an overhang
  #this sets those parties to their electorate count whilst leaving everyone else with their iterative count
 
  list_seats = seats - electorates #obvious
 
  vote_pcnt = round((votes / sum(votes)) * 100, 2) #normally I prefer decimals, but percentages were cleaner in the table, I thought, if I wanted more precision (2dp % = 4dp decimal)
 
  MP_pcnt = round((seats / sum(seats)) * 100, 2)
  vote_eff_pcnt = round((alice / sum(alice)) * 100, 2) #the all important effective share
 
  decimal.form = votes / sum(votes) #for convenience, should have this further up... not changing in case I introduce another mistake
 

  pop_pcnt = if(is.na(enrolled)){
    pop.pcnt = "-"
    #print(pop.pcnt)
  }else{
    if(sum(votes) == 1){
      voting.pop = turnout * 1000
      standardised.voters = votes * voting.pop
      pop.pcnt = standardised.voters / 1000
      #print(pop.pcnt)
    }else{
      pop.pcnt = votes / enrolled
      #print(pop.pcnt)
    }
  }
  #because I was reading https://www.interest.co.nz/opinion/107554/economist-brian-easton-says-mmp-may-deliver-parliament-which-reflects-us-frequently
 
  results = data.frame(parties, electorates, list_seats, seats, MP_pcnt, vote_pcnt, vote_eff_pcnt, pop_pcnt = round(pop_pcnt * 100, 2)) 
 
  blackbox = data.frame(parties, vote_share = alice, share.ovr.thresh = hatter, iterated_seats = current_seats[120,])
  #for primitive diagnosis/investigation of the iterative outcome
 
  proportionality = data.frame(parties, vote_pcnt, MP_pcnt,
                               diff = vote_pcnt - MP_pcnt,
                               crude_seats = round(decimal.form * parliament, 0),
                               actual = seats,
                               ca_diff = round(decimal.form * parliament, 0) - seats,
                               cr_cmltv = cumsum(round(decimal.form * parliament, 0)),
                               cr_incl_v = cumsum(alice)
                               )
  #more detailed investigation
 
    size = data.frame(overhang = sum(seats) - parliament, theoretical_s = parliament, actual_s = sum(seats), counted_v = sum(alice), wasted_v = 1 - sum(alice), disproportionality = sum( (proportionality$diff )^2 ))
    #quick summary of the quirks of the system
    #now has a sum of squares measure of disproportionality... squared difference of vote and MP share.
 
  rownames(results) <- c()
  rownames(blackbox) <- c()
  rownames(proportionality) <- c()
  #tidier presentation with my cheapo tables
 
  #Returning Useful Stuff
  #and ensuring that more useful stuff is available on demand
 
  details = list(iterative_seats = current_seats, iterative_quotients = quotients,
       proportionality = proportionality)
 
  output = list(blackbox = blackbox[order(results$seats, decreasing = TRUE), ],
       results = results[order(results$seats, decreasing = TRUE), ],
       size = size)
 
  outcome.object = list(details = details, output = output)
   
  ifelse(detail.req == FALSE, return(output),  return(outcome.object))
    #must be return() not print()... at least for return(output)
}



(Don't worry, I did go to sleep...)

So, without further introductions... THE NO THRESHOLD PARLIAMENT:

We'd have four extra parties (NZ First, New Conservatives, TOP and Advance) with half of their collective eight seats coming off Labour's actual preliminary total and three from National... which should drive home just how incentivised the big parties are to NOT change the threshold (even if the Electoral Commission's 2012 report's recommended 4% threshold wouldn't change anything in this election... before the specials are considered).

There is no world in which this parliament is worse than the one we're likely to have. Unless you're so perfectly aligned with this piss-poor Labour government that what you really want is for it to have an even bigger outright majority. Even those of you who wouldn't want a Labour/Greens government benefit from this because now Labour can temper the Greens' negotiations by threatening to sidle up to NZ First, TOP and Maori (any one of them would be sufficient to hit 61 seats).

* I sort of already explained why "better" needs qualification with the Calculator coding discussion.

** At least, if the parties are in parliament. Being under-represented by 1 seat if you've got no seats is worse than if you're on 2 seats with the missing seat. And I'm not sure if having, say, The Aotearoa Legalise Cannabis Party in parliament with 1 MP is actually just as bad as having the Greens have one more seat than they should have, or not. I'm kind of tempted to say it's okay to have parties that shouldn't be in parliament so long as their numbers are tiny...

[Nobel Hobos 2]

Well that's the first time I've seen the Code box used on General, for actual code.
I give the above post 4 Nerds:

[Forsher]

Well that's the first time I've seen the Code box used on General, for actual code.
I give the above post 4 Nerds:

If you ignore the turnouts, assume US presidential vote shares are analogous to party votes and assume a 725 seat (72/120 = 06 => 435/.6 = 725) US House of Representatives with the same 435 electorate seats then... the 2016 US election with MMP for their House of Representatives would've been vastly more representative than our one just now: lower wasted vote (0.05729842) and lower disproportionality (28.9876).

[Angermanland]

That's down almost entirely to the historically high proportion of wasted vote created by our absurd threshold.

I have some better* numbers for the preliminary results now. And some really lazy tables.

So, three main things to pay attention to here. On the far left we've got the hypothetical threshold, in "wasted_v" we have the wasted vote (i.e. the votes that are made and quite literally discarded) and in "disproportionality" we have a sum of squares measure of the difference between the % of votes and the % of seats won. Actually... I'll show you how that works with another table:

Parties which have a "negative difference" are OVER REPRESENTED in the preliminary parliament (i.e. that parliament calculated from the preliminary results). We square the differences for two reasons. One, to make everything positive so we can more easily summarise (literally by using sums) all the information (remember, the direction of failed representation doesn't really matter**). Two, to reflect that bigger numbers are worse than others (highlighted)... technically we should be thinking about loss functions in here but I haven't done that. And then, of course, we sum everything up... I've used a cumulative sum so the final number we care about is actually in the bottom right corner... which is why that's highlighted.

It's important to note that over-hangs (which might happen if there's a big swing in one of the close Maori seats and the Maori Party also doesn't gain many party votes based on the specials) are probably actually worse for representation than the threshold effects. I don't want to understate that the badness of the threshold effects, however (so, yeah, loss functions... not a thing to just gloss over, really).

Anyway, so here's the preliminary results in one of my cheapo tables:

You might wonder how the Population Share is so low... it seems disturbing that we can have a party with more than 53% of the seats in Parliament that represents barely a third of the enrolled voters, right? Well, that's because the preliminary results have the same denominator as the final results when it comes to turnout (about 3.4 million) but about 480,000 fewer votes (this year), meaning turnout is effectively 68.35% at the moment (not 82.5%).

But here's the interesting question... what would parliament look like without a threshold? Now, I tried to find a calculator to do this (the Electoral Commission's is great but it doesn't allow one to change the threshold... maybe because it'd show what a crock of shit their recommendation of 4% is!!) so I had to "make" one. And, well, I've had about six different versions of it because I keep finding BIG problems with my work. Even the one I'm about to show you tries to give the Greens an extra seat relative to the Commission's Calculator in this insane hypothetical (133 seat parliament!) and I'm not sure why. I think it comes down to how ties are handled but I couldn't find a better explanation of how we actually implement the Sainte-Laguë Method than Wikipedia . So, I fudged it and just have the calculator give the final iterative seat to the largest tied party in the case of a tied quotient (rather than the smallest, which is its behaviour... assuming no more screw ups... elsewhere). It's not great R code, but you can have a look at my calculator for yourselves:

Code: Select all


mmp.ties = function(parties, votes, electorates, threshold, parliament, enrolled = NA, detail.req = FALSE, turnout = 1){
  #I should really allow this to take a data argument, but that's not something I remember off the top of my head and I've been doing this for practically 24 hours now
 
  alice = hatter = votes / sum(votes)
  wonderland = electorates

  alice[alice < threshold & wonderland == 0] = 0
  #attenuates parties with NEITHER an electorate or enough votes to clear the threshold
  hatter[hatter < threshold] = 0
  #attenuates the parties below the threshold only
 
  counted = sum(alice)
  #proportion of votes actually counted
  wasted = 1 - counted
 
  #Setting up the Iterative Process
 
  columns = (length(parties))
 
  current_seats = quotients = matrix(0, nrow = parliament, ncol = columns) #matrix of zeroes
    colnames(current_seats) = parties   
    colnames(quotients) = parties   
    #I needed the names to be here because I use the names to resolve a duplicate maximum problem and if they're not before the actual iterations start it screws up
 
  i = 1
    qs = alice / (2 * current_seats[i, ] + 1) #the all important quotient
    quotients[i, ] = qs #which I will now be saving for investigation at the end
   
    random.initial = sample(which(quotients[i, ] == max(quotients[i, ], na.rm = TRUE)), 1)
    #in theory, if the initial seat is a tie, this allow it to be resolved... maybe I should've (23:56 18/10/20) tested this one
   
    index = ifelse(sum(quotients[i, ] - quotients[i, which.max(quotients[i, ])] == 0) > 1, random.initial, which.max(qs))
    #choosing the index (i.e. the party) that gets the seat, with an ifelse() to handle ties
 
    current_seats[i, index] = current_seats[i, index] + 1
    #this is why I have i=1 out of the loop
   
 
  for(i in 2:parliament){
    qs = alice / (2 * current_seats[i - 1, ] + 1) #current is possibly misleading, it's more "immediately prior" seats but it IS current from the perspective of "how many do they have right now" (and now is defined as after i=1 but before i=2, which hasn't happened yet)
    quotients[i, ] = qs
   
    drawn.loser = ifelse(i < parliament,
                         names(which.min(current_seats[i - 1, which(quotients[i, ] == max(quotients[i, ], na.rm = TRUE))])),
                         names(which.max(current_seats[i - 1, which(quotients[i, ] == max(quotients[i, ], na.rm = TRUE))])))
    tied.loser = which(parties == (drawn.loser))
    #handling BOTH the which.max() duplicate problem AND a discrepancy between my numbers and the Electoral Commission's
    #the discrepancy arose with the final seat... my minimum seats rule would have given it over to the Greens whereas they give it to National and I can't see why
    #giving the larger party the final seat probably makes it more likely that a party with more than 50% actually clears that in terms of seats won, so maybe it's reasonable
    #ad hoc rationalisations ftw! (!!!)
    #I think what the Commission's done is generate the quotients for every party over the threshold or with an electorate seat under the scenarios of everything between 0 and 122 previous seats; I'll have a look at that in a minute
   
    index = ifelse(sum(quotients[i, ] - quotients[i, which.max(quotients[i, ])] == 0) > 1, tied.loser, which.max(qs))
    #subtly different to the i=1 case because i>1 is different
   
    current_seats[i, index] = current_seats[i, index] + 1
    #the seat the party that just won, wins
    current_seats[i, ] = current_seats[i,] + current_seats[i - 1,]
    #I guess I could sum these all at the end, but this way you see the whole count each session... I think that's easier to understand/read
    quotients[i, ] = qs
    #probably superfluous but I've been doing this too long to want to confirm that
  }
  #Loops are inefficient but easy
   
  seats = current_seats[120, ] #this would be where the colSums happen
  seats[current_seats[120,] < electorates] = electorates[current_seats[120,] < electorates]
  #parties with more electorate seats than iterative seats cause an overhang
  #this sets those parties to their electorate count whilst leaving everyone else with their iterative count
 
  list_seats = seats - electorates #obvious
 
  vote_pcnt = round((votes / sum(votes)) * 100, 2) #normally I prefer decimals, but percentages were cleaner in the table, I thought, if I wanted more precision (2dp % = 4dp decimal)
 
  MP_pcnt = round((seats / sum(seats)) * 100, 2)
  vote_eff_pcnt = round((alice / sum(alice)) * 100, 2) #the all important effective share
 
  decimal.form = votes / sum(votes) #for convenience, should have this further up... not changing in case I introduce another mistake
 

  pop_pcnt = if(is.na(enrolled)){
    pop.pcnt = "-"
    #print(pop.pcnt)
  }else{
    if(sum(votes) == 1){
      voting.pop = turnout * 1000
      standardised.voters = votes * voting.pop
      pop.pcnt = standardised.voters / 1000
      #print(pop.pcnt)
    }else{
      pop.pcnt = votes / enrolled
      #print(pop.pcnt)
    }
  }
  #because I was reading https://www.interest.co.nz/opinion/107554/economist-brian-easton-says-mmp-may-deliver-parliament-which-reflects-us-frequently
 
  results = data.frame(parties, electorates, list_seats, seats, MP_pcnt, vote_pcnt, vote_eff_pcnt, pop_pcnt = round(pop_pcnt * 100, 2)) 
 
  blackbox = data.frame(parties, vote_share = alice, share.ovr.thresh = hatter, iterated_seats = current_seats[120,])
  #for primitive diagnosis/investigation of the iterative outcome
 
  proportionality = data.frame(parties, vote_pcnt, MP_pcnt,
                               diff = vote_pcnt - MP_pcnt,
                               crude_seats = round(decimal.form * parliament, 0),
                               actual = seats,
                               ca_diff = round(decimal.form * parliament, 0) - seats,
                               cr_cmltv = cumsum(round(decimal.form * parliament, 0)),
                               cr_incl_v = cumsum(alice)
                               )
  #more detailed investigation
 
    size = data.frame(overhang = sum(seats) - parliament, theoretical_s = parliament, actual_s = sum(seats), counted_v = sum(alice), wasted_v = 1 - sum(alice), disproportionality = sum( (proportionality$diff )^2 ))
    #quick summary of the quirks of the system
    #now has a sum of squares measure of disproportionality... squared difference of vote and MP share.
 
  rownames(results) <- c()
  rownames(blackbox) <- c()
  rownames(proportionality) <- c()
  #tidier presentation with my cheapo tables
 
  #Returning Useful Stuff
  #and ensuring that more useful stuff is available on demand
 
  details = list(iterative_seats = current_seats, iterative_quotients = quotients,
       proportionality = proportionality)
 
  output = list(blackbox = blackbox[order(results$seats, decreasing = TRUE), ],
       results = results[order(results$seats, decreasing = TRUE), ],
       size = size)
 
  outcome.object = list(details = details, output = output)
   
  ifelse(detail.req == FALSE, return(output),  return(outcome.object))
    #must be return() not print()... at least for return(output)
}



(Don't worry, I did go to sleep...)

So, without further introductions... THE NO THRESHOLD PARLIAMENT:

We'd have four extra parties (NZ First, New Conservatives, TOP and Advance) with half of their collective eight seats coming off Labour's actual preliminary total and three from National... which should drive home just how incentivised the big parties are to NOT change the threshold (even if the Electoral Commission's 2012 report's recommended 4% threshold wouldn't change anything in this election... before the specials are considered).

There is no world in which this parliament is worse than the one we're likely to have. Unless you're so perfectly aligned with this piss-poor Labour government that what you really want is for it to have an even bigger outright majority. Even those of you who wouldn't want a Labour/Greens government benefit from this because now Labour can temper the Greens' negotiations by threatening to sidle up to NZ First, TOP and Maori (any one of them would be sufficient to hit 61 seats).

* I sort of already explained why "better" needs qualification with the Calculator coding discussion.

** At least, if the parties are in parliament. Being under-represented by 1 seat if you've got no seats is worse than if you're on 2 seats with the missing seat. And I'm not sure if having, say, The Aotearoa Legalise Cannabis Party in parliament with 1 MP is actually just as bad as having the Greens have one more seat than they should have, or not. I'm kind of tempted to say it's okay to have parties that shouldn't be in parliament so long as their numbers are tiny...

You know, I used to agree with you on the whole threshold thing...
After all, it exists entirely* because it was copied from the German system, where it was included as part of their "that whole Nazi thing was a bad idea, let's not do that again" measures. Nazis aren't exactly a problem here, after all.
This election has successfully changed my mind, despite my vote being one of those wasted in this election, and the argument, while convincing, wasn't terribly complicated.
I shall summarize it:

Advance is a thing and would have actually got a seat.

*well, not quite entirely, it does make it harder for new parties to get off the ground, too, which made the whole thing more appealing to those already in power at the time than would otherwise have been the case.

[Nobel Hobos 2]

There is no world in which this parliament is worse than the one we're likely to have. Unless you're so perfectly aligned with this piss-poor Labour government that what you really want is for it to have an even bigger outright majority. Even those of you who wouldn't want a Labour/Greens government benefit from this because now Labour can temper the Greens' negotiations by threatening to sidle up to NZ First, TOP and Maori (any one of them would be sufficient to hit 61 seats).

* I sort of already explained why "better" needs qualification with the Calculator coding discussion.

** At least, if the parties are in parliament. Being under-represented by 1 seat if you've got no seats is worse than if you're on 2 seats with the missing seat. And I'm not sure if having, say, The Aotearoa Legalise Cannabis Party in parliament with 1 MP is actually just as bad as having the Greens have one more seat than they should have, or not. I'm kind of tempted to say it's okay to have parties that shouldn't be in parliament so long as their numbers are tiny...

You know, I used to agree with you on the whole threshold thing...
After all, it exists entirely because it was copied from the German system, where it was included as part of their "that whole Nazi thing was a bad idea, let's not do that again" measures. Nazis aren't exactly a problem here, after all.
This election has successfully changed my mind, despite my vote being one of those wasted in this election, and the argument, while convincing, wasn't terribly complicated.
I shall summarize it:

Advance is a thing and would have actually got a seat.

1 seat means nothing. They speak in Parliament. Everyone boos. Maybe next time they get two seats. Still nothing.

[Forsher]

1 seat means nothing. They speak in Parliament. Everyone boos. Maybe next time they get two seats. Still nothing.

This is the most frustrating thing about ant-threshold reduction people... they always fall back on "BUT TOO MANY MINOR PARTIES" or, explicitly, "I don't support X therefore it's a good thing". We've had multiple single (or otherwise tiny) parties in parliament many times. And, unless the leader goes on Dancing with Stars, they don't get anywhere.

[Angermanland]

1 seat means nothing. They speak in Parliament. Everyone boos. Maybe next time they get two seats. Still nothing.

This is the most frustrating thing about ant-threshold reduction people... they always fall back on "BUT TOO MANY MINOR PARTIES" or, explicitly, "I don't support X therefore it's a good thing". We've had multiple single (or otherwise tiny) parties in parliament many times. And, unless the leader goes on Dancing with Stars, they don't get anywhere.

Ehh, I still figure the whole thing could do with a lot of improvement, I'm just no longer of the opinion that 'just get rid of the threshold entirely' is the correct answer. Reducing it to some degree is probably reasonable. Getting rid of it entirely would require coming up with something that actually worked better at it's (supposed) main job.
For one thing, I'm fully in favour of using single transferable vote (or, less complicated but also perhaps less suitable, approval voting) for selecting MPs in electorates. Selecting for 'least dis-preferred', and thus greatly reduce the 'strategic voting' problem.
"two ticks" needs to go die in a fire. It's basically "don't think about who your electoral representative is! Only the party matters!" and is pushed hard by Labour and National pretty much entirely because it sabotages anyone else's chances of winning electorate seats to make up for missing the party vote threshold.
... We could do with a way of killing the incessant media insistence on treating the campaigns as a two party presidential race.
More parties actually getting in would be Great. particularly if part of the process of getting them was breaking up support for national and labour. Just, you know, not the sort of idiocy that results in things like the Advance party.

I would Love to see a socially conservative party (that is, a party that is more difficult to convince that a given issue is a problem that needs fixing rather than an acceptable trade off to keep other things ticking over as well as they are (of course, the more Actually broken something is, the smaller the difference is), and then requires quite a bit more evidence that the proposed solution will actually help and not have overly problematic side effects, but once convinced of those has no particular reason to oppose any given progressive idea. Or if they do it's because they're conservative And something else.) that aren't ignorant lunatics... or not-so-crypto- right wing reactionary nutjobs of some stripe.
An environmentalist party Not married to hyper-progressive (relatively speaking) social policy? Keep the lunatics out and that'd eat votes from the right quite happily (and also probably some from the Greens), if it can actually get enough momentum for people to take it seriously.
There are a bunch of other types of parties that'd be great to have.

Strangely enough, that's a group with no overlap with those who would consider Fox and various US fascists and conspiracy nuts to be sources of factual information and/or good ideas <_< (edit: For clarity, the Chinese Government is not a better source of good ideas, and is only barely a better source of factual information (which is a very low bar).)

[Forsher]

I just found an April 2018 assessment of what I wanted from the current coalition. It explains a lot about why I really loathe this result.

Ending Jacindamania: A Pivot Towards Democracy

So... not only did Jacindamania continue but it's become so normalised people barely even talk about it now. Even worse, Jacinda Ardern (and Judith Collins) both came out in favour of a four year term, which is anti-democratic.

Not Pursuing the Spirit of their Immigration Policy

This was true in the sense that neither Labour nor NZ First ultimately did all that much about their xenophobic promises. OTOH, we shut the borders basically as soon as it was obvious that it was a tool on the table and they're still closed. Indeed, one of the clear contrasts between Labour and National at the election was that National would try and find a way of getting the international students the sector relies on for funding into the country. Did Labour want a bar of that? Nope. Similarly, the travel bubble is one way even though I think some Australian states have done a better control job than we have.

(There's an exception for a couple of hundred postgraduates and PhDs, admittedly.)

Sacking Phil Twyford: Urban Policy

Ardern (and Little before her) have spent their political power protecting Phil Twyford for a long time. What's more, Ardern promoted Twyford up the list in the run up to the election. So, given the majority (that's expected to increase after the specials are counted), it seems extremely unlikely that the Auckland's Nero is going anywhere. He may even get portfolios he lost back in order to put Megan Woods in Health and put Hipkins back/exclusively (can't remember what happened) in Education (which he apparently likes).

Honestly, I think the only way Twyford was going anywhere was if he was scapegoated by "coalition negotiations" that quite simply aren't going to happen. (Sure, there could be a coalition, but the negotiating process that would have been blamed for Twyford's well earned political demise is irrelevant barring something extraordinary like Labour's Maori seats rebelling/a bunch of Labour MPs doing a Jami-Lee Ross and Ardern's not wielding the anti-Waka jumping axe.)

Get the Core Auckland Transport Projects Started



I was an idiot. None of the four (Northwestern LRT, Mangere LRT, completed electrification, third main) have gone anywhere. Mangere LRT has even been completely canned by the incompetence of our dear friend Phil.

Ardern tells us she wants a four year term so she can actually do things. Complete bullshit. This government has proven itself time and time again to be utterly useless and that's why it hasn't got anything done. NZ First didn't give a fuck about anything in Auckland once they got their provincial funds and Auckland Transport had literally done all the work for Mangere LRT before Labour was even elected... they just needed money (that was it, nothing else... but it got transferred to Twyford's personal domain and never saw the light of day before he killed it).

And fat lot of good getting anything remotely left leaning from a John Key government. Sorry, Jacinda Ardern. I find it hard to tell them apart.

[Major-Tom]

GG for Jacinda, I will concur with Shrilland though, she has been remarkably lucky in a sense. Obviously, dealing with crisis after crisis is hard for the country of NZ, and my point isn't that Jacinda is lucky to have to deal with them as well, she's been lucky that her record is based largely on her responses to that and seldom on any other policies.

Because, really, besides being a competent crisis manager, I feel that her government didn't actually do all too much. Might change with a majority.

[Outer Sparta]

GG for Jacinda, I will concur with Shrilland though, she has been remarkably lucky in a sense. Obviously, dealing with crisis after crisis is hard for the country of NZ, and my point isn't that Jacinda is lucky to have to deal with them as well, she's been lucky that her record is based largely on her responses to that and seldom on any other policies.

Because, really, besides being a competent crisis manager, I feel that her government didn't actually do all too much. Might change with a majority.

Of course it's worth noting she had the NZ First in government. A Labour majority or one with Green help would be less constrained with normal governing with no NZ First in the way.

[Shrillland]

In the last election-based news, Ardern won't be in charge of a Labour-only government after all: https://www.tvnz.co.nz/one-news/new-zealand/greens-say-best-both-worlds-after-signing-deal-labour

It will be a Labour-Greens Coalition.

[Costa Fierro]

In the last election-based news, Ardern won't be in charge of a Labour-only government after all: https://www.tvnz.co.nz/one-news/new-zealand/greens-say-best-both-worlds-after-signing-deal-labour

It will be a Labour-Greens Coalition.

Nope.

It will be a Labour government with two gagged Green ministers and some "freedoms" to criticise, although not too much because that would look unprofessional.