1
00:00:05,239 --> 00:00:07,259
[music]

2
00:00:09,825 --> 00:00:11,845
[music]

3
00:00:15,280 --> 00:00:18,080
Thank you everyone for coming. Um going

4
00:00:18,080 --> 00:00:20,400
to start by saying this talk was

5
00:00:20,400 --> 00:00:22,720
motivated by a slight desire to break

6
00:00:22,720 --> 00:00:25,600
things and see what chaos I can find.

7
00:00:25,600 --> 00:00:27,680
I've had slightly too many people when I

8
00:00:27,680 --> 00:00:29,840
tell tell them about this talk who are

9
00:00:29,840 --> 00:00:32,320
like, "Oh, so who's going to use this?"

10
00:00:32,320 --> 00:00:34,480
I was like, "Hopefully no one. No one.

11
00:00:34,480 --> 00:00:36,880
No one should do this. We We're just

12
00:00:36,880 --> 00:00:38,960
having some fun. It's Saturday. It's

13
00:00:38,960 --> 00:00:41,920
after lunch. Time for fun."

14
00:00:41,920 --> 00:00:45,760
Yay. Elections. So much fun. Okay.

15
00:00:45,760 --> 00:00:48,399
Jerrymandering. Gerrymandering got its

16
00:00:48,399 --> 00:00:50,160
name because of this guy, Elbridge

17
00:00:50,160 --> 00:00:53,600
Jerry. He was governor of Massachusetts,

18
00:00:53,600 --> 00:00:57,120
I believe it was. In 1812,

19
00:00:57,120 --> 00:00:59,760
he redrew the state borders so that his

20
00:00:59,760 --> 00:01:02,480
party would win. [snorts]

21
00:01:02,480 --> 00:01:04,799
Uh local newspaper looked at those

22
00:01:04,799 --> 00:01:06,720
borders going they look that looks like

23
00:01:06,720 --> 00:01:09,119
a salamander

24
00:01:09,119 --> 00:01:12,560
and drew a political comic about it.

25
00:01:12,560 --> 00:01:15,760
Um that being Yeah. And so ever since

26
00:01:15,760 --> 00:01:19,479
then it's been called gerrymandering

27
00:01:20,159 --> 00:01:23,280
deciding borders. So

28
00:01:23,280 --> 00:01:25,920
we have two groups here, yellow, blue.

29
00:01:25,920 --> 00:01:28,080
We want to make them into five separate

30
00:01:28,080 --> 00:01:30,159
groups.

31
00:01:30,159 --> 00:01:32,159
What should that look like? We got heaps

32
00:01:32,159 --> 00:01:34,880
of options. We could do um equal

33
00:01:34,880 --> 00:01:38,320
distribution. So the underlying the

34
00:01:38,320 --> 00:01:40,479
overall distri um distribution yellow

35
00:01:40,479 --> 00:01:43,600
has 40%, blue has 60% is represented in

36
00:01:43,600 --> 00:01:46,320
each seat. Um but that means blue wins

37
00:01:46,320 --> 00:01:50,799
all five seats, yellow wins none. Um,

38
00:01:50,799 --> 00:01:52,640
we could have that the actual seat

39
00:01:52,640 --> 00:01:54,079
distribution matches the overall

40
00:01:54,079 --> 00:01:57,040
distribution. So, yellow wins two, blue

41
00:01:57,040 --> 00:02:00,560
wins three, that's an option. Um, or

42
00:02:00,560 --> 00:02:03,759
something else, some other combination.

43
00:02:03,759 --> 00:02:05,840
And bit of questions like what is fair?

44
00:02:05,840 --> 00:02:09,200
What makes sense? And even if we do

45
00:02:09,200 --> 00:02:11,280
something that we collectively decide is

46
00:02:11,280 --> 00:02:14,640
fair, people move, things change. It

47
00:02:14,640 --> 00:02:16,800
might be fair now, might not be fair

48
00:02:16,800 --> 00:02:19,280
later.

49
00:02:19,280 --> 00:02:21,760
What does Australia do?

50
00:02:21,760 --> 00:02:24,319
So, Australia at the federal level, we

51
00:02:24,319 --> 00:02:26,319
have the AEC,

52
00:02:26,319 --> 00:02:27,840
the Australian Electoral Commission.

53
00:02:27,840 --> 00:02:30,319
Yes, [applause]

54
00:02:30,319 --> 00:02:32,959
we do love the AEC.

55
00:02:32,959 --> 00:02:35,360
Um, the wiki page for gerrymandering

56
00:02:35,360 --> 00:02:36,879
even has like a picture of some

57
00:02:36,879 --> 00:02:38,480
Australian borders saying Australia has

58
00:02:38,480 --> 00:02:40,560
the independent AEC that defines

59
00:02:40,560 --> 00:02:44,879
borders, which is very sweet.

60
00:02:44,879 --> 00:02:47,280
So, borders change. The AEC calls it a

61
00:02:47,280 --> 00:02:50,640
redistribution um when the entitlement

62
00:02:50,640 --> 00:02:52,800
or a state of a state or territory

63
00:02:52,800 --> 00:02:55,599
change. So the entitlement is how many

64
00:02:55,599 --> 00:02:59,040
seats that state or territory should

65
00:02:59,040 --> 00:03:02,480
roughly get. Um it's based on relative

66
00:03:02,480 --> 00:03:04,640
population of states. [snorts]

67
00:03:04,640 --> 00:03:07,760
Um when electorate when electorate

68
00:03:07,760 --> 00:03:10,159
contains unequal population. So I'm

69
00:03:10,159 --> 00:03:12,319
going to have to read this out exactly.

70
00:03:12,319 --> 00:03:15,200
They say an unequal population is

71
00:03:15,200 --> 00:03:17,040
onethird of the electorate within a

72
00:03:17,040 --> 00:03:19,840
state or one electorate in the ACT or

73
00:03:19,840 --> 00:03:21,840
Northern Territory deviates from the

74
00:03:21,840 --> 00:03:23,840
average state or territory enrollment by

75
00:03:23,840 --> 00:03:26,159
more than 10% for a continuous period of

76
00:03:26,159 --> 00:03:28,800
more than 2 months.

77
00:03:28,800 --> 00:03:31,599
Um now you know

78
00:03:31,599 --> 00:03:34,319
um also when a redistribution hasn't

79
00:03:34,319 --> 00:03:38,120
happened in seven years.

80
00:03:38,239 --> 00:03:40,720
Um I do have QR codes. This goes if you

81
00:03:40,720 --> 00:03:43,840
really want to hear more they have uh

82
00:03:43,840 --> 00:03:45,760
the redistribution

83
00:03:45,760 --> 00:03:48,000
um section on the AEC that goes in so

84
00:03:48,000 --> 00:03:51,599
much detail. Um it is great. This this

85
00:03:51,599 --> 00:03:53,519
talk nearly became everything about

86
00:03:53,519 --> 00:03:56,400
redistribution.

87
00:03:56,400 --> 00:03:59,040
Okay. So how they do it? They work out

88
00:03:59,040 --> 00:04:01,040
the entitlements, how many seats each

89
00:04:01,040 --> 00:04:03,519
state or territory should have. They do

90
00:04:03,519 --> 00:04:05,519
a public consultation period for one

91
00:04:05,519 --> 00:04:08,400
month. And

92
00:04:08,400 --> 00:04:10,400
anyone can s suggest ideas. So of course

93
00:04:10,400 --> 00:04:11,840
the political parties are going to do

94
00:04:11,840 --> 00:04:13,760
it. Various lobby groups can do it.

95
00:04:13,760 --> 00:04:16,880
Think takes etc. But you can also do it.

96
00:04:16,880 --> 00:04:20,000
Um they see on their page it's like want

97
00:04:20,000 --> 00:04:21,680
to get involved here are order

98
00:04:21,680 --> 00:04:26,000
redistributions currently happening.

99
00:04:26,000 --> 00:04:27,680
So then they form a committee cuz it's

100
00:04:27,680 --> 00:04:30,080
government. Um and they propose

101
00:04:30,080 --> 00:04:32,720
boundaries and they give names for the

102
00:04:32,720 --> 00:04:34,960
new boundary new electorates if there

103
00:04:34,960 --> 00:04:36,479
are any. and like they actually give

104
00:04:36,479 --> 00:04:40,160
reasons why they think it's a good idea.

105
00:04:40,160 --> 00:04:42,880
Uh so the main criteria, the main things

106
00:04:42,880 --> 00:04:45,840
they're trying to look for when defining

107
00:04:45,840 --> 00:04:48,400
a new boundary, new electorate,

108
00:04:48,400 --> 00:04:50,160
community interest within the proposed

109
00:04:50,160 --> 00:04:52,160
electorate, including economic, social,

110
00:04:52,160 --> 00:04:54,880
and regional interests. So if it's a

111
00:04:54,880 --> 00:04:56,560
town that's roughly the right size,

112
00:04:56,560 --> 00:04:58,080
they'll try and keep the town together.

113
00:04:58,080 --> 00:05:02,080
They'll try and um

114
00:05:02,080 --> 00:05:03,840
yeah where people go shopping together

115
00:05:03,840 --> 00:05:07,600
like that sort of grouping. Uh they have

116
00:05:07,600 --> 00:05:09,440
means of communication and travel within

117
00:05:09,440 --> 00:05:11,520
the proposed electorate. Uh so if

118
00:05:11,520 --> 00:05:13,919
there's a big mountain range in between

119
00:05:13,919 --> 00:05:17,039
they won't put electroate spanning that

120
00:05:17,039 --> 00:05:19,280
uh physical features any area of the pro

121
00:05:19,280 --> 00:05:20,880
proposed electorate. So that double up

122
00:05:20,880 --> 00:05:22,479
with mountain range and existing

123
00:05:22,479 --> 00:05:24,320
boundaries. They don't redraw everything

124
00:05:24,320 --> 00:05:26,400
each time. They just try and poke things

125
00:05:26,400 --> 00:05:29,280
a little bit.

126
00:05:30,160 --> 00:05:32,960
more public consultation and they also

127
00:05:32,960 --> 00:05:35,680
form a augmented committee that has reps

128
00:05:35,680 --> 00:05:37,600
from states.

129
00:05:37,600 --> 00:05:40,960
They this one looks um they look at the

130
00:05:40,960 --> 00:05:42,880
um original proposal, they look at the

131
00:05:42,880 --> 00:05:44,639
public suggestions, they make their own

132
00:05:44,639 --> 00:05:47,199
proposal, and if they're basically the

133
00:05:47,199 --> 00:05:49,520
same, they're like, "Great, we're done."

134
00:05:49,520 --> 00:05:51,600
If it's not the same, then it goes into

135
00:05:51,600 --> 00:05:53,600
more consultation and it takes a bit

136
00:05:53,600 --> 00:05:55,600
more time.

137
00:05:55,600 --> 00:05:58,000
Finally, boundaries are gazetted and

138
00:05:58,000 --> 00:06:01,800
used for the next election.

139
00:06:02,320 --> 00:06:04,400
Uh, as an aside, one thing I learned

140
00:06:04,400 --> 00:06:06,479
that I'm kind of surprised I didn't know

141
00:06:06,479 --> 00:06:09,759
before is that we don't have a set

142
00:06:09,759 --> 00:06:12,800
number of seats in a House of Reps. Uh,

143
00:06:12,800 --> 00:06:16,319
currently it's 150. Um, the previous few

144
00:06:16,319 --> 00:06:20,560
years has been 151. Has been 150 since

145
00:06:20,560 --> 00:06:25,199
2001 or so. And then like when before

146
00:06:25,199 --> 00:06:27,280
that I was young enough not to care that

147
00:06:27,280 --> 00:06:30,000
much about elections which maybe why I'm

148
00:06:30,000 --> 00:06:32,080
not remembering

149
00:06:32,080 --> 00:06:34,880
a few reasons why this is the case. Um

150
00:06:34,880 --> 00:06:37,039
the number of seats we have in the House

151
00:06:37,039 --> 00:06:41,199
of Reps is tied to a number of senators.

152
00:06:41,199 --> 00:06:44,960
So there's 77 senators which leads to

153
00:06:44,960 --> 00:06:48,400
144 seats kind of but they use that 144

154
00:06:48,400 --> 00:06:51,039
to work out the entitlement

155
00:06:51,039 --> 00:06:55,440
of each state. But then the states have

156
00:06:55,440 --> 00:06:57,759
at least five seats per state, which

157
00:06:57,759 --> 00:06:59,759
Tasmania means Tasmania gets more than

158
00:06:59,759 --> 00:07:01,680
it should. But just because Tasmania

159
00:07:01,680 --> 00:07:02,880
gets more than it should, doesn't mean

160
00:07:02,880 --> 00:07:05,759
that other states lose out. Um,

161
00:07:05,759 --> 00:07:09,360
territories get two at least. And also

162
00:07:09,360 --> 00:07:10,880
rounding issues, cuz of course that has

163
00:07:10,880 --> 00:07:13,840
to happen. Um, [snorts]

164
00:07:13,840 --> 00:07:18,080
so if if your entitlement is 47.6,

165
00:07:18,080 --> 00:07:21,120
they're going to round you up to 48. Uh

166
00:07:21,120 --> 00:07:23,440
apparently we can you can set it up so

167
00:07:23,440 --> 00:07:30,520
you have uh 175 seats in a house.

168
00:07:30,800 --> 00:07:33,680
Another thing too we have external

169
00:07:33,680 --> 00:07:35,680
territories in Australia. So think

170
00:07:35,680 --> 00:07:38,400
Christmas Island, Norfolk Island,

171
00:07:38,400 --> 00:07:41,039
Australian Antarctic Territory.

172
00:07:41,039 --> 00:07:43,599
And if we had more than a few hundred,

173
00:07:43,599 --> 00:07:45,199
few thousand people living in any of

174
00:07:45,199 --> 00:07:48,400
those places, if we had suddenly 200,000

175
00:07:48,400 --> 00:07:50,080
or whatever the entitlement is right

176
00:07:50,080 --> 00:07:52,479
now, they would get their own seat. They

177
00:07:52,479 --> 00:07:54,000
wouldn't be then like told to tie in

178
00:07:54,000 --> 00:07:57,039
with some other state. So theoretically,

179
00:07:57,039 --> 00:07:59,199
if we get a whole bunch of people move

180
00:07:59,199 --> 00:08:02,319
to herd Ireland, say,

181
00:08:02,319 --> 00:08:03,840
and all the other states will still have

182
00:08:03,840 --> 00:08:05,440
their entitlement, we could do quite we

183
00:08:05,440 --> 00:08:09,400
could so much chaos.

184
00:08:09,599 --> 00:08:10,960
Yeah.

185
00:08:10,960 --> 00:08:13,680
Uh, how does voting work? Refresher for

186
00:08:13,680 --> 00:08:16,639
everyone. Um,

187
00:08:16,639 --> 00:08:18,639
in Australia, federal I'm not going to

188
00:08:18,639 --> 00:08:20,400
comment on Tasmania's voting system

189
00:08:20,400 --> 00:08:23,700
because I do not understand it.

190
00:08:23,700 --> 00:08:25,280
[laughter]

191
00:08:25,280 --> 00:08:27,840
Preferential voting. So, you get your

192
00:08:27,840 --> 00:08:29,680
ballot, you have names, you go, I like

193
00:08:29,680 --> 00:08:31,599
this person the most. Or more likely go,

194
00:08:31,599 --> 00:08:33,200
I hate this person the most. And work

195
00:08:33,200 --> 00:08:35,200
your way up to someone who you hate the

196
00:08:35,200 --> 00:08:38,200
least.

197
00:08:38,640 --> 00:08:41,919
and you put preferences down and then

198
00:08:41,919 --> 00:08:43,839
you get told, okay, they count all these

199
00:08:43,839 --> 00:08:47,200
votes and say, at the moment, Tom is

200
00:08:47,200 --> 00:08:50,000
leading, Rachel, unfortunately only

201
00:08:50,000 --> 00:08:53,519
second. Uh, I say this is from the AEC

202
00:08:53,519 --> 00:08:56,720
web page. I did not make up these names.

203
00:08:56,720 --> 00:08:58,240
Out of my control. What's happening

204
00:08:58,240 --> 00:08:59,760
here?

205
00:08:59,760 --> 00:09:03,200
Um, look, go Lily, sorry. You have the

206
00:09:03,200 --> 00:09:04,720
least votes. We're going to look at your

207
00:09:04,720 --> 00:09:07,279
second preferences and distribute them

208
00:09:07,279 --> 00:09:11,040
to Omar, Tom, and Rachel.

209
00:09:11,040 --> 00:09:13,279
And so now has caught up with Rachel and

210
00:09:13,279 --> 00:09:16,080
we're both equal last.

211
00:09:16,080 --> 00:09:18,080
But as Omar has the least first

212
00:09:18,080 --> 00:09:21,680
preferences, he is removed.

213
00:09:21,680 --> 00:09:24,240
So Omar's votes um third second

214
00:09:24,240 --> 00:09:27,920
preferences form gets distributed out.

215
00:09:27,920 --> 00:09:29,920
And hey, look, Rachel now has more than

216
00:09:29,920 --> 00:09:33,040
50% of the votes. Uh she is the winner

217
00:09:33,040 --> 00:09:36,440
for this state.

218
00:09:36,720 --> 00:09:39,736
Yay. [laughter]

219
00:09:40,480 --> 00:09:42,160
Yeah.

220
00:09:42,160 --> 00:09:46,080
Okay. So what data is available?

221
00:09:46,080 --> 00:09:47,839
There is actually quite a bit of data

222
00:09:47,839 --> 00:09:51,040
and I was excited to find that you can

223
00:09:51,040 --> 00:09:52,320
so it's actually down to the booth

224
00:09:52,320 --> 00:09:55,040
level. So where you go and vote uh you

225
00:09:55,040 --> 00:09:57,839
can go to the A web page. They've hidden

226
00:09:57,839 --> 00:10:00,000
it quite recently a bit more. in the

227
00:10:00,000 --> 00:10:05,040
tally room section. Um, and download the

228
00:10:05,040 --> 00:10:08,880
first preferences for the booth or the

229
00:10:08,880 --> 00:10:11,279
two party preferred for the booth. And

230
00:10:11,279 --> 00:10:13,839
you can also see the preference flow,

231
00:10:13,839 --> 00:10:17,760
but the preference flow is limited to um

232
00:10:17,760 --> 00:10:20,079
what actually happened in that in that

233
00:10:20,079 --> 00:10:23,079
seat.

234
00:10:24,160 --> 00:10:26,000
Um,

235
00:10:26,000 --> 00:10:27,279
at this point we're going to have to

236
00:10:27,279 --> 00:10:28,800
make some compromise cuz while it would

237
00:10:28,800 --> 00:10:31,279
be lovely

238
00:10:31,279 --> 00:10:33,839
to do the whole preference flow, we

239
00:10:33,839 --> 00:10:36,800
don't have the data because if you have

240
00:10:36,800 --> 00:10:39,519
two seats that have like it's a

241
00:10:39,519 --> 00:10:42,240
competition between different parties in

242
00:10:42,240 --> 00:10:44,560
those seats. We can't map these flows

243
00:10:44,560 --> 00:10:46,640
properly and and we don't have the full

244
00:10:46,640 --> 00:10:50,320
data to try and uh work like to

245
00:10:50,320 --> 00:10:53,839
interpolate even half decently.

246
00:10:53,839 --> 00:10:57,600
Uh side note in the in their AC's also

247
00:10:57,600 --> 00:10:59,839
recently included a bunch of SA1 level

248
00:10:59,839 --> 00:11:02,480
data uh for their election things. I'm

249
00:11:02,480 --> 00:11:04,720
not quite sure how they work it out and

250
00:11:04,720 --> 00:11:06,399
there are warnings saying this is not

251
00:11:06,399 --> 00:11:09,279
like true vote counts. These are maps

252
00:11:09,279 --> 00:11:12,839
went into this.

253
00:11:12,959 --> 00:11:14,560
So these are where all the booths are in

254
00:11:14,560 --> 00:11:18,160
Australia. Um we have lots. It's your

255
00:11:18,160 --> 00:11:20,480
local public school. Your My local one

256
00:11:20,480 --> 00:11:22,160
was the archery club for a bit. I did

257
00:11:22,160 --> 00:11:25,760
not know I had a local archery club.

258
00:11:25,760 --> 00:11:29,040
Uh there are 8,875

259
00:11:29,040 --> 00:11:33,120
in this data set. Yeah. And that's not

260
00:11:33,120 --> 00:11:35,440
even including like mobile um mobile

261
00:11:35,440 --> 00:11:38,240
sites. So there are mobile sites that go

262
00:11:38,240 --> 00:11:41,440
out to more rural areas to hospitals to

263
00:11:41,440 --> 00:11:43,920
jails. Uh not looking at them because

264
00:11:43,920 --> 00:11:48,240
they move that complicates things. We're

265
00:11:48,240 --> 00:11:49,920
just going to focus on the ones you can

266
00:11:49,920 --> 00:11:52,800
vote at on election day. Uh so some

267
00:11:52,800 --> 00:11:54,399
people might be going, "Oh, but what

268
00:11:54,399 --> 00:11:57,233
about pre-polling?" Yes, very much. Yes.

269
00:11:57,233 --> 00:11:59,253
[snorts]

270
00:11:59,600 --> 00:12:02,000
Okay. So, let's start with something a

271
00:12:02,000 --> 00:12:03,680
little bit small. Going to look at

272
00:12:03,680 --> 00:12:05,279
Wooling Gong. Wingong is great. You

273
00:12:05,279 --> 00:12:10,240
should all visit. Um [snorts]

274
00:12:10,240 --> 00:12:11,839
okay, these are where the booths are.

275
00:12:11,839 --> 00:12:14,000
And we're just going we're going to do

276
00:12:14,000 --> 00:12:15,600
basic stuff. We're going to be like

277
00:12:15,600 --> 00:12:19,040
people go to the closest booth to them.

278
00:12:19,040 --> 00:12:21,839
So therefore, anyone who lives in these

279
00:12:21,839 --> 00:12:23,600
polygons will go to the booth assigned

280
00:12:23,600 --> 00:12:25,360
to their polygon. So this is a voronoi

281
00:12:25,360 --> 00:12:27,760
tessellation.

282
00:12:27,760 --> 00:12:31,120
It works. It looks good.

283
00:12:31,120 --> 00:12:32,880
We will fix it up later if it causes

284
00:12:32,880 --> 00:12:35,880
problems.

285
00:12:37,200 --> 00:12:39,839
But tried something bigger. And this is

286
00:12:39,839 --> 00:12:41,040
like trying to do it for all of New

287
00:12:41,040 --> 00:12:42,639
South Wales. And it's something

288
00:12:42,639 --> 00:12:45,639
happening.

289
00:12:45,680 --> 00:12:47,040
Um, does anyone want to have a guess

290
00:12:47,040 --> 00:12:47,839
what's happening?

291
00:12:47,839 --> 00:12:49,633
It's biblically accurate projection.

292
00:12:49,633 --> 00:12:51,653
[laughter]

293
00:12:51,760 --> 00:12:53,680
Uh, so biblically accurate projection

294
00:12:53,680 --> 00:12:55,920
was the suggestion for the recording.

295
00:12:55,920 --> 00:12:58,720
Zero external territories.

296
00:12:58,720 --> 00:13:01,120
Uh, so we had zeros in external

297
00:13:01,120 --> 00:13:03,519
territories. No. Um, external good

298
00:13:03,519 --> 00:13:05,760
spotting on the external is Lord How

299
00:13:05,760 --> 00:13:07,440
Island not actually an external

300
00:13:07,440 --> 00:13:10,720
territory for New South Wales? Um,

301
00:13:10,720 --> 00:13:12,240
they've decided to New South Wales.

302
00:13:12,240 --> 00:13:13,920
They're a bit

303
00:13:13,920 --> 00:13:16,399
in state in federal elections I believe

304
00:13:16,399 --> 00:13:18,160
it's part of the Sydney electorate and

305
00:13:18,160 --> 00:13:20,639
in state ones is part of Port Stevens

306
00:13:20,639 --> 00:13:25,160
which is like 300ks north of Sydney

307
00:13:26,160 --> 00:13:29,920
questions no what's happening is I a bit

308
00:13:29,920 --> 00:13:32,079
naively was long latter coordinates

309
00:13:32,079 --> 00:13:35,120
let's just chuck it in Voronoi

310
00:13:35,120 --> 00:13:36,959
algorithms are like no we want flat

311
00:13:36,959 --> 00:13:41,600
stuff your sphere is not flat

312
00:13:41,600 --> 00:13:44,160
so little bit of a detour learning about

313
00:13:44,160 --> 00:13:49,040
some GIS stuff. Uh I ended up using EPSG

314
00:13:49,040 --> 00:13:52,800
9473 which is set up to be decent for

315
00:13:52,800 --> 00:13:57,120
Australia as a whole. Um

316
00:13:57,120 --> 00:14:00,880
and look, it works for New South Wales.

317
00:14:00,880 --> 00:14:03,360
Um you'll see a bit later there's a bit

318
00:14:03,360 --> 00:14:05,920
of funkiness with other states and like

319
00:14:05,920 --> 00:14:10,040
that's fine. We'll cope.

320
00:14:10,720 --> 00:14:13,839
Um this looks nice. This looks good. Um,

321
00:14:13,839 --> 00:14:15,839
going to be even fancier and clip it by

322
00:14:15,839 --> 00:14:17,839
the state borders.

323
00:14:17,839 --> 00:14:21,120
And I'm I'm very happy that we got uh

324
00:14:21,120 --> 00:14:24,560
acts clipped out. And you might see a

325
00:14:24,560 --> 00:14:26,720
little bit there that's white. That's

326
00:14:26,720 --> 00:14:30,639
Jarvis Bay. Uh Javas Bay is uh a

327
00:14:30,639 --> 00:14:32,800
territory that's for electoral purposes

328
00:14:32,800 --> 00:14:35,519
part of CRA.

329
00:14:35,519 --> 00:14:38,519
Um,

330
00:14:39,600 --> 00:14:42,079
it's it exists to be a Navy station

331
00:14:42,079 --> 00:14:45,079
essentially.

332
00:14:46,320 --> 00:14:48,160
Okay, we're going pretty good for time.

333
00:14:48,160 --> 00:14:51,360
I was thought we might go way over.

334
00:14:51,360 --> 00:14:53,519
Okay, on to actual gerrymandering. We've

335
00:14:53,519 --> 00:14:55,519
done some

336
00:14:55,519 --> 00:14:58,670
got the fundamentals down.

337
00:14:58,670 --> 00:14:59,600
[snorts]

338
00:14:59,600 --> 00:15:01,440
All right.

339
00:15:01,440 --> 00:15:03,199
So essentially what we want to do with

340
00:15:03,199 --> 00:15:05,279
this area mandering we have our voronoi

341
00:15:05,279 --> 00:15:08,720
regions and we want to select them such

342
00:15:08,720 --> 00:15:13,680
that um we how can we select regions so

343
00:15:13,680 --> 00:15:16,480
that the divisions or electorates are

344
00:15:16,480 --> 00:15:18,560
chose are won in such a way that a

345
00:15:18,560 --> 00:15:20,560
particular party gets the most votes in

346
00:15:20,560 --> 00:15:22,480
the most divisions

347
00:15:22,480 --> 00:15:24,480
at the heart of it that's like a

348
00:15:24,480 --> 00:15:27,440
assignment optimization problem I did a

349
00:15:27,440 --> 00:15:29,680
lot of linear integer programming why

350
00:15:29,680 --> 00:15:31,680
not this sounds like a data skews to get

351
00:15:31,680 --> 00:15:34,880
back into some of that. Uh so for those

352
00:15:34,880 --> 00:15:38,959
uninitiated, integer linear programming,

353
00:15:38,959 --> 00:15:41,519
it's got its start in World War II where

354
00:15:41,519 --> 00:15:44,880
the US military's problem was we have to

355
00:15:44,880 --> 00:15:46,880
feed lots of people. We need to feed

356
00:15:46,880 --> 00:15:48,399
lots of people so they have the right

357
00:15:48,399 --> 00:15:50,959
amount of nutrients. They have enough

358
00:15:50,959 --> 00:15:53,279
energy, they don't get too much salt,

359
00:15:53,279 --> 00:15:56,399
but we want to do it as cheaply as

360
00:15:56,399 --> 00:15:58,959
possible.

361
00:15:58,959 --> 00:16:01,519
Um, so that that was like sort of the

362
00:16:01,519 --> 00:16:04,560
starting problem

363
00:16:04,560 --> 00:16:09,279
that spawned into a whole field.

364
00:16:09,279 --> 00:16:11,680
You might have heard the story about the

365
00:16:11,680 --> 00:16:13,680
student coming late into class and there

366
00:16:13,680 --> 00:16:15,920
were two problems on the board and he

367
00:16:15,920 --> 00:16:18,160
thought they were homework exercises and

368
00:16:18,160 --> 00:16:19,920
he came back to class the next week

369
00:16:19,920 --> 00:16:22,160
going, "Well, they were hard. I only

370
00:16:22,160 --> 00:16:24,880
solved one of them." And turns out they

371
00:16:24,880 --> 00:16:28,240
were like examples of plastic unsolved

372
00:16:28,240 --> 00:16:31,920
problems. So yeah, he got his PhD thesis

373
00:16:31,920 --> 00:16:34,639
in a weak work or something like that.

374
00:16:34,639 --> 00:16:36,800
Um but that that problem that he solved

375
00:16:36,800 --> 00:16:38,959
led directly into the development of the

376
00:16:38,959 --> 00:16:40,639
simplex algorithm which is one of the

377
00:16:40,639 --> 00:16:43,199
main algorithms used for solving integer

378
00:16:43,199 --> 00:16:45,600
linear programs.

379
00:16:45,600 --> 00:16:48,399
Um the other two people Allison

380
00:16:48,399 --> 00:16:51,360
Hardcourt Alyssa land they developed one

381
00:16:51,360 --> 00:16:53,120
of the other fundamental algorithms in

382
00:16:53,120 --> 00:16:55,759
this area called branch andbound uh

383
00:16:55,759 --> 00:16:58,320
they're both Australian they were in

384
00:16:58,320 --> 00:16:59,600
London at the time when they were

385
00:16:59,600 --> 00:17:01,839
working on it and the paper that they

386
00:17:01,839 --> 00:17:04,160
published was published under AG Doick

387
00:17:04,160 --> 00:17:07,520
which is Allison's maiden name and no

388
00:17:07,520 --> 00:17:09,120
one has realized that there were two

389
00:17:09,120 --> 00:17:12,400
women who did that for decades

390
00:17:12,400 --> 00:17:15,199
and it's quite recently just in 2019 19

391
00:17:15,199 --> 00:17:18,400
where Allison got uh Victoria Senior of

392
00:17:18,400 --> 00:17:20,240
the Year in recognition of the work she

393
00:17:20,240 --> 00:17:23,039
did there.

394
00:17:23,039 --> 00:17:26,880
Okay. So should Okay, I'm going to

395
00:17:26,880 --> 00:17:28,720
explain what linear integer programming

396
00:17:28,720 --> 00:17:32,160
is on the fly while we're doing it.

397
00:17:32,160 --> 00:17:34,559
We got a bunch of stuff that we know.

398
00:17:34,559 --> 00:17:36,080
I'm not going to go into too much detail

399
00:17:36,080 --> 00:17:37,120
because this is just going to be a lot

400
00:17:37,120 --> 00:17:39,520
of maths and it's last day of

401
00:17:39,520 --> 00:17:41,120
conference. No one wants to do this much

402
00:17:41,120 --> 00:17:43,600
maths.

403
00:17:43,600 --> 00:17:45,600
We define variables. So in this case,

404
00:17:45,600 --> 00:17:50,160
our key variable is X subscript RD. So

405
00:17:50,160 --> 00:17:52,799
it's a boolean one. Is this region

406
00:17:52,799 --> 00:17:55,360
chosen to be part of this division or

407
00:17:55,360 --> 00:17:57,679
electorate?

408
00:17:57,679 --> 00:18:00,720
Um we want to select them all so that

409
00:18:00,720 --> 00:18:02,400
each region must be in exactly one

410
00:18:02,400 --> 00:18:04,799
division division or region electorate.

411
00:18:04,799 --> 00:18:09,000
You can't miss mix and match.

412
00:18:09,360 --> 00:18:12,400
Um let's set not going to go too much.

413
00:18:12,400 --> 00:18:14,000
When I talk about integer linear

414
00:18:14,000 --> 00:18:16,799
programming, it's about defining

415
00:18:16,799 --> 00:18:19,520
variables, defining constraints like

416
00:18:19,520 --> 00:18:22,880
that sum I have at the bottom and then

417
00:18:22,880 --> 00:18:24,320
using different algorithms like the

418
00:18:24,320 --> 00:18:26,480
simplex, branch and bound to find

419
00:18:26,480 --> 00:18:29,919
optimal solutions.

420
00:18:29,919 --> 00:18:31,440
Okay, so we got this. We got a very

421
00:18:31,440 --> 00:18:35,360
basic one. It

422
00:18:35,360 --> 00:18:38,160
finds it divides up camber act into

423
00:18:38,160 --> 00:18:40,320
different seats. At the moment they're

424
00:18:40,320 --> 00:18:42,320
just randomly assigned. I haven't put

425
00:18:42,320 --> 00:18:46,320
any condition in to say this is things

426
00:18:46,320 --> 00:18:49,120
should be next to each other and kind of

427
00:18:49,120 --> 00:18:51,280
makes sense to have contiguous um

428
00:18:51,280 --> 00:18:52,799
electorates.

429
00:18:52,799 --> 00:18:54,400
Um that's actually that's something that

430
00:18:54,400 --> 00:18:56,080
A cares about when they talk about

431
00:18:56,080 --> 00:18:59,200
transport options and things like that.

432
00:18:59,200 --> 00:19:02,080
Um so we have [snorts] that

433
00:19:02,080 --> 00:19:05,440
contiguous regions. I yes I'm looking at

434
00:19:05,440 --> 00:19:07,200
the time going maybe not going so well

435
00:19:07,200 --> 00:19:10,160
for time now.

436
00:19:10,160 --> 00:19:13,440
One of my favorite things about

437
00:19:13,440 --> 00:19:16,240
um linear programming is just how often

438
00:19:16,240 --> 00:19:18,160
you can reduce problems down to graph

439
00:19:18,160 --> 00:19:19,919
theory.

440
00:19:19,919 --> 00:19:21,600
So

441
00:19:21,600 --> 00:19:24,880
this we got wool gong again. Wllingong's

442
00:19:24,880 --> 00:19:27,600
nice and small for this example. I can

443
00:19:27,600 --> 00:19:30,640
be like well boos are just vertices,

444
00:19:30,640 --> 00:19:34,559
adjacencies are edges.

445
00:19:34,559 --> 00:19:37,280
um zoom in a little bit

446
00:19:37,280 --> 00:19:40,000
and we can apply uh this is just really

447
00:19:40,000 --> 00:19:43,200
a max like not a max flow but a flow

448
00:19:43,200 --> 00:19:46,559
problem. We want to define

449
00:19:46,559 --> 00:19:48,720
if we want to split this up into two um

450
00:19:48,720 --> 00:19:50,240
divisions. We got this yellow one, we

451
00:19:50,240 --> 00:19:52,880
got this purple one say it looks like

452
00:19:52,880 --> 00:19:56,000
that and we want to have flow like be a

453
00:19:56,000 --> 00:19:57,840
how are we going to work out the to make

454
00:19:57,840 --> 00:19:59,600
it contiguous? We're going to say well

455
00:19:59,600 --> 00:20:00,880
there must be some sort of flow. So,

456
00:20:00,880 --> 00:20:02,320
think like a water pipe going through

457
00:20:02,320 --> 00:20:04,480
it. And there must be water going to

458
00:20:04,480 --> 00:20:07,720
every spot.

459
00:20:08,160 --> 00:20:10,400
I'm going to skip the actual details on

460
00:20:10,400 --> 00:20:13,039
this bit, but it's fun. Go back and look

461
00:20:13,039 --> 00:20:17,003
at it later. Um, but the point is

462
00:20:17,003 --> 00:20:18,160
[snorts] you can't have things like

463
00:20:18,160 --> 00:20:20,240
that. It's great. That's what we don't

464
00:20:20,240 --> 00:20:22,899
want to have.

465
00:20:22,899 --> 00:20:24,080
[snorts]

466
00:20:24,080 --> 00:20:25,520
And then we have more. We have so much

467
00:20:25,520 --> 00:20:28,320
maps to define this. Few things I'll

468
00:20:28,320 --> 00:20:29,919
point out. We got like a whole bunch of

469
00:20:29,919 --> 00:20:32,400
extra variables. Like we got f that has

470
00:20:32,400 --> 00:20:34,960
three subscripts. Two of them are for

471
00:20:34,960 --> 00:20:37,120
regions. And you think about those 8,000

472
00:20:37,120 --> 00:20:39,520
booths that we had.

473
00:20:39,520 --> 00:20:42,559
Think how big this is going to be. Yeah.

474
00:20:42,559 --> 00:20:45,840
More foreshadowing. [snorts]

475
00:20:45,840 --> 00:20:47,679
Uh more maths. This is all just for

476
00:20:47,679 --> 00:20:51,200
flow. And we do that and we get this. Uh

477
00:20:51,200 --> 00:20:53,679
we got I included JavaS Bay and Northolk

478
00:20:53,679 --> 00:20:55,280
Island now because they're part of CRA

479
00:20:55,280 --> 00:20:58,799
for elections.

480
00:20:58,799 --> 00:21:02,880
Um to point Javas Bay and Norfolk,

481
00:21:02,880 --> 00:21:04,640
they're not contiguous to CRA. They're a

482
00:21:04,640 --> 00:21:08,640
few hundred thousand kilometers away. My

483
00:21:08,640 --> 00:21:10,880
first instinct was to say, well then no

484
00:21:10,880 --> 00:21:12,320
nowhere near. So why don't I make them

485
00:21:12,320 --> 00:21:14,480
adjacent to everything

486
00:21:14,480 --> 00:21:17,360
that blew up the problem so much like

487
00:21:17,360 --> 00:21:20,320
solving CRA takes like no more than a

488
00:21:20,320 --> 00:21:22,320
few minutes for this sort of thing. And

489
00:21:22,320 --> 00:21:24,480
this was still running like an hour or

490
00:21:24,480 --> 00:21:26,240
two later. And I was like, okay, no, bad

491
00:21:26,240 --> 00:21:27,360
idea.

492
00:21:27,360 --> 00:21:29,919
I I think in the end I said closest to

493
00:21:29,919 --> 00:21:32,240
is adjacent to the 10 closest things to

494
00:21:32,240 --> 00:21:36,480
it and that that worked.

495
00:21:36,480 --> 00:21:38,159
Also need to consider how many people

496
00:21:38,159 --> 00:21:41,120
are in each division cuz

497
00:21:41,120 --> 00:21:44,400
yeah kind of make things a bit fair.

498
00:21:44,400 --> 00:21:46,480
We need to define who is actually

499
00:21:46,480 --> 00:21:48,960
winning each division. So which party.

500
00:21:48,960 --> 00:21:52,799
So this is defined as the party that has

501
00:21:52,799 --> 00:21:55,679
the most first preference votes. So that

502
00:21:55,679 --> 00:21:58,880
that's kind of what we had the data for.

503
00:21:58,880 --> 00:22:01,120
And then finally finally finally we want

504
00:22:01,120 --> 00:22:03,440
to maximize the number of divisions that

505
00:22:03,440 --> 00:22:08,080
a particular party has won.

506
00:22:08,080 --> 00:22:10,240
And finally and 23 minutes into this

507
00:22:10,240 --> 00:22:15,000
talk we're on to actual gerrymandering.

508
00:22:15,200 --> 00:22:16,799
So let's start small. Let's start with

509
00:22:16,799 --> 00:22:19,280
ACT going

510
00:22:19,280 --> 00:22:21,600
not going alphabetical order um but

511
00:22:21,600 --> 00:22:24,320
close enough. Started with the ALP, wins

512
00:22:24,320 --> 00:22:26,559
free seats. Great. That's what they

513
00:22:26,559 --> 00:22:29,280
currently have.

514
00:22:29,280 --> 00:22:31,200
Tried with the um LP in this case,

515
00:22:31,200 --> 00:22:34,960
Liberal Party, no seats, which I'm bit

516
00:22:34,960 --> 00:22:38,159
surprised about. Tried with Greens, also

517
00:22:38,159 --> 00:22:41,840
no seats. Bit surprised.

518
00:22:41,840 --> 00:22:43,679
Had a look at the data. And so the

519
00:22:43,679 --> 00:22:47,280
purple column is the amount of uh boosts

520
00:22:47,280 --> 00:22:49,600
that they've got, amount of first

521
00:22:49,600 --> 00:22:52,080
preferences at that booth. um like where

522
00:22:52,080 --> 00:22:54,000
they've won the booth and it's just

523
00:22:54,000 --> 00:22:56,480
overwhelmingly ALP

524
00:22:56,480 --> 00:22:58,159
and this is going back to the

525
00:22:58,159 --> 00:23:01,679
pre-polling pre-polling and postal votes

526
00:23:01,679 --> 00:23:03,919
and what people actually do on the day

527
00:23:03,919 --> 00:23:06,480
is all a bit different and I suspect

528
00:23:06,480 --> 00:23:08,480
that's why it's different here. I don't

529
00:23:08,480 --> 00:23:10,320
have confirmation.

530
00:23:10,320 --> 00:23:12,320
Um

531
00:23:12,320 --> 00:23:14,320
but yeah, it's a bit weird this data

532
00:23:14,320 --> 00:23:16,240
from what it looks like and I suspect

533
00:23:16,240 --> 00:23:19,520
that is why it's weird.

534
00:23:19,520 --> 00:23:20,960
Like I really wanted to break something.

535
00:23:20,960 --> 00:23:24,600
I wanted some chaos.

536
00:23:24,880 --> 00:23:26,880
So I did notice independence.

537
00:23:26,880 --> 00:23:29,760
Independence won something. So like

538
00:23:29,760 --> 00:23:34,080
cool. Independence win two seats.

539
00:23:34,080 --> 00:23:35,840
When you let a division just be a single

540
00:23:35,840 --> 00:23:38,801
booth, [laughter]

541
00:23:39,520 --> 00:23:43,600
which okay, we try we did we got there.

542
00:23:43,600 --> 00:23:45,600
And then I was like New South Wales,

543
00:23:45,600 --> 00:23:48,240
sure, why not? And I was like, it's

544
00:23:48,240 --> 00:23:49,760
going to take a while. And I ran it

545
00:23:49,760 --> 00:23:51,440
overnight. And then my laptop was very

546
00:23:51,440 --> 00:23:53,039
sad the next morning and had to have a

547
00:23:53,039 --> 00:23:54,960
hard reboot.

548
00:23:54,960 --> 00:23:56,960
And I tried to do I I was like, okay, I

549
00:23:56,960 --> 00:23:59,919
set up some cloud stuff and that was

550
00:23:59,919 --> 00:24:02,159
also sad.

551
00:24:02,159 --> 00:24:04,400
And I fiddled with it and I looked at

552
00:24:04,400 --> 00:24:06,559
going, do I need to do fancy math stuff

553
00:24:06,559 --> 00:24:09,600
to make it better? And then I was like,

554
00:24:09,600 --> 00:24:12,480
oh, looked up the stuff. This is using o

555
00:24:12,480 --> 00:24:15,600
tools the python package which I'm

556
00:24:15,600 --> 00:24:19,919
regretting because it's has a two um 2 g

557
00:24:19,919 --> 00:24:22,480
by model limit which is tiny it's really

558
00:24:22,480 --> 00:24:25,600
tiny um and that's because of protobuff

559
00:24:25,600 --> 00:24:28,640
um pro every yeah exactly everything's

560
00:24:28,640 --> 00:24:31,760
protobuff's fault like

561
00:24:31,760 --> 00:24:33,360
if you've worked with me you have heard

562
00:24:33,360 --> 00:24:36,360
rants

563
00:24:36,400 --> 00:24:39,200
and this is not all looking at first

564
00:24:39,200 --> 00:24:42,080
preferences nationwide I'd like I cut

565
00:24:42,080 --> 00:24:43,520
out a lot of the minor parties here.

566
00:24:43,520 --> 00:24:44,799
These are just roughly who's in

567
00:24:44,799 --> 00:24:47,360
parliament at the moment. And it is

568
00:24:47,360 --> 00:24:51,360
quite um towards the major parties. Like

569
00:24:51,360 --> 00:24:52,640
we do have there's three different

570
00:24:52,640 --> 00:24:54,559
columns for what's essentially the L&P

571
00:24:54,559 --> 00:24:58,080
here, but it's like quite

572
00:24:58,080 --> 00:24:59,679
not much variation that I was hoping

573
00:24:59,679 --> 00:25:02,400
for.

574
00:25:02,400 --> 00:25:05,360
Um cuz when minor parties wins, they do

575
00:25:05,360 --> 00:25:07,600
tend to do it on preferences, which

576
00:25:07,600 --> 00:25:10,480
we're just ignoring.

577
00:25:10,480 --> 00:25:11,919
And I was like, well, could we just look

578
00:25:11,919 --> 00:25:14,320
at the two party preferred for each seat

579
00:25:14,320 --> 00:25:16,240
and like just get some bigger numbers

580
00:25:16,240 --> 00:25:18,240
out of each one?

581
00:25:18,240 --> 00:25:20,640
But then we're kind of a bit limited too

582
00:25:20,640 --> 00:25:25,120
because most seats are still uh classic

583
00:25:25,120 --> 00:25:29,440
ALP LMP com um combination.

584
00:25:29,440 --> 00:25:32,240
And even if it wasn't one of those two,

585
00:25:32,240 --> 00:25:34,880
a neighboring like you like say it's an

586
00:25:34,880 --> 00:25:37,200
independent versus a major. If you want

587
00:25:37,200 --> 00:25:38,640
to do any fun gerrymandering for

588
00:25:38,640 --> 00:25:39,679
independence, you would want a

589
00:25:39,679 --> 00:25:41,520
neighboring seat that's also independent

590
00:25:41,520 --> 00:25:43,679
on two party preferred. And how often is

591
00:25:43,679 --> 00:25:45,520
that happening

592
00:25:45,520 --> 00:25:47,440
like so you then sort of have things

593
00:25:47,440 --> 00:25:50,559
like you're inherently encoding existing

594
00:25:50,559 --> 00:25:51,919
electoral boundaries, which is kind of

595
00:25:51,919 --> 00:25:54,640
what the AEC likes to do too, but we're

596
00:25:54,640 --> 00:25:56,720
trying not to be the AEC a bit bit as

597
00:25:56,720 --> 00:25:59,679
well.

598
00:25:59,679 --> 00:26:01,520
Um,

599
00:26:01,520 --> 00:26:04,960
oh yeah, Anthony Green, great blog. Uh

600
00:26:04,960 --> 00:26:07,520
this is the

601
00:26:07,520 --> 00:26:09,279
how cuz yeah that's something that has

602
00:26:09,279 --> 00:26:11,120
been changing over recent elections.

603
00:26:11,120 --> 00:26:12,640
We're getting many more independents in

604
00:26:12,640 --> 00:26:14,720
the parliament and minor parties. So

605
00:26:14,720 --> 00:26:16,880
we're having way less

606
00:26:16,880 --> 00:26:23,799
uh classic ALP LMP seat contestants.

607
00:26:24,240 --> 00:26:26,240
U also rise in pre-polling like it's

608
00:26:26,240 --> 00:26:28,720
getting close to 40% of votes are now

609
00:26:28,720 --> 00:26:30,960
prepled and that's not reflected in the

610
00:26:30,960 --> 00:26:34,720
data I have. Um, and I probably could do

611
00:26:34,720 --> 00:26:37,440
something, but the there's more less

612
00:26:37,440 --> 00:26:39,039
pre-polling sites. So, you're then

613
00:26:39,039 --> 00:26:41,279
looking over a much bigger area, which

614
00:26:41,279 --> 00:26:43,039
you could, but I feel sad for missing

615
00:26:43,039 --> 00:26:47,120
out the details and

616
00:26:47,120 --> 00:26:51,919
lots of things to consider. Um, lots and

617
00:26:51,919 --> 00:26:53,440
yeah, I want to work on this more. I

618
00:26:53,440 --> 00:26:55,440
want to try things. I have to rewrite it

619
00:26:55,440 --> 00:26:59,600
out of O tools cuz not the best choice.

620
00:26:59,600 --> 00:27:01,360
And we are running out of time. two

621
00:27:01,360 --> 00:27:03,440
minutes as I just told. So, let's go to

622
00:27:03,440 --> 00:27:05,600
some questions.

623
00:27:05,600 --> 00:27:09,058
Thank you. [applause]

624
00:27:13,520 --> 00:27:15,840
Thank you, Rachel. Uh, the first note I

625
00:27:15,840 --> 00:27:18,799
wrote down was 8,000 polling booths.

626
00:27:18,799 --> 00:27:20,880
That's a lot of sausage sizzles. We have

627
00:27:20,880 --> 00:27:26,039
time for one question here.

628
00:27:27,600 --> 00:27:29,679
So what you're saying is we could do a

629
00:27:29,679 --> 00:27:31,679
better job gerrymandering if we had more

630
00:27:31,679 --> 00:27:33,440
data from the AEC.

631
00:27:33,440 --> 00:27:36,440
Yes.

632
00:27:37,120 --> 00:27:41,240
We have time for another question.

633
00:27:42,559 --> 00:27:45,039
So evidently like polling places are not

634
00:27:45,039 --> 00:27:46,400
representative of the electoral district

635
00:27:46,400 --> 00:27:48,559
as a whole. Are you aware of any other

636
00:27:48,559 --> 00:27:50,640
data sources or like polling sources

637
00:27:50,640 --> 00:27:52,960
that are a bit open to give more

638
00:27:52,960 --> 00:27:54,640
information to like what's in a suburb,

639
00:27:54,640 --> 00:27:56,240
what's in a street about the the

640
00:27:56,240 --> 00:27:58,000
political climate there?

641
00:27:58,000 --> 00:28:02,799
Uh yeah, that's a great question. I

642
00:28:02,799 --> 00:28:04,640
was fully intending to go down that path

643
00:28:04,640 --> 00:28:07,120
and look at things too. Um ABS is the

644
00:28:07,120 --> 00:28:10,799
obvious choice. Um actually now that the

645
00:28:10,799 --> 00:28:13,120
AC is doing some SA1 data, you could do

646
00:28:13,120 --> 00:28:14,880
quite a bit of interesting stuff. Um so

647
00:28:14,880 --> 00:28:16,399
essay one if you're not familiar is the

648
00:28:16,399 --> 00:28:18,720
statistical areas it's roughly like half

649
00:28:18,720 --> 00:28:21,360
a suburb in size like think in a city

650
00:28:21,360 --> 00:28:24,799
suburb half uh it's matched by um number

651
00:28:24,799 --> 00:28:27,440
of people so they're good good good

652
00:28:27,440 --> 00:28:29,760
comparison tool

653
00:28:29,760 --> 00:28:33,200
um yeah I'm not I'm not aware like I

654
00:28:33,200 --> 00:28:34,720
would like to be aware of but I am not

655
00:28:34,720 --> 00:28:38,679
of other good sources.

656
00:28:39,440 --> 00:28:41,039
Okay, I don't think we have time for any

657
00:28:41,039 --> 00:28:44,720
more, but I'd just like to say when I

658
00:28:44,720 --> 00:28:46,799
get up here, thank you very much,

659
00:28:46,799 --> 00:28:49,219
Rachel. Here's your very important mug.

660
00:28:49,219 --> 00:28:51,239
[applause]