1
00:00:00,480 --> 00:00:03,480
foreign

2
00:00:10,219 --> 00:00:16,619
hey thank you okay it's lightning talks

3
00:00:13,620 --> 00:00:19,080
time uh you know the deal five minutes

4
00:00:16,619 --> 00:00:22,859
of talk uh when we get to the last 10 or

5
00:00:19,080 --> 00:00:24,900
so seconds finger claps can we practice

6
00:00:22,859 --> 00:00:26,880
and then as the time goes down to the

7
00:00:24,900 --> 00:00:29,820
last two or three seconds we turn it

8
00:00:26,880 --> 00:00:31,560
into hand Applause proper claps and then

9
00:00:29,820 --> 00:00:34,140
an outrageous noise completely

10
00:00:31,560 --> 00:00:35,170
uproarious as we welcome to the stage

11
00:00:34,140 --> 00:00:38,679
Tom Eastman

12
00:00:35,170 --> 00:00:38,679
[Applause]

13
00:00:42,500 --> 00:00:46,379
okay I've got five minutes to try to

14
00:00:45,059 --> 00:00:47,760
convince you that these little lumps of

15
00:00:46,379 --> 00:00:49,500
hexadecimal are actually secretly

16
00:00:47,760 --> 00:00:51,180
incredibly fascinating

17
00:00:49,500 --> 00:00:52,920
um you've all probably seen uuids before

18
00:00:51,180 --> 00:00:54,539
you may or may not use them a lot you

19
00:00:52,920 --> 00:00:55,920
probably just use one version you're

20
00:00:54,539 --> 00:00:58,860
probably wondering why there's five

21
00:00:55,920 --> 00:01:00,780
versions of uuids you're probably

22
00:00:58,860 --> 00:01:02,760
wondering why even though there's five

23
00:01:00,780 --> 00:01:05,339
versions every company invents their own

24
00:01:02,760 --> 00:01:07,439
thing and why we're making even more

25
00:01:05,339 --> 00:01:09,360
versions so

26
00:01:07,439 --> 00:01:11,100
this is a whole bunch of uuid version

27
00:01:09,360 --> 00:01:12,720
ones this is the original design they're

28
00:01:11,100 --> 00:01:14,820
supposed to sort of be a unique

29
00:01:12,720 --> 00:01:15,960
identifier that lasts forever and is

30
00:01:14,820 --> 00:01:18,360
never going to collide with anything

31
00:01:15,960 --> 00:01:19,979
else about half of it is a time stamp

32
00:01:18,360 --> 00:01:21,420
and about half of it is a node ID and

33
00:01:19,979 --> 00:01:22,860
the node ID is like the MAC address of

34
00:01:21,420 --> 00:01:25,200
your computer

35
00:01:22,860 --> 00:01:27,900
um and they decided to put the time

36
00:01:25,200 --> 00:01:30,840
stamp in like little Indian order so

37
00:01:27,900 --> 00:01:32,100
that these things would not like uh grow

38
00:01:30,840 --> 00:01:33,600
down the bottom they wouldn't get bigger

39
00:01:32,100 --> 00:01:35,340
over time

40
00:01:33,600 --> 00:01:37,380
um this is kind of what the schematic of

41
00:01:35,340 --> 00:01:38,640
them looks like so there's like a how

42
00:01:37,380 --> 00:01:40,979
about half of it is the time and about

43
00:01:38,640 --> 00:01:43,020
half of it is the node ID

44
00:01:40,979 --> 00:01:45,540
um they've been building them like this

45
00:01:43,020 --> 00:01:47,820
since the 1990s

46
00:01:45,540 --> 00:01:50,100
um uuid4 is the probably was probably

47
00:01:47,820 --> 00:01:51,720
the one that you've see used a lot these

48
00:01:50,100 --> 00:01:52,979
days it's the one that's entirely random

49
00:01:51,720 --> 00:01:56,520
it didn't bother with the timestamp and

50
00:01:52,979 --> 00:01:58,439
it didn't bother with um the node ID and

51
00:01:56,520 --> 00:02:00,360
so that's what these ones look like or

52
00:01:58,439 --> 00:02:02,399
these These are those ones so they're

53
00:02:00,360 --> 00:02:04,140
just completely random

54
00:02:02,399 --> 00:02:06,119
um very unguessable very sort of

55
00:02:04,140 --> 00:02:08,580
cryptographically secure quite you often

56
00:02:06,119 --> 00:02:10,380
used as like um session IDs and tokens

57
00:02:08,580 --> 00:02:12,300
like that um they still make pretty good

58
00:02:10,380 --> 00:02:14,220
primary keys for your database but

59
00:02:12,300 --> 00:02:16,379
they start falling over for reasons why

60
00:02:14,220 --> 00:02:18,739
which we'll get into

61
00:02:16,379 --> 00:02:21,480
um there's a couple other versions so um

62
00:02:18,739 --> 00:02:23,400
uuid 5 is a weird one that you can use

63
00:02:21,480 --> 00:02:25,140
for creating taking anything you've

64
00:02:23,400 --> 00:02:27,540
already got and turning it into a uuid

65
00:02:25,140 --> 00:02:29,520
so I create my own little namespace and

66
00:02:27,540 --> 00:02:31,379
in that namespace the string one becomes

67
00:02:29,520 --> 00:02:33,060
this uuid and the string two becomes

68
00:02:31,379 --> 00:02:34,860
this uuid

69
00:02:33,060 --> 00:02:36,840
um that can be really useful if you have

70
00:02:34,860 --> 00:02:38,400
like a uuid shaped hole and need to put

71
00:02:36,840 --> 00:02:39,840
your own kind of primary key or

72
00:02:38,400 --> 00:02:42,680
identifier into it and there's like

73
00:02:39,840 --> 00:02:44,580
well-defined namespaces for like DNS or

74
00:02:42,680 --> 00:02:45,840
x509 certificates whatever the hell

75
00:02:44,580 --> 00:02:48,060
those are

76
00:02:45,840 --> 00:02:49,440
um so in Python there's five versions

77
00:02:48,060 --> 00:02:51,180
version one is the one I just showed you

78
00:02:49,440 --> 00:02:52,319
version two doesn't exist don't use it

79
00:02:51,180 --> 00:02:53,940
it's a weird Microsoft thing version

80
00:02:52,319 --> 00:02:55,980
three is a bad version of version five

81
00:02:53,940 --> 00:02:57,120
version four is the random one that you

82
00:02:55,980 --> 00:03:00,300
probably see everywhere version five is

83
00:02:57,120 --> 00:03:02,700
the one I just showed you why

84
00:03:00,300 --> 00:03:04,920
do these why do why do these disappoint

85
00:03:02,700 --> 00:03:07,319
people why um whenever you get to a big

86
00:03:04,920 --> 00:03:09,780
company why did all of these guys you

87
00:03:07,319 --> 00:03:12,540
know Instagram Twitter uh Twitter like

88
00:03:09,780 --> 00:03:13,980
three times actually mongodb all of

89
00:03:12,540 --> 00:03:16,860
these people kind of invented their own

90
00:03:13,980 --> 00:03:19,260
versions of unique identifiers

91
00:03:16,860 --> 00:03:21,300
uh all of them looked at uuids and

92
00:03:19,260 --> 00:03:23,819
decided they just weren't any use

93
00:03:21,300 --> 00:03:25,920
and the reason for that is

94
00:03:23,819 --> 00:03:27,180
once you once you're talking about it a

95
00:03:25,920 --> 00:03:30,000
big scale

96
00:03:27,180 --> 00:03:32,700
databases don't really like primary keys

97
00:03:30,000 --> 00:03:34,620
that aren't sortable people actually

98
00:03:32,700 --> 00:03:36,659
want to be able to generate uids that

99
00:03:34,620 --> 00:03:37,980
don't Collide but they want them to sort

100
00:03:36,659 --> 00:03:40,500
of appear next to each other if they

101
00:03:37,980 --> 00:03:42,180
were generated roughly the same time

102
00:03:40,500 --> 00:03:44,879
um there was a few other weird decisions

103
00:03:42,180 --> 00:03:46,799
made like in uuid1 they use the

104
00:03:44,879 --> 00:03:48,299
Gregorian Epoch which is a bit annoying

105
00:03:46,799 --> 00:03:49,860
to calculate compared to a Unix

106
00:03:48,299 --> 00:03:51,180
timestamp people got Unix timestamps a

107
00:03:49,860 --> 00:03:53,640
lot more better

108
00:03:51,180 --> 00:03:55,620
um you have to parse the time sequence

109
00:03:53,640 --> 00:03:56,159
weirdly if you want to

110
00:03:55,620 --> 00:03:59,040
um

111
00:03:56,159 --> 00:04:00,299
and using the node ID was decided to be

112
00:03:59,040 --> 00:04:01,500
a bad idea because you were exposing the

113
00:04:00,299 --> 00:04:03,900
MAC address of whatever generated the

114
00:04:01,500 --> 00:04:05,640
uuids so the number one reason was

115
00:04:03,900 --> 00:04:07,560
everyone kind of abandoned these things

116
00:04:05,640 --> 00:04:10,260
because we wanted number to go up we

117
00:04:07,560 --> 00:04:11,580
have um the ietc or crypto Burrs in a

118
00:04:10,260 --> 00:04:14,340
way

119
00:04:11,580 --> 00:04:16,799
so they're inventing new ones and the

120
00:04:14,340 --> 00:04:18,479
new ones modeled after

121
00:04:16,799 --> 00:04:20,220
um all of those things that I just

122
00:04:18,479 --> 00:04:23,220
showed you all of those pretty much all

123
00:04:20,220 --> 00:04:25,139
of those alternative implementations

124
00:04:23,220 --> 00:04:27,060
basically had the Common Thread that

125
00:04:25,139 --> 00:04:28,919
they all went up over time they were all

126
00:04:27,060 --> 00:04:30,740
just orderable so the First new version

127
00:04:28,919 --> 00:04:34,080
uid V6

128
00:04:30,740 --> 00:04:36,780
is exactly the same as uuid1 but all the

129
00:04:34,080 --> 00:04:38,639
fields are kind of reorganized

130
00:04:36,780 --> 00:04:40,500
um they're just reshuffled so that you

131
00:04:38,639 --> 00:04:41,880
actually have the big end of the time at

132
00:04:40,500 --> 00:04:44,580
the start kind of like you know year

133
00:04:41,880 --> 00:04:45,780
month day instead of like seriously it

134
00:04:44,580 --> 00:04:47,639
was kind of like almost American style

135
00:04:45,780 --> 00:04:49,199
it's like Year Day month or some [ __ ]

136
00:04:47,639 --> 00:04:51,660
like that I don't know

137
00:04:49,199 --> 00:04:54,840
um and then uuid V7 is actually

138
00:04:51,660 --> 00:04:57,120
incredibly simple to implement and um

139
00:04:54,840 --> 00:04:58,259
is the stand the the standard that will

140
00:04:57,120 --> 00:04:59,940
probably be the one everyone expects you

141
00:04:58,259 --> 00:05:02,220
to use going forward where the first

142
00:04:59,940 --> 00:05:03,900
half of it is like the Unix timestamp in

143
00:05:02,220 --> 00:05:05,780
milliseconds with some sub millisecond

144
00:05:03,900 --> 00:05:07,979
Precision down to like four thousand uh

145
00:05:05,780 --> 00:05:09,600
nanoseconds on the python implementation

146
00:05:07,979 --> 00:05:11,160
and then the rest of it is just random

147
00:05:09,600 --> 00:05:12,120
data and there's a couple of things in

148
00:05:11,160 --> 00:05:14,699
there

149
00:05:12,120 --> 00:05:16,199
um the uui 87 is pretty cool and it's

150
00:05:14,699 --> 00:05:17,759
fun to play with and it was fun to

151
00:05:16,199 --> 00:05:19,680
implement and you should probably use it

152
00:05:17,759 --> 00:05:21,240
in database Keys going forward

153
00:05:19,680 --> 00:05:23,940
um and then finally they invented uuid

154
00:05:21,240 --> 00:05:25,080
V8 which is just screw it do it yourself

155
00:05:23,940 --> 00:05:26,460
um you can put your own version

156
00:05:25,080 --> 00:05:28,380
specifier in there and then you can fill

157
00:05:26,460 --> 00:05:30,360
it up with whatever information is

158
00:05:28,380 --> 00:05:32,460
useful for your application so long

159
00:05:30,360 --> 00:05:35,100
story short if you want good primary

160
00:05:32,460 --> 00:05:36,780
keys do you use the new uidv7 if you

161
00:05:35,100 --> 00:05:39,120
want utterly unguessable ones you use

162
00:05:36,780 --> 00:05:42,860
uuid4 if you're doing weird stuff you

163
00:05:39,120 --> 00:05:42,860
might want uuid5 thank you so much

164
00:05:43,320 --> 00:05:48,120
thanks Tom

165
00:05:46,139 --> 00:05:50,639
Tom was betting to me that he wouldn't

166
00:05:48,120 --> 00:05:52,560
uh he wouldn't finish in time so uh

167
00:05:50,639 --> 00:05:54,600
thanks Tom

168
00:05:52,560 --> 00:05:56,639
um so up on Deck we have uh Noah

169
00:05:54,600 --> 00:05:58,020
kantrowitz who I think has heard that

170
00:05:56,639 --> 00:06:01,440
he's busy taking a microphone off

171
00:05:58,020 --> 00:06:03,360
someone but first Alex Ware and Eleanor

172
00:06:01,440 --> 00:06:04,979
Williams awesome uh so we came up with

173
00:06:03,360 --> 00:06:07,740
this idea last night the slides did not

174
00:06:04,979 --> 00:06:10,020
exist two hours ago let's do this we had

175
00:06:07,740 --> 00:06:11,940
the uh Canberra python user organizers

176
00:06:10,020 --> 00:06:15,060
us and Andrew Hicks in the crowd shout

177
00:06:11,940 --> 00:06:17,160
out and also Jonah um so how did this

178
00:06:15,060 --> 00:06:20,340
all start it's like Perth python News

179
00:06:17,160 --> 00:06:21,180
Group many years ago uh the girls from a

180
00:06:20,340 --> 00:06:22,560
couple of girls from Europe were

181
00:06:21,180 --> 00:06:24,539
attaching of this event called Jenga

182
00:06:22,560 --> 00:06:26,100
girls that invented two pycon Europe EU

183
00:06:24,539 --> 00:06:28,199
we went we've got six weeks can we do

184
00:06:26,100 --> 00:06:30,780
one in Brisbane and we did

185
00:06:28,199 --> 00:06:32,460
um uh we went and did I went did a lot

186
00:06:30,780 --> 00:06:34,740
of familiar faces there by the way

187
00:06:32,460 --> 00:06:36,900
um uh did a bunch of them around the

188
00:06:34,740 --> 00:06:38,819
world uh there had it's been a

189
00:06:36,900 --> 00:06:40,800
tremendously successful initiative

190
00:06:38,819 --> 00:06:42,479
um like literally tens of thousands of

191
00:06:40,800 --> 00:06:44,639
people lots of events in uh the current

192
00:06:42,479 --> 00:06:46,919
president is from Zimbabwe more events

193
00:06:44,639 --> 00:06:48,780
in Africa than in America

194
00:06:46,919 --> 00:06:50,759
um and we're really proud to pay

195
00:06:48,780 --> 00:06:53,580
excellent resources really proud to be

196
00:06:50,759 --> 00:06:55,680
part of it but it is now 2023 I have an

197
00:06:53,580 --> 00:06:56,819
organizer and in a while because um I

198
00:06:55,680 --> 00:06:59,340
had a family

199
00:06:56,819 --> 00:07:01,199
um but uh we wanted to do some research

200
00:06:59,340 --> 00:07:03,479
and Outreach initiatives

201
00:07:01,199 --> 00:07:05,880
um and noticing that still like half the

202
00:07:03,479 --> 00:07:07,440
people here don't kind of look like us

203
00:07:05,880 --> 00:07:10,139
so there's this organization called 50

204
00:07:07,440 --> 00:07:12,660
50 at the Anu that we decided to do some

205
00:07:10,139 --> 00:07:14,340
stuff with uh yep and then from that

206
00:07:12,660 --> 00:07:16,979
discussion we came about of what we

207
00:07:14,340 --> 00:07:18,840
wanted to do next and uh there was yeah

208
00:07:16,979 --> 00:07:20,099
I've organized a bunch of these why

209
00:07:18,840 --> 00:07:22,380
don't we haven't done one in ages let's

210
00:07:20,099 --> 00:07:23,460
organize one of these in Canberra and

211
00:07:22,380 --> 00:07:25,020
one of the lovely things about the

212
00:07:23,460 --> 00:07:26,759
application process for Django girls was

213
00:07:25,020 --> 00:07:28,139
it was so fast that we had an event and

214
00:07:26,759 --> 00:07:29,460
like a website up and running before

215
00:07:28,139 --> 00:07:30,960
we'd actually thought through whether it

216
00:07:29,460 --> 00:07:32,160
was a good idea to plan one for eight

217
00:07:30,960 --> 00:07:33,780
weeks

218
00:07:32,160 --> 00:07:34,919
um but at this point it was real so we

219
00:07:33,780 --> 00:07:36,419
were doing it

220
00:07:34,919 --> 00:07:38,039
um luckily Canberra has a bunch of

221
00:07:36,419 --> 00:07:40,500
incredible organizations that we were

222
00:07:38,039 --> 00:07:42,840
able to just in a huge Panic email and

223
00:07:40,500 --> 00:07:44,639
ask for help and we got support uh

224
00:07:42,840 --> 00:07:45,780
involvement promotion from all of these

225
00:07:44,639 --> 00:07:47,340
different people if you are doing any

226
00:07:45,780 --> 00:07:48,900
kinds of Grassroots organizing

227
00:07:47,340 --> 00:07:50,460
definitely check who else in your

228
00:07:48,900 --> 00:07:51,960
community has already done stuff and

229
00:07:50,460 --> 00:07:53,280
shout it to Linux Australia if you

230
00:07:51,960 --> 00:07:55,020
aren't aware of the community grants

231
00:07:53,280 --> 00:07:56,479
program we had an absolutely incredible

232
00:07:55,020 --> 00:07:58,919
experience

233
00:07:56,479 --> 00:08:00,360
so lovely in supporting us through the

234
00:07:58,919 --> 00:08:01,800
process of applying for money and they

235
00:08:00,360 --> 00:08:03,720
allowed us to pay for Pizza on the day

236
00:08:01,800 --> 00:08:04,319
very important

237
00:08:03,720 --> 00:08:06,720
um

238
00:08:04,319 --> 00:08:08,639
as well in organizing something like

239
00:08:06,720 --> 00:08:11,460
this having a user group with a massive

240
00:08:08,639 --> 00:08:13,380
mailing list very handy uh one of the

241
00:08:11,460 --> 00:08:14,580
best experiences was when we sent out a

242
00:08:13,380 --> 00:08:17,340
big email to the mailing list going

243
00:08:14,580 --> 00:08:19,800
please help we promised there will be an

244
00:08:17,340 --> 00:08:21,240
event hopefully uh looking the next day

245
00:08:19,800 --> 00:08:22,740
at all of the different people who had

246
00:08:21,240 --> 00:08:24,780
signed up as mentors especially the

247
00:08:22,740 --> 00:08:26,160
people we didn't necessarily expect once

248
00:08:24,780 --> 00:08:27,780
we had all those people signed up we

249
00:08:26,160 --> 00:08:29,099
then had to pair them up into groups

250
00:08:27,780 --> 00:08:31,860
where we were looking at roughly a group

251
00:08:29,099 --> 00:08:33,360
of like three attendees to One Mentor so

252
00:08:31,860 --> 00:08:35,219
social engineering for good we were

253
00:08:33,360 --> 00:08:37,140
trying to match people on any kind of

254
00:08:35,219 --> 00:08:39,599
similar interests this experience

255
00:08:37,140 --> 00:08:41,760
greatly improved by being at a pub with

256
00:08:39,599 --> 00:08:44,940
beer and burgers as we're figuring out

257
00:08:41,760 --> 00:08:46,320
that problem uh also in terms of things

258
00:08:44,940 --> 00:08:47,880
you learn when organizing stuff we

259
00:08:46,320 --> 00:08:49,620
forgot to tell people what the end time

260
00:08:47,880 --> 00:08:52,200
of the event was turns out event

261
00:08:49,620 --> 00:08:54,839
schedules useful who knew

262
00:08:52,200 --> 00:08:56,010
um the other bit we almost stayed on

263
00:08:54,839 --> 00:08:59,260
budget

264
00:08:56,010 --> 00:08:59,260
[Music]

265
00:08:59,760 --> 00:09:03,959
sorry the event was a huge success like

266
00:09:01,920 --> 00:09:06,540
speaks for itself um we had a really

267
00:09:03,959 --> 00:09:08,100
great day we had like a 30 applicants of

268
00:09:06,540 --> 00:09:09,180
military applicants many people in the

269
00:09:08,100 --> 00:09:11,220
room and many people are off the photo

270
00:09:09,180 --> 00:09:13,680
here who didn't want to be part of it

271
00:09:11,220 --> 00:09:14,880
um we had a great time thanks to

272
00:09:13,680 --> 00:09:16,200
everyone who sponsored us but what are

273
00:09:14,880 --> 00:09:18,420
we doing next

274
00:09:16,200 --> 00:09:19,740
um I'm a web guy uh Alex is a data

275
00:09:18,420 --> 00:09:21,660
person I don't we want to alternate

276
00:09:19,740 --> 00:09:23,459
between a web and a data and science

277
00:09:21,660 --> 00:09:25,140
sort of thing so we want to do one in

278
00:09:23,459 --> 00:09:26,640
six months the software carpentry we'd

279
00:09:25,140 --> 00:09:28,440
never run one of them so if anyone here

280
00:09:26,640 --> 00:09:30,000
has done a software commentary Workshop

281
00:09:28,440 --> 00:09:31,500
please can you come talk to us we want

282
00:09:30,000 --> 00:09:35,100
to know how to do it

283
00:09:31,500 --> 00:09:37,380
um so we would love help yes secondly

284
00:09:35,100 --> 00:09:38,519
um oh so we created a meetup group so if

285
00:09:37,380 --> 00:09:39,660
you're in around Canberra and you would

286
00:09:38,519 --> 00:09:41,279
like to be involved in one of these

287
00:09:39,660 --> 00:09:42,959
events in future are we really looking

288
00:09:41,279 --> 00:09:44,880
to just try and improve gender diversity

289
00:09:42,959 --> 00:09:46,260
in Tech um so while we do use women in

290
00:09:44,880 --> 00:09:48,720
the name uh we are certainly not

291
00:09:46,260 --> 00:09:50,160
limiting and we want any capacity you've

292
00:09:48,720 --> 00:09:52,019
thought that there was a space for

293
00:09:50,160 --> 00:09:53,279
another thing in between like women in

294
00:09:52,019 --> 00:09:55,140
big data and all of those sorts of

295
00:09:53,279 --> 00:09:57,360
things and the python user group that

296
00:09:55,140 --> 00:09:58,380
there's just you know safety in numbers

297
00:09:57,360 --> 00:10:00,060
guys

298
00:09:58,380 --> 00:10:01,800
um and also I wanted to shout out hi

299
00:10:00,060 --> 00:10:03,899
ladies con I actually have no idea what

300
00:10:01,800 --> 00:10:05,279
this is but it looks really fun and I

301
00:10:03,899 --> 00:10:06,899
just it hadn't been chatted out so I'm

302
00:10:05,279 --> 00:10:11,580
doing that now

303
00:10:06,899 --> 00:10:13,320
um and finally uh yeah uh so the Django

304
00:10:11,580 --> 00:10:14,640
girls Adelaide is happening tomorrow uh

305
00:10:13,320 --> 00:10:16,320
we're not directly involved but we are

306
00:10:14,640 --> 00:10:17,880
highly supportive we're super excited to

307
00:10:16,320 --> 00:10:19,320
see what is going on

308
00:10:17,880 --> 00:10:21,360
um we believe that there will be some

309
00:10:19,320 --> 00:10:23,399
information through the Discord

310
00:10:21,360 --> 00:10:25,320
um but go look for row at the Kraken

311
00:10:23,399 --> 00:10:26,820
table they are still looking for some

312
00:10:25,320 --> 00:10:27,660
mentors so if this is the sort of thing

313
00:10:26,820 --> 00:10:28,800
you would like to help out with

314
00:10:27,660 --> 00:10:30,720
especially if you've got web dev

315
00:10:28,800 --> 00:10:32,760
experience but certainly not limited in

316
00:10:30,720 --> 00:10:34,260
any capacity or if you're interested in

317
00:10:32,760 --> 00:10:35,580
knowing how these kind of the Dynamics

318
00:10:34,260 --> 00:10:37,080
of these kind of events work because

319
00:10:35,580 --> 00:10:38,760
they're like really successful we should

320
00:10:37,080 --> 00:10:40,440
be doing more Outreach you know getting

321
00:10:38,760 --> 00:10:42,000
us all together

322
00:10:40,440 --> 00:10:43,980
um so go check that out tomorrow if

323
00:10:42,000 --> 00:10:47,120
you're here for the Sprints and we've

324
00:10:43,980 --> 00:10:47,120
got like time to spare

325
00:10:47,399 --> 00:10:50,399
amazing

326
00:10:52,320 --> 00:10:58,079
you have a choice Noah

327
00:10:55,019 --> 00:11:00,180
uh you either present without video or

328
00:10:58,079 --> 00:11:01,980
banno does interpretive dance for the

329
00:11:00,180 --> 00:11:04,820
three minutes while you take your laptop

330
00:11:01,980 --> 00:11:04,820
to get configured

331
00:11:05,399 --> 00:11:07,560
dance

332
00:11:06,600 --> 00:11:08,180
[Applause]

333
00:11:07,560 --> 00:11:11,180
dance

334
00:11:08,180 --> 00:11:11,180
okay

335
00:11:11,240 --> 00:11:16,560
if you can get set up on this side

336
00:11:13,560 --> 00:11:17,579
you're on deck and uh I'll start the

337
00:11:16,560 --> 00:11:19,800
clock

338
00:11:17,579 --> 00:11:20,940
now we're cutting whenever you're ready

339
00:11:19,800 --> 00:11:22,980
all right I'm going to give this one

340
00:11:20,940 --> 00:11:26,820
last shot and then we're just going to

341
00:11:22,980 --> 00:11:31,940
listen to my talk which will not be as

342
00:11:26,820 --> 00:11:31,940
interesting but Banner fill in the time

343
00:11:32,399 --> 00:11:42,970
nope doesn't like it

344
00:11:34,140 --> 00:11:46,090
[Applause]

345
00:11:42,970 --> 00:11:46,090
[Music]

346
00:11:46,339 --> 00:11:49,920
nope

347
00:11:48,180 --> 00:11:52,620
all right

348
00:11:49,920 --> 00:11:55,200
Noah all right

349
00:11:52,620 --> 00:11:56,760
so you can't see my slides but I'm going

350
00:11:55,200 --> 00:11:58,140
to talk about how to look at space and I

351
00:11:56,760 --> 00:11:59,760
guess I'm just going to read out some of

352
00:11:58,140 --> 00:12:03,839
the URLs and these will be posted onto

353
00:11:59,760 --> 00:12:06,540
Discord later uh all right uh so the

354
00:12:03,839 --> 00:12:07,800
first thing uh maybe I just turned my

355
00:12:06,540 --> 00:12:10,800
screen around

356
00:12:07,800 --> 00:12:13,019
let's do that

357
00:12:10,800 --> 00:12:15,000
[Applause]

358
00:12:13,019 --> 00:12:17,700
all right

359
00:12:15,000 --> 00:12:20,339
yeah well it's mostly black all right uh

360
00:12:17,700 --> 00:12:23,160
so things so the first one this is the

361
00:12:20,339 --> 00:12:24,779
NASA eyes orery uh this is a great way

362
00:12:23,160 --> 00:12:26,160
to see a 3D representation of everything

363
00:12:24,779 --> 00:12:28,860
that is currently in our solar system

364
00:12:26,160 --> 00:12:30,779
planets missions comets all the 3D

365
00:12:28,860 --> 00:12:34,140
representations of everything

366
00:12:30,779 --> 00:12:35,700
um it is at eyes.nasa.gov app slash Ori

367
00:12:34,140 --> 00:12:36,959
I'm not going to read all of these URLs

368
00:12:35,700 --> 00:12:39,360
out because some of them do get long and

369
00:12:36,959 --> 00:12:41,220
weird all right the next thing the NASA

370
00:12:39,360 --> 00:12:43,440
deep space Network so this is an array

371
00:12:41,220 --> 00:12:44,519
of multiple radio telescopes around the

372
00:12:43,440 --> 00:12:47,399
world one of which is actually in

373
00:12:44,519 --> 00:12:49,260
Canberra uh that gives a full coverage

374
00:12:47,399 --> 00:12:50,639
view of all deep space missions

375
00:12:49,260 --> 00:12:52,139
throughout the solar system and there is

376
00:12:50,639 --> 00:12:54,360
a real-time view of everything it is

377
00:12:52,139 --> 00:12:57,060
connected to including a fun thing if

378
00:12:54,360 --> 00:12:58,440
you go down to the extended view you can

379
00:12:57,060 --> 00:12:59,760
see the real-time data rates this was

380
00:12:58,440 --> 00:13:02,040
actually a couple days ago when it was

381
00:12:59,760 --> 00:13:03,660
talking to Juno outside Jupiter and you

382
00:13:02,040 --> 00:13:06,060
cannot see but if someone very close to

383
00:13:03,660 --> 00:13:08,040
uh maybe could it's getting 200 kilobits

384
00:13:06,060 --> 00:13:09,300
per second from Jupiter the fact that I

385
00:13:08,040 --> 00:13:12,420
cannot get that from my mobile phone

386
00:13:09,300 --> 00:13:14,339
very frequently is a very sad thing uh

387
00:13:12,420 --> 00:13:15,720
all right but that's just sort of the

388
00:13:14,339 --> 00:13:18,300
general idea of space so we want to look

389
00:13:15,720 --> 00:13:20,220
at actual space a service called Mast

390
00:13:18,300 --> 00:13:21,660
this is a database that contains

391
00:13:20,220 --> 00:13:24,060
information from most of our space

392
00:13:21,660 --> 00:13:27,320
telescopes so James Webb Hubble Spitzer

393
00:13:24,060 --> 00:13:31,860
Kepler all in this service called Mast

394
00:13:27,320 --> 00:13:32,820
also it has an API Astro query.mast so

395
00:13:31,860 --> 00:13:35,279
if you want to play with any of this

396
00:13:32,820 --> 00:13:37,980
data you can use that

397
00:13:35,279 --> 00:13:39,600
next up Helio viewer helioviewer.org

398
00:13:37,980 --> 00:13:41,639
this is a site that Aggregates a lot of

399
00:13:39,600 --> 00:13:43,380
information about solar observation

400
00:13:41,639 --> 00:13:44,880
satellites the one that's currently most

401
00:13:43,380 --> 00:13:46,980
interesting and still active is the

402
00:13:44,880 --> 00:13:49,200
Solar Dynamics Observatory but the old

403
00:13:46,980 --> 00:13:50,519
stereo a spacecraft is still there just

404
00:13:49,200 --> 00:13:53,040
most of its instruments are offline but

405
00:13:50,519 --> 00:13:55,200
still makes really pretty pictures

406
00:13:53,040 --> 00:13:56,880
uh all right the bomb if you want to

407
00:13:55,200 --> 00:13:58,920
figure out what weather is happening in

408
00:13:56,880 --> 00:14:02,519
Space the bomb has a space weather

409
00:13:58,920 --> 00:14:03,779
prediction center forecast Center this

410
00:14:02,519 --> 00:14:05,639
gives you information about incoming

411
00:14:03,779 --> 00:14:08,459
geometrical uh yeah that's my company

412
00:14:05,639 --> 00:14:10,560
name geomagnetic events

413
00:14:08,459 --> 00:14:12,779
and also a fun thing it has an aurora

414
00:14:10,560 --> 00:14:14,700
prediction map uh it's probably never

415
00:14:12,779 --> 00:14:16,079
going to get quite this far north but it

416
00:14:14,700 --> 00:14:17,459
happened to be in Tasmania and you want

417
00:14:16,079 --> 00:14:19,920
to see is there going to be an aurora

418
00:14:17,459 --> 00:14:22,800
visible they've got a prediction for you

419
00:14:19,920 --> 00:14:25,139
all right next up Mars so the best way I

420
00:14:22,800 --> 00:14:27,959
know if to get Mars vid Mars pictures is

421
00:14:25,139 --> 00:14:30,779
from the NASA Mission sites

422
00:14:27,959 --> 00:14:33,360
so these give you uh the sort of process

423
00:14:30,779 --> 00:14:35,700
jpegs these are not science quality but

424
00:14:33,360 --> 00:14:37,680
this is what Mars looked like about 12

425
00:14:35,700 --> 00:14:40,800
hours ago this is from the I believe

426
00:14:37,680 --> 00:14:43,139
left nav cam on perseverance and this is

427
00:14:40,800 --> 00:14:45,180
from the right nav cam on curiosity

428
00:14:43,139 --> 00:14:46,440
three days ago

429
00:14:45,180 --> 00:14:47,880
if you want to get the real science

430
00:14:46,440 --> 00:14:50,160
quality images there's a program called

431
00:14:47,880 --> 00:14:52,440
Mars viewer it is a really gnarly Java

432
00:14:50,160 --> 00:14:55,079
app but it will give you the raw data if

433
00:14:52,440 --> 00:14:56,820
you know what you're looking for

434
00:14:55,079 --> 00:14:59,220
if you want something that isn't or

435
00:14:56,820 --> 00:15:01,800
isn't the Sun or Mars the next step is

436
00:14:59,220 --> 00:15:03,540
the NASA PDS image Atlas so this gives

437
00:15:01,800 --> 00:15:05,699
you basically every other spacecraft

438
00:15:03,540 --> 00:15:07,380
that NASA is in charge of

439
00:15:05,699 --> 00:15:08,699
um fun ones include the Mars

440
00:15:07,380 --> 00:15:10,740
reconnaissance Orbiter they just have

441
00:15:08,699 --> 00:15:12,839
these giant track images they're all

442
00:15:10,740 --> 00:15:15,540
pngs relatively easy to work with that

443
00:15:12,839 --> 00:15:17,040
are just scans of the surface of Mars

444
00:15:15,540 --> 00:15:18,899
similarly there's the planetary science

445
00:15:17,040 --> 00:15:20,519
archive from the esa so this is for all

446
00:15:18,899 --> 00:15:22,320
spacecraft that the European Space

447
00:15:20,519 --> 00:15:23,399
Agency is in charge of such as beppy

448
00:15:22,320 --> 00:15:25,500
Colombo

449
00:15:23,399 --> 00:15:27,180
and if you want to actually like look at

450
00:15:25,500 --> 00:15:29,279
things with your face you'll need to

451
00:15:27,180 --> 00:15:31,440
know where they are uh now for big

452
00:15:29,279 --> 00:15:33,480
things like say Jupiter or stars or

453
00:15:31,440 --> 00:15:35,279
galaxies that don't move very often you

454
00:15:33,480 --> 00:15:37,019
can find those very easily online but if

455
00:15:35,279 --> 00:15:38,399
you want to look at say a satellite you

456
00:15:37,019 --> 00:15:39,720
need its orbital information and the

457
00:15:38,399 --> 00:15:41,820
place you get that is a website called

458
00:15:39,720 --> 00:15:43,560
sellustrack the story of this one's

459
00:15:41,820 --> 00:15:45,480
actually really funny uh it originally

460
00:15:43,560 --> 00:15:47,940
started with a guy who's running a BBS

461
00:15:45,480 --> 00:15:50,339
in the early 90s and he would call into

462
00:15:47,940 --> 00:15:52,320
the US military once a day and request

463
00:15:50,339 --> 00:15:55,380
military information and they would just

464
00:15:52,320 --> 00:15:57,300
give it to him and eventually just

465
00:15:55,380 --> 00:15:58,620
became so important that when the office

466
00:15:57,300 --> 00:16:00,779
that was giving the information was shut

467
00:15:58,620 --> 00:16:03,120
down no rad uh just started sending him

468
00:16:00,779 --> 00:16:06,120
the data directly because everyone was

469
00:16:03,120 --> 00:16:07,740
dependent on his website it also has a

470
00:16:06,120 --> 00:16:09,660
nice 3D viewer so if you want to do that

471
00:16:07,740 --> 00:16:12,300
you can see 3D positions of satellites

472
00:16:09,660 --> 00:16:13,860
and also it has a python API in a

473
00:16:12,300 --> 00:16:14,699
library called skyfield thank you very

474
00:16:13,860 --> 00:16:17,000
much

475
00:16:14,699 --> 00:16:17,000
foreign

476
00:16:19,440 --> 00:16:22,920
here in the python world we all like to

477
00:16:21,180 --> 00:16:26,420
think that space is significant so uh

478
00:16:22,920 --> 00:16:26,420
thank you Noah for giving us more space

479
00:16:28,560 --> 00:16:32,180
oh we gave Libby a mic did we

480
00:16:32,579 --> 00:16:38,339
hello

481
00:16:34,920 --> 00:16:40,440
the voice of Bob Libby Berry everyone uh

482
00:16:38,339 --> 00:16:44,840
up on Deck we have Matthew Perry but

483
00:16:40,440 --> 00:16:50,519
first a first-time speaker Anisha sarkar

484
00:16:44,840 --> 00:16:53,060
[Applause]

485
00:16:50,519 --> 00:16:53,060
ting

486
00:16:53,579 --> 00:16:58,139
um my name is Anisha sarkar and I'm a

487
00:16:55,740 --> 00:17:01,139
very passionate student about natural

488
00:16:58,139 --> 00:17:03,000
language processing but at this picon I

489
00:17:01,139 --> 00:17:04,799
learned about multiple tools especially

490
00:17:03,000 --> 00:17:07,020
about guis

491
00:17:04,799 --> 00:17:09,720
so I decided to implement it because

492
00:17:07,020 --> 00:17:11,400
there's no better way to test my

493
00:17:09,720 --> 00:17:14,819
understanding other than to do a

494
00:17:11,400 --> 00:17:16,980
practical project so ever since I came

495
00:17:14,819 --> 00:17:19,439
to Australia I was intrigued by the way

496
00:17:16,980 --> 00:17:22,980
pedestrian buttons worked and the very

497
00:17:19,439 --> 00:17:25,319
cool music that came along with it

498
00:17:22,980 --> 00:17:27,299
this form the basis of my project to

499
00:17:25,319 --> 00:17:30,000
simulate the traffic signals using Code

500
00:17:27,299 --> 00:17:32,220
so how do the traffic lights work we've

501
00:17:30,000 --> 00:17:34,380
got buttons we've got timer and we've

502
00:17:32,220 --> 00:17:36,240
got sound when you press the button if

503
00:17:34,380 --> 00:17:39,120
your luck is good the green line turns

504
00:17:36,240 --> 00:17:41,460
on right away and then you can cross and

505
00:17:39,120 --> 00:17:43,980
when in that process there's also sound

506
00:17:41,460 --> 00:17:45,660
so so that people can know like people

507
00:17:43,980 --> 00:17:48,419
can just hear and know that the lights

508
00:17:45,660 --> 00:17:50,760
on and now can I simulate this on

509
00:17:48,419 --> 00:17:53,039
computer of course I can at least a

510
00:17:50,760 --> 00:17:55,440
simple version I decided to start this

511
00:17:53,039 --> 00:18:00,679
project during pycon AU

512
00:17:55,440 --> 00:18:02,580
[Applause]

513
00:18:00,679 --> 00:18:04,080
because I know when I go back to

514
00:18:02,580 --> 00:18:06,480
University I probably wouldn't have the

515
00:18:04,080 --> 00:18:08,520
time my workflow I would have the scope

516
00:18:06,480 --> 00:18:10,260
and prerequisites uh then I would build

517
00:18:08,520 --> 00:18:12,660
a dual light system which is a red light

518
00:18:10,260 --> 00:18:15,360
and the green light add some timer build

519
00:18:12,660 --> 00:18:18,059
the Dual light system sound get some

520
00:18:15,360 --> 00:18:20,940
sound test it and fix it and also most

521
00:18:18,059 --> 00:18:24,419
importantly make the documentation

522
00:18:20,940 --> 00:18:27,480
so the tools I used were tikenter play

523
00:18:24,419 --> 00:18:29,400
sound to play the audio a timer and

524
00:18:27,480 --> 00:18:31,620
threading so that multiple processes

525
00:18:29,400 --> 00:18:33,120
could happen at the same time I'll show

526
00:18:31,620 --> 00:18:35,100
you the demo real soon but I wanted to

527
00:18:33,120 --> 00:18:36,660
go through my slides real quickly

528
00:18:35,100 --> 00:18:38,280
um in the future I would like to

529
00:18:36,660 --> 00:18:40,260
implement so during the springtime which

530
00:18:38,280 --> 00:18:42,299
I'll be here for uh have a mock

531
00:18:40,260 --> 00:18:44,400
implementation using the adult Greenman

532
00:18:42,299 --> 00:18:47,640
proposal which I could probably link to

533
00:18:44,400 --> 00:18:49,799
it in the GitHub uh sorry in the Discord

534
00:18:47,640 --> 00:18:51,900
um and finally I would also like to have

535
00:18:49,799 --> 00:18:53,940
some pedestrian lights turn green right

536
00:18:51,900 --> 00:18:55,559
away when the button is pressed so I

537
00:18:53,940 --> 00:18:57,980
would also like to randomize the process

538
00:18:55,559 --> 00:18:57,980
a bit

539
00:18:58,799 --> 00:19:03,919
um time for the demo it's not too long

540
00:19:01,620 --> 00:19:03,919
but

541
00:19:13,220 --> 00:19:29,910
[Applause]

542
00:19:27,010 --> 00:19:31,120
[Laughter]

543
00:19:29,910 --> 00:19:31,760
[Applause]

544
00:19:31,120 --> 00:19:35,490
[Music]

545
00:19:31,760 --> 00:19:35,490
[Applause]

546
00:19:42,420 --> 00:19:47,089
[Applause]

547
00:19:49,620 --> 00:19:55,679
I'm so grateful uh that it got to be

548
00:19:52,919 --> 00:19:58,740
here today at Phi Khan and learn so much

549
00:19:55,679 --> 00:20:01,140
through these talks and I hope I'll be

550
00:19:58,740 --> 00:20:04,080
able to come to more attend more pycons

551
00:20:01,140 --> 00:20:07,280
learn more and just try out new things

552
00:20:04,080 --> 00:20:07,280
thank you faicon

553
00:20:10,740 --> 00:20:15,900
thank you that was fantastic

554
00:20:12,679 --> 00:20:18,660
uh Tessa Bradbury you are up on Deck

555
00:20:15,900 --> 00:20:20,880
first it does look like this lectern is

556
00:20:18,660 --> 00:20:23,900
actually working oh so uh please welcome

557
00:20:20,880 --> 00:20:23,900
Matthew Perry

558
00:20:25,520 --> 00:20:29,039
cool hi everyone

559
00:20:27,600 --> 00:20:30,660
um so today I'm going to talk about how

560
00:20:29,039 --> 00:20:33,600
you can make your tests five times

561
00:20:30,660 --> 00:20:34,919
faster by parallelizing them so why

562
00:20:33,600 --> 00:20:36,900
would you want to do parallel testing

563
00:20:34,919 --> 00:20:38,880
well if you run on a single core then

564
00:20:36,900 --> 00:20:40,679
you have a single test queue and the

565
00:20:38,880 --> 00:20:43,080
test runs sequentially so you it's quite

566
00:20:40,679 --> 00:20:44,520
slow okay but if you have multiple calls

567
00:20:43,080 --> 00:20:46,559
on your computer which most do these

568
00:20:44,520 --> 00:20:48,539
days then you can have multiple test

569
00:20:46,559 --> 00:20:50,700
queues and split the tests to run across

570
00:20:48,539 --> 00:20:51,840
each core across your computer and make

571
00:20:50,700 --> 00:20:53,820
it faster

572
00:20:51,840 --> 00:20:56,340
so what are some of the issues doing

573
00:20:53,820 --> 00:20:58,500
this within Django So within Django

574
00:20:56,340 --> 00:21:00,240
there's no inbuilt cache isolation so

575
00:20:58,500 --> 00:21:02,400
you have this lovely class called the

576
00:21:00,240 --> 00:21:04,320
transaction test case within Django for

577
00:21:02,400 --> 00:21:06,539
testing which means that your database

578
00:21:04,320 --> 00:21:08,460
transactions are isolated so this means

579
00:21:06,539 --> 00:21:09,780
that if you don't have a cache within

580
00:21:08,460 --> 00:21:12,480
your application then you can probably

581
00:21:09,780 --> 00:21:14,100
just turn on Parallel testing and it's

582
00:21:12,480 --> 00:21:15,840
not going to have isolation issues there

583
00:21:14,100 --> 00:21:17,400
won't be bugs and it's going to be a lot

584
00:21:15,840 --> 00:21:20,039
quicker very quickly

585
00:21:17,400 --> 00:21:22,440
but in our case we had use a login

586
00:21:20,039 --> 00:21:24,419
session stored within the cache and then

587
00:21:22,440 --> 00:21:26,400
what that means is that because a lot of

588
00:21:24,419 --> 00:21:28,380
our tests would clear the cache at the

589
00:21:26,400 --> 00:21:30,000
start of every setup method that it

590
00:21:28,380 --> 00:21:32,580
would clear those login details those

591
00:21:30,000 --> 00:21:34,559
login session details from the cache and

592
00:21:32,580 --> 00:21:35,940
then the test would fail incorrectly so

593
00:21:34,559 --> 00:21:37,620
you might have 100 tests failing

594
00:21:35,940 --> 00:21:39,720
randomly at the same time

595
00:21:37,620 --> 00:21:42,659
you can also have issues with unisolated

596
00:21:39,720 --> 00:21:45,059
tests so for example tests relying on

597
00:21:42,659 --> 00:21:46,440
state outside of the test classes and

598
00:21:45,059 --> 00:21:48,299
tests relying on the order in which

599
00:21:46,440 --> 00:21:50,280
they're running

600
00:21:48,299 --> 00:21:52,620
um so another aspect which made it a bit

601
00:21:50,280 --> 00:21:55,080
more complex here is we were using

602
00:21:52,620 --> 00:21:56,700
features specific to the redis cache so

603
00:21:55,080 --> 00:21:58,860
for most use cases you would probably

604
00:21:56,700 --> 00:22:00,960
just want to mock the cache with a local

605
00:21:58,860 --> 00:22:03,360
memory cache because those are isolated

606
00:22:00,960 --> 00:22:05,100
automatically but the redest occasions

607
00:22:03,360 --> 00:22:06,840
are isolated you're working with an

608
00:22:05,100 --> 00:22:09,600
instance really so you need to build

609
00:22:06,840 --> 00:22:11,039
cache isolation into it

610
00:22:09,600 --> 00:22:13,080
um another aspect was I didn't want to

611
00:22:11,039 --> 00:22:15,179
just Mock and ignore the cache because

612
00:22:13,080 --> 00:22:16,860
we had about 3 200 tests and several

613
00:22:15,179 --> 00:22:19,080
hundred thousand lines of python code

614
00:22:16,860 --> 00:22:20,700
and I wanted to actually test the cages

615
00:22:19,080 --> 00:22:22,919
being used correctly

616
00:22:20,700 --> 00:22:25,440
so my solution was to have essentially a

617
00:22:22,919 --> 00:22:27,960
different case database uh per CPU

618
00:22:25,440 --> 00:22:30,179
that's running tests

619
00:22:27,960 --> 00:22:33,360
so you can you do this using a library

620
00:22:30,179 --> 00:22:36,659
called ply test exist so very easy to

621
00:22:33,360 --> 00:22:38,039
pip install and you specify the number

622
00:22:36,659 --> 00:22:39,600
of cores you're going to be using on

623
00:22:38,039 --> 00:22:42,000
your computer using the dash in argument

624
00:22:39,600 --> 00:22:43,500
so if you use the auto keyword that's

625
00:22:42,000 --> 00:22:45,780
going to use all the cores on your

626
00:22:43,500 --> 00:22:48,179
computer you can also specify the number

627
00:22:45,780 --> 00:22:50,400
of cores that you want to use in our

628
00:22:48,179 --> 00:22:52,919
case because there's a maximum of 16

629
00:22:50,400 --> 00:22:55,799
redis databases on a various instance by

630
00:22:52,919 --> 00:22:58,400
default we capped it to 16 and specified

631
00:22:55,799 --> 00:23:01,260
to use that number of cores

632
00:22:58,400 --> 00:23:03,419
so one of the main classes used in the

633
00:23:01,260 --> 00:23:05,640
implementation here was the redis

634
00:23:03,419 --> 00:23:07,740
manager which is essentially a class

635
00:23:05,640 --> 00:23:10,799
pretending or a Singleton pretending to

636
00:23:07,740 --> 00:23:12,480
be a redis cache but it's actually when

637
00:23:10,799 --> 00:23:15,020
when you run a method for the redis

638
00:23:12,480 --> 00:23:18,480
cache class it's acquiring a lock

639
00:23:15,020 --> 00:23:21,419
setting the database for the cache

640
00:23:18,480 --> 00:23:24,059
object according to the apply test Pi

641
00:23:21,419 --> 00:23:25,559
test Exodus worker ID so that means

642
00:23:24,059 --> 00:23:27,480
every

643
00:23:25,559 --> 00:23:28,919
test queue is going to be working with a

644
00:23:27,480 --> 00:23:30,900
different case database

645
00:23:28,919 --> 00:23:32,820
then we get the result from running the

646
00:23:30,900 --> 00:23:34,679
corresponding method release the lock

647
00:23:32,820 --> 00:23:36,120
and return that result

648
00:23:34,679 --> 00:23:38,159
um so one thing you might notice here is

649
00:23:36,120 --> 00:23:40,559
that this is pretty inefficient

650
00:23:38,159 --> 00:23:41,940
um so you every single time you run a

651
00:23:40,559 --> 00:23:45,600
redis method you're going to be running

652
00:23:41,940 --> 00:23:46,919
a method to change the database but you

653
00:23:45,600 --> 00:23:48,840
still get a very nice speed up

654
00:23:46,919 --> 00:23:50,520
regardless so if anyone that has done

655
00:23:48,840 --> 00:23:51,960
this before I'd be really Keen to know

656
00:23:50,520 --> 00:23:53,039
how you got around this if you've done

657
00:23:51,960 --> 00:23:55,260
so

658
00:23:53,039 --> 00:23:57,419
so this is essentially a picture of the

659
00:23:55,260 --> 00:23:59,940
architecture I settled with at the end

660
00:23:57,419 --> 00:24:01,799
you essentially have that single turn

661
00:23:59,940 --> 00:24:03,179
for a rediscation manager pretending to

662
00:24:01,799 --> 00:24:04,559
be a reddish cache but it's just

663
00:24:03,179 --> 00:24:06,659
redirecting the request to the

664
00:24:04,559 --> 00:24:09,600
corresponding database

665
00:24:06,659 --> 00:24:12,419
so additionally to this we also have a

666
00:24:09,600 --> 00:24:14,820
within conf test we set up load file

667
00:24:12,419 --> 00:24:17,039
scheduling using the pi test xdesmax

668
00:24:14,820 --> 00:24:18,120
scheduler you can also use load scope

669
00:24:17,039 --> 00:24:20,220
scheduling

670
00:24:18,120 --> 00:24:22,500
so load far scheduling means that every

671
00:24:20,220 --> 00:24:25,080
test file is going to be associated with

672
00:24:22,500 --> 00:24:27,120
just one CPU okay so this just made it

673
00:24:25,080 --> 00:24:28,679
simpler to set this all up but if you

674
00:24:27,120 --> 00:24:30,600
wanted to have a small speed up and get

675
00:24:28,679 --> 00:24:33,539
rid of any isolation issues you could

676
00:24:30,600 --> 00:24:35,039
move to large scope scheduling also I

677
00:24:33,539 --> 00:24:37,140
had to add a couple of fixtures to avoid

678
00:24:35,039 --> 00:24:41,280
breaking the coverage and we essentially

679
00:24:37,140 --> 00:24:43,100
inject the redis Singleton through the

680
00:24:41,280 --> 00:24:46,940
import system by replacing

681
00:24:43,100 --> 00:24:46,940
django.call.cash and django.com

682
00:24:47,220 --> 00:24:51,659
um so the eventual results was the tests

683
00:24:49,740 --> 00:24:54,179
became about five times faster to

684
00:24:51,659 --> 00:24:55,620
execute that's using eight cores

685
00:24:54,179 --> 00:24:57,360
um getting rid of that database

686
00:24:55,620 --> 00:24:59,880
switching aspect might make it a lot

687
00:24:57,360 --> 00:25:02,640
faster we also have the capacity to

688
00:24:59,880 --> 00:25:04,320
scale it up with a 72 core machine uh

689
00:25:02,640 --> 00:25:06,600
it's going to be easier to detect test

690
00:25:04,320 --> 00:25:08,039
isolation issues in future if we uh

691
00:25:06,600 --> 00:25:09,840
randomize the order in which they run

692
00:25:08,039 --> 00:25:11,280
more and we don't have as much of a test

693
00:25:09,840 --> 00:25:12,419
budget anymore which is really great we

694
00:25:11,280 --> 00:25:15,080
can test more

695
00:25:12,419 --> 00:25:15,080
thanks everyone

696
00:25:15,659 --> 00:25:18,840
thanks Matthew

697
00:25:17,340 --> 00:25:20,700
uh

698
00:25:18,840 --> 00:25:24,240
so now I'm going to try and say Pi test

699
00:25:20,700 --> 00:25:25,440
xdisk five times quickly

700
00:25:24,240 --> 00:25:28,799
or not

701
00:25:25,440 --> 00:25:31,380
um do it Lachlan Jacob is up on Deck but

702
00:25:28,799 --> 00:25:32,820
first uh Tessa Bradbury with a uh with a

703
00:25:31,380 --> 00:25:34,620
war story

704
00:25:32,820 --> 00:25:37,320
so hello everyone

705
00:25:34,620 --> 00:25:40,260
so I've been a web developer and an SRE

706
00:25:37,320 --> 00:25:41,700
for a decade and about a year ago I

707
00:25:40,260 --> 00:25:42,659
brought down production for the first

708
00:25:41,700 --> 00:25:44,000
time

709
00:25:42,659 --> 00:25:46,799
oh

710
00:25:44,000 --> 00:25:48,299
there's a bunch of technical details in

711
00:25:46,799 --> 00:25:50,100
the public postmodern we wrote up but

712
00:25:48,299 --> 00:25:52,320
behind every incident there are humans

713
00:25:50,100 --> 00:25:53,159
so this is my story of what that was

714
00:25:52,320 --> 00:25:55,020
like

715
00:25:53,159 --> 00:25:57,720
so I've had the power to break

716
00:25:55,020 --> 00:26:00,120
everything in so many ways for so many

717
00:25:57,720 --> 00:26:02,520
years and I've seen so many incidents

718
00:26:00,120 --> 00:26:05,100
unfold but this was the first time it's

719
00:26:02,520 --> 00:26:07,080
been me who clicked the button and it

720
00:26:05,100 --> 00:26:09,120
was surprisingly weird and complex thing

721
00:26:07,080 --> 00:26:11,580
to unpack like I've been around

722
00:26:09,120 --> 00:26:13,679
reliability and resilient circles for

723
00:26:11,580 --> 00:26:15,600
long enough not to blame the person who

724
00:26:13,679 --> 00:26:18,539
triggered the outage even when it was me

725
00:26:15,600 --> 00:26:20,820
but it's always far more useful to

726
00:26:18,539 --> 00:26:22,679
explore why it happened and not just the

727
00:26:20,820 --> 00:26:26,580
technical details but the human and

728
00:26:22,679 --> 00:26:29,039
organizational sites so that day it was

729
00:26:26,580 --> 00:26:30,900
a Monday I get into work and as always

730
00:26:29,039 --> 00:26:33,059
on a Monday there's a bunch of stuff to

731
00:26:30,900 --> 00:26:34,679
catch up on from the weekend and plus I

732
00:26:33,059 --> 00:26:36,900
was involved with two incidents last

733
00:26:34,679 --> 00:26:38,700
week that I need to follow up so I turn

734
00:26:36,900 --> 00:26:40,320
up and my goal is to tick off as much as

735
00:26:38,700 --> 00:26:41,340
possible so I can get back to the real

736
00:26:40,320 --> 00:26:43,620
work

737
00:26:41,340 --> 00:26:45,179
and one of my tasks was to check on a

738
00:26:43,620 --> 00:26:47,340
backfill that I'd been running that was

739
00:26:45,179 --> 00:26:49,260
tracking its progress in redis for those

740
00:26:47,340 --> 00:26:51,779
that don't know redis is a in-memory key

741
00:26:49,260 --> 00:26:53,400
value data store and the backfill it

742
00:26:51,779 --> 00:26:54,900
finished but it had some weird errors

743
00:26:53,400 --> 00:26:57,059
right toward the end complaining that

744
00:26:54,900 --> 00:26:58,559
redis had run out of memory but those

745
00:26:57,059 --> 00:27:01,200
errors had gone away and it had got

746
00:26:58,559 --> 00:27:03,240
better and so it had given the backfill

747
00:27:01,200 --> 00:27:05,760
it finished I figured the obvious thing

748
00:27:03,240 --> 00:27:07,860
was to just remove that tracking data

749
00:27:05,760 --> 00:27:09,360
because we didn't need it anymore and I

750
00:27:07,860 --> 00:27:10,500
kind of need to do this before the end

751
00:27:09,360 --> 00:27:12,720
of the day because a lot of our

752
00:27:10,500 --> 00:27:13,919
customers are us-based and so this is

753
00:27:12,720 --> 00:27:15,720
Monday and they're all going to come

754
00:27:13,919 --> 00:27:17,400
online overnight and I need to fix this

755
00:27:15,720 --> 00:27:19,559
before the load comes

756
00:27:17,400 --> 00:27:21,480
so how do I remove this temporary data

757
00:27:19,559 --> 00:27:24,120
well there's a command from redis it's

758
00:27:21,480 --> 00:27:26,520
called Dell delete stuff so I guess I'll

759
00:27:24,120 --> 00:27:28,260
log into prod and I'll just run it and

760
00:27:26,520 --> 00:27:30,480
like if it doesn't work or it takes too

761
00:27:28,260 --> 00:27:32,220
long it's going to time out

762
00:27:30,480 --> 00:27:33,240
um or I can just stop it and it'll be

763
00:27:32,220 --> 00:27:36,179
fine

764
00:27:33,240 --> 00:27:38,039
so I run my command and it times out but

765
00:27:36,179 --> 00:27:40,020
it's the client timing out waiting for

766
00:27:38,039 --> 00:27:43,200
the server to finish the recommend not

767
00:27:40,020 --> 00:27:45,960
actually the command timing out and this

768
00:27:43,200 --> 00:27:48,600
is when the Panic sets in you know that

769
00:27:45,960 --> 00:27:50,880
fight flight freeze thing well I was in

770
00:27:48,600 --> 00:27:54,240
full freeze mode my brain literally

771
00:27:50,880 --> 00:27:55,860
stopped working but I managed to open a

772
00:27:54,240 --> 00:27:58,080
dashboard to see if there were errors

773
00:27:55,860 --> 00:28:00,240
and everything was red

774
00:27:58,080 --> 00:28:03,059
um after the incident I learned that

775
00:28:00,240 --> 00:28:05,100
that Dell command if you run it on a

776
00:28:03,059 --> 00:28:07,799
data structure deletes each element one

777
00:28:05,100 --> 00:28:10,380
at a time and blocks all other requests

778
00:28:07,799 --> 00:28:12,360
from running while it's doing so I was

779
00:28:10,380 --> 00:28:15,299
deleting a set with approximately 200

780
00:28:12,360 --> 00:28:18,299
million entries in it

781
00:28:15,299 --> 00:28:20,880
um so but so what did I do with this

782
00:28:18,299 --> 00:28:22,740
back at a previous employer I did some

783
00:28:20,880 --> 00:28:25,200
SRE training and they told a story

784
00:28:22,740 --> 00:28:27,539
someone had made some changes and broke

785
00:28:25,200 --> 00:28:28,980
something very important and while the

786
00:28:27,539 --> 00:28:31,380
incident responders were digging around

787
00:28:28,980 --> 00:28:33,480
the person that made the change stuck

788
00:28:31,380 --> 00:28:35,340
his head around the corner and said I

789
00:28:33,480 --> 00:28:37,080
don't know if it's relevant but I did

790
00:28:35,340 --> 00:28:39,299
this thing

791
00:28:37,080 --> 00:28:41,520
um and the way they told the story that

792
00:28:39,299 --> 00:28:43,140
guy was the hero if he hadn't mentioned

793
00:28:41,520 --> 00:28:45,240
what had changed it would have taken

794
00:28:43,140 --> 00:28:47,400
much longer to bring everything back and

795
00:28:45,240 --> 00:28:49,200
this is why we don't cast blame when

796
00:28:47,400 --> 00:28:51,120
people feel safe they mention things

797
00:28:49,200 --> 00:28:52,799
they would never say otherwise and that

798
00:28:51,120 --> 00:28:54,900
knowledge is a superpower

799
00:28:52,799 --> 00:28:57,000
so I held on to that story with both

800
00:28:54,900 --> 00:28:59,220
hands and I managed to post something in

801
00:28:57,000 --> 00:29:01,679
the incident response Channel it was not

802
00:28:59,220 --> 00:29:03,600
good but I couldn't wait for the Panic

803
00:29:01,679 --> 00:29:05,700
to chill enough to come up with the

804
00:29:03,600 --> 00:29:07,140
right words I just had to post something

805
00:29:05,700 --> 00:29:10,140
and Hope

806
00:29:07,140 --> 00:29:12,360
and that moment is the moment I'm most

807
00:29:10,140 --> 00:29:13,500
proud of out of any incident I've been

808
00:29:12,360 --> 00:29:16,320
involved with

809
00:29:13,500 --> 00:29:18,360
in retrospect it wasn't a risk but in

810
00:29:16,320 --> 00:29:19,980
that moment when my brain was melting

811
00:29:18,360 --> 00:29:22,200
and I couldn't think through the

812
00:29:19,980 --> 00:29:24,320
consequences I still chose to call out

813
00:29:22,200 --> 00:29:27,000
my mistake so we could fix things faster

814
00:29:24,320 --> 00:29:29,279
and also I'm very proud of the company

815
00:29:27,000 --> 00:29:31,020
that I work for where that was the right

816
00:29:29,279 --> 00:29:32,760
call to make which is not true

817
00:29:31,020 --> 00:29:35,520
everywhere

818
00:29:32,760 --> 00:29:37,320
so what do I do with this story well I

819
00:29:35,520 --> 00:29:39,960
could have just kept it to myself but

820
00:29:37,320 --> 00:29:41,580
it's important I held on to a story in

821
00:29:39,960 --> 00:29:43,860
that moment and I wanted to be able to

822
00:29:41,580 --> 00:29:46,020
give that to other people so I wrote

823
00:29:43,860 --> 00:29:48,240
this out and this is basically a recap

824
00:29:46,020 --> 00:29:50,880
of a internal blog post that I made at

825
00:29:48,240 --> 00:29:52,380
the time and I made it and everyone was

826
00:29:50,880 --> 00:29:53,520
nice and said lovely things and it was

827
00:29:52,380 --> 00:29:55,260
great

828
00:29:53,520 --> 00:29:57,659
um but the real punchline of the story

829
00:29:55,260 --> 00:29:59,520
was months later it was someone else's

830
00:29:57,659 --> 00:30:01,980
turn to click the button and break

831
00:29:59,520 --> 00:30:03,899
everything and they also chose to post

832
00:30:01,980 --> 00:30:05,520
in the incident response Channel and

833
00:30:03,899 --> 00:30:08,700
their referenced my post when they did

834
00:30:05,520 --> 00:30:11,159
it and that is the power of our stories

835
00:30:08,700 --> 00:30:14,640
stories shape the culture of who we are

836
00:30:11,159 --> 00:30:17,039
together and who we choose to be and we

837
00:30:14,640 --> 00:30:18,720
all have the power of our stories and we

838
00:30:17,039 --> 00:30:21,179
all have the choice of when to speak

839
00:30:18,720 --> 00:30:23,220
those stories and when to listen and so

840
00:30:21,179 --> 00:30:25,559
I invite you to be bold and to be

841
00:30:23,220 --> 00:30:28,640
vulnerable and to build a culture of who

842
00:30:25,559 --> 00:30:28,640
we want to be together

843
00:30:30,179 --> 00:30:35,220
thanks Tessa

844
00:30:32,460 --> 00:30:36,899
had a uh a privileged position in the in

845
00:30:35,220 --> 00:30:40,380
the front of the room watching this wave

846
00:30:36,899 --> 00:30:42,720
of wints just uh wash over the entirety

847
00:30:40,380 --> 00:30:45,960
of the audience here hi Andrew

848
00:30:42,720 --> 00:30:48,240
uh up on Deck we have achinta but first

849
00:30:45,960 --> 00:30:50,340
uh Lachlan Jacob who's going to talk

850
00:30:48,240 --> 00:30:52,880
about something that may not have aged

851
00:30:50,340 --> 00:30:52,880
all that well

852
00:30:53,700 --> 00:30:58,140
I'm LG um and I'm today talking about

853
00:30:56,159 --> 00:31:00,120
the bless you Channel

854
00:30:58,140 --> 00:31:01,980
so what is the bless you Channel

855
00:31:00,120 --> 00:31:03,720
um we have a dedicated slack Channel at

856
00:31:01,980 --> 00:31:05,279
work for blessing colleagues when they

857
00:31:03,720 --> 00:31:08,460
sneeze

858
00:31:05,279 --> 00:31:10,020
um it started in 2017. um it was a bit

859
00:31:08,460 --> 00:31:11,880
of a joke you know it's funnier to type

860
00:31:10,020 --> 00:31:14,039
someone's name into slack than to

861
00:31:11,880 --> 00:31:15,120
actually use words to tell them bless

862
00:31:14,039 --> 00:31:18,480
you

863
00:31:15,120 --> 00:31:20,220
um but you know due to things in 2020 it

864
00:31:18,480 --> 00:31:23,480
died off a little bit despite probably

865
00:31:20,220 --> 00:31:23,480
more sneezes happening

866
00:31:23,520 --> 00:31:27,299
um coming out of 2020 there and as we

867
00:31:25,380 --> 00:31:29,100
started returning to the office

868
00:31:27,299 --> 00:31:32,580
um reports of bless you's death were

869
00:31:29,100 --> 00:31:34,380
greatly exaggerated and in 2022 of April

870
00:31:32,580 --> 00:31:35,760
it received an uptick in usage as people

871
00:31:34,380 --> 00:31:37,919
came back

872
00:31:35,760 --> 00:31:41,659
this uptick led to ultimately the

873
00:31:37,919 --> 00:31:41,659
creation of sneezebot

874
00:31:41,880 --> 00:31:47,940
sneezebot since it's been added to us

875
00:31:44,460 --> 00:31:49,919
slack it's um invoked a lot of fierce

876
00:31:47,940 --> 00:31:51,299
competition as people race to be the

877
00:31:49,919 --> 00:31:52,980
first to bless someone

878
00:31:51,299 --> 00:31:55,320
and it's made coming back to the office

879
00:31:52,980 --> 00:31:57,539
a little bit more fun

880
00:31:55,320 --> 00:31:59,760
um the way it works is pretty simple

881
00:31:57,539 --> 00:32:01,799
um firstly a sneeze occurs

882
00:31:59,760 --> 00:32:04,679
secondly you tag the user in the channel

883
00:32:01,799 --> 00:32:06,179
as fast as you can and third to sort of

884
00:32:04,679 --> 00:32:08,399
verify that that sneeze did in fact

885
00:32:06,179 --> 00:32:10,679
occur the user who was tagged has to

886
00:32:08,399 --> 00:32:12,779
react to the message

887
00:32:10,679 --> 00:32:14,820
and importantly only the first message

888
00:32:12,779 --> 00:32:18,419
counts

889
00:32:14,820 --> 00:32:21,240
so we also added a command to slack that

890
00:32:18,419 --> 00:32:22,380
would post a sneeze aboard of um you

891
00:32:21,240 --> 00:32:24,960
know how people are doing with their

892
00:32:22,380 --> 00:32:26,820
sneezing and blessing which was pretty

893
00:32:24,960 --> 00:32:29,279
exciting

894
00:32:26,820 --> 00:32:30,720
um and yeah taking a little bit of look

895
00:32:29,279 --> 00:32:33,179
under the hood the way the slackbot

896
00:32:30,720 --> 00:32:35,220
works it listens to messages it gets the

897
00:32:33,179 --> 00:32:38,580
text finds the tag in this case John

898
00:32:35,220 --> 00:32:41,100
shout out to him and adds the message to

899
00:32:38,580 --> 00:32:42,720
an unverified blesses map

900
00:32:41,100 --> 00:32:45,059
then on the flip side of that when he

901
00:32:42,720 --> 00:32:47,520
reacts we look at the we listen for

902
00:32:45,059 --> 00:32:49,799
reactions and when they occur if we find

903
00:32:47,520 --> 00:32:51,720
the message we add points to the

904
00:32:49,799 --> 00:32:53,640
scoreboards and remove the message so no

905
00:32:51,720 --> 00:32:58,140
one else can get points

906
00:32:53,640 --> 00:33:01,020
this code on to such an extent

907
00:32:58,140 --> 00:33:03,779
that trophies were created and some

908
00:33:01,020 --> 00:33:07,020
special emojis for people who did well

909
00:33:03,779 --> 00:33:08,640
in the various sneezins the sneezing's

910
00:33:07,020 --> 00:33:11,220
mapped Real World seasons so we have a

911
00:33:08,640 --> 00:33:12,470
summer sneeze and Autumn season and so

912
00:33:11,220 --> 00:33:14,760
on

913
00:33:12,470 --> 00:33:16,320
[Applause]

914
00:33:14,760 --> 00:33:18,419
um one of the things we noticed once

915
00:33:16,320 --> 00:33:19,980
people started using the tool was that

916
00:33:18,419 --> 00:33:22,320
one of the most exciting things that can

917
00:33:19,980 --> 00:33:24,240
happen is if there's a close call two

918
00:33:22,320 --> 00:33:26,940
people race to bless and someone just

919
00:33:24,240 --> 00:33:29,279
misses out and people tended to pile on

920
00:33:26,940 --> 00:33:32,419
a little bit and give them a letter L or

921
00:33:29,279 --> 00:33:32,419
indicating they lost

922
00:33:32,640 --> 00:33:35,760
um we wanted to encourage this sort of

923
00:33:34,380 --> 00:33:37,620
shaming so

924
00:33:35,760 --> 00:33:40,019
um

925
00:33:37,620 --> 00:33:43,380
one of the enhancements was to add an

926
00:33:40,019 --> 00:33:45,000
elder board for if you lose

927
00:33:43,380 --> 00:33:47,159
it was even the case that people would

928
00:33:45,000 --> 00:33:48,840
delete messages to delete messages to

929
00:33:47,159 --> 00:33:52,760
avoid the embarrassment

930
00:33:48,840 --> 00:33:52,760
and this is the elder board now

931
00:33:53,880 --> 00:33:59,820
um there's lots of potential future

932
00:33:56,580 --> 00:34:01,980
ideas for what we can do with this we're

933
00:33:59,820 --> 00:34:05,340
really only just getting started but

934
00:34:01,980 --> 00:34:07,320
um there's uh ideas of analytics API to

935
00:34:05,340 --> 00:34:08,820
get some information a sneeze graph so

936
00:34:07,320 --> 00:34:10,500
you can see who blesses who and maybe

937
00:34:08,820 --> 00:34:11,359
match that up with where people sit in

938
00:34:10,500 --> 00:34:14,040
the office

939
00:34:11,359 --> 00:34:15,419
stats commands achievements and many

940
00:34:14,040 --> 00:34:17,159
more things so there's plenty of

941
00:34:15,419 --> 00:34:19,040
exciting development still to come with

942
00:34:17,159 --> 00:34:22,200
snesbot

943
00:34:19,040 --> 00:34:25,020
and some quick stats um we've so far

944
00:34:22,200 --> 00:34:28,159
detected 809 sneezes in the office and

945
00:34:25,020 --> 00:34:31,560
145 of those belong to one person so

946
00:34:28,159 --> 00:34:33,599
good job to them we won't name them and

947
00:34:31,560 --> 00:34:35,040
all in all a great deal of wasted office

948
00:34:33,599 --> 00:34:37,820
productivity

949
00:34:35,040 --> 00:34:37,820
thank you

950
00:34:39,720 --> 00:34:43,339
uh because will that be available

951
00:34:43,460 --> 00:34:48,720
uh another first time speaker indeed uh

952
00:34:46,740 --> 00:34:51,720
I'm really looking forward to seeing you

953
00:34:48,720 --> 00:34:53,820
stop I'm looking forward to uh to seeing

954
00:34:51,720 --> 00:34:55,679
a proposal on that uh in next year's uh

955
00:34:53,820 --> 00:34:57,359
in next year's call for proposals will

956
00:34:55,679 --> 00:34:59,160
that project be available during the

957
00:34:57,359 --> 00:35:00,119
Sprints asking for a friend that friend

958
00:34:59,160 --> 00:35:02,180
is me

959
00:35:00,119 --> 00:35:05,160
okay

960
00:35:02,180 --> 00:35:09,180
uh right so up on Deck we have uh Jack

961
00:35:05,160 --> 00:35:11,700
reichelt but first achinta

962
00:35:09,180 --> 00:35:12,780
uh hello my name is achinta I'm an

963
00:35:11,700 --> 00:35:15,060
undergraduate student from the

964
00:35:12,780 --> 00:35:16,440
Australian National University and I

965
00:35:15,060 --> 00:35:20,520
haven't done a speech since year 12

966
00:35:16,440 --> 00:35:23,040
English so if this sucks I am so sorry

967
00:35:20,520 --> 00:35:24,900
uh okay machine learning everybody loves

968
00:35:23,040 --> 00:35:26,280
it it helps us derive meaning out of

969
00:35:24,900 --> 00:35:28,560
massive amounts of data and make

970
00:35:26,280 --> 00:35:30,720
predictions based on patterns most

971
00:35:28,560 --> 00:35:32,520
people have some interaction with it on

972
00:35:30,720 --> 00:35:34,680
a daily basis without even realizing it

973
00:35:32,520 --> 00:35:36,660
and machine learning now is progressing

974
00:35:34,680 --> 00:35:38,640
at an incredible rate with increasingly

975
00:35:36,660 --> 00:35:40,079
complex models all the way from your

976
00:35:38,640 --> 00:35:43,200
regression models and decision tree

977
00:35:40,079 --> 00:35:45,660
models uh to think deep neural networks

978
00:35:43,200 --> 00:35:48,540
cnns and the list goes on

979
00:35:45,660 --> 00:35:50,520
uh despite the ever ramping accuracy of

980
00:35:48,540 --> 00:35:52,260
these complex models and techniques one

981
00:35:50,520 --> 00:35:53,579
of the most significant drawbacks is the

982
00:35:52,260 --> 00:35:55,619
lack of transparency behind Model

983
00:35:53,579 --> 00:35:57,660
Behavior as these models get

984
00:35:55,619 --> 00:35:59,339
increasingly complex the nature of the

985
00:35:57,660 --> 00:36:01,380
model turns into somewhat of a black box

986
00:35:59,339 --> 00:36:03,420
this is actually quite detrimental

987
00:36:01,380 --> 00:36:05,339
because what it does is it starts to

988
00:36:03,420 --> 00:36:06,780
severely encumber their users in

989
00:36:05,339 --> 00:36:09,599
critical critical decision-making

990
00:36:06,780 --> 00:36:11,400
applications regardless of the accuracy

991
00:36:09,599 --> 00:36:13,740
of a model if you can't explain why it

992
00:36:11,400 --> 00:36:15,780
came to a decision that it did that is

993
00:36:13,740 --> 00:36:18,119
not good even conceptually you're

994
00:36:15,780 --> 00:36:19,980
putting your faith sometimes blindly in

995
00:36:18,119 --> 00:36:22,560
something you don't understand and and

996
00:36:19,980 --> 00:36:24,780
that is problematic uh you might be

997
00:36:22,560 --> 00:36:27,000
thinking if the model is accurate and is

998
00:36:24,780 --> 00:36:29,640
doing what I ask it to why should I care

999
00:36:27,000 --> 00:36:31,619
well as many of you would know our

1000
00:36:29,640 --> 00:36:33,900
machine learning models are really are

1001
00:36:31,619 --> 00:36:35,579
only as good as our data and making sure

1002
00:36:33,900 --> 00:36:37,740
you have good data is is bloody

1003
00:36:35,579 --> 00:36:39,540
difficult and because depend because

1004
00:36:37,740 --> 00:36:40,740
depending on what you're looking at data

1005
00:36:39,540 --> 00:36:43,020
can be filled with biases and

1006
00:36:40,740 --> 00:36:45,900
inaccuracies so making sure you have

1007
00:36:43,020 --> 00:36:47,280
good data is not enough uh you can

1008
00:36:45,900 --> 00:36:48,660
imagine that when you take machine

1009
00:36:47,280 --> 00:36:50,700
learning in the context of something

1010
00:36:48,660 --> 00:36:52,980
like healthcare it can be very difficult

1011
00:36:50,700 --> 00:36:55,859
to ensure our models uh are representing

1012
00:36:52,980 --> 00:36:57,480
people of all backgrounds equally when

1013
00:36:55,859 --> 00:36:59,579
we don't understand why it is making the

1014
00:36:57,480 --> 00:37:03,180
decision it did so what you end up with

1015
00:36:59,579 --> 00:37:04,980
are headlines like these and even more

1016
00:37:03,180 --> 00:37:07,200
marginalization of certain groups of

1017
00:37:04,980 --> 00:37:09,420
people and yes these articles were not

1018
00:37:07,200 --> 00:37:12,240
hard to find I'm sure you guys have come

1019
00:37:09,420 --> 00:37:15,240
across topics like these two and know it

1020
00:37:12,240 --> 00:37:17,460
is not just a data issue as Doshi Velez

1021
00:37:15,240 --> 00:37:19,079
and Kim have put it the problem is that

1022
00:37:17,460 --> 00:37:21,180
a single metric such as classification

1023
00:37:19,079 --> 00:37:23,520
accuracy or even a collection of metrics

1024
00:37:21,180 --> 00:37:25,200
is in incomplete description of most

1025
00:37:23,520 --> 00:37:26,700
real world tasks

1026
00:37:25,200 --> 00:37:28,700
this is where the concept of

1027
00:37:26,700 --> 00:37:31,260
interpretability comes in

1028
00:37:28,700 --> 00:37:32,520
as Christoph molnap would say it's hard

1029
00:37:31,260 --> 00:37:34,920
to mathematically Define what

1030
00:37:32,520 --> 00:37:37,200
interpretability is but intuitively it

1031
00:37:34,920 --> 00:37:38,820
is the same as explainability and we can

1032
00:37:37,200 --> 00:37:41,760
kind of think of it as equipping these

1033
00:37:38,820 --> 00:37:43,920
ml models with the tools display their

1034
00:37:41,760 --> 00:37:46,440
machinations to us in a manner that our

1035
00:37:43,920 --> 00:37:47,880
monkey human brains can understand so it

1036
00:37:46,440 --> 00:37:51,960
really is just placing a bit more

1037
00:37:47,880 --> 00:37:53,579
emphasis on why as opposed to the how

1038
00:37:51,960 --> 00:37:55,560
um yeah so as outlined by doing

1039
00:37:53,579 --> 00:37:57,060
colleagues in this great article there's

1040
00:37:55,560 --> 00:37:59,280
two classifications for the techniques

1041
00:37:57,060 --> 00:38:00,839
that achieve interpretability and those

1042
00:37:59,280 --> 00:38:03,180
are intrinsic intermability and

1043
00:38:00,839 --> 00:38:04,980
post-talk interpretability as the name

1044
00:38:03,180 --> 00:38:06,599
suggests intrinsic interpretability

1045
00:38:04,980 --> 00:38:08,460
involves having interpretability

1046
00:38:06,599 --> 00:38:10,859
constraints constructed into the model

1047
00:38:08,460 --> 00:38:13,380
structure itself uh on the other hand

1048
00:38:10,859 --> 00:38:15,599
post-talk interpretability techniques

1049
00:38:13,380 --> 00:38:18,180
involve examining the model after the

1050
00:38:15,599 --> 00:38:19,920
fact typically through creating a second

1051
00:38:18,180 --> 00:38:22,440
model in parallel to explain the

1052
00:38:19,920 --> 00:38:24,359
behaviors of the initial model so both

1053
00:38:22,440 --> 00:38:25,800
come with their own trade-offs and it's

1054
00:38:24,359 --> 00:38:29,099
typically between the performance of

1055
00:38:25,800 --> 00:38:32,040
your model and explanation Fidelity so

1056
00:38:29,099 --> 00:38:33,900
and intrinsically an interpretable model

1057
00:38:32,040 --> 00:38:36,000
may become more interpretable as you

1058
00:38:33,900 --> 00:38:38,220
stack on these constraints but then you

1059
00:38:36,000 --> 00:38:40,380
tend to uh

1060
00:38:38,220 --> 00:38:41,820
get a decrease in model performance as

1061
00:38:40,380 --> 00:38:44,339
you bottleneck the features used in

1062
00:38:41,820 --> 00:38:46,560
prediction in the post talk case it's

1063
00:38:44,339 --> 00:38:50,099
really just an approximation of the

1064
00:38:46,560 --> 00:38:52,619
Model Behavior and so the Fidelity

1065
00:38:50,099 --> 00:38:55,320
Fidelity of the explanation we receive

1066
00:38:52,619 --> 00:38:58,260
may not be as clear but you've retained

1067
00:38:55,320 --> 00:39:01,500
the underlying model performance so what

1068
00:38:58,260 --> 00:39:03,140
other end results of this a massive net

1069
00:39:01,500 --> 00:39:05,760
positive

1070
00:39:03,140 --> 00:39:07,619
uh sorry

1071
00:39:05,760 --> 00:39:09,660
uh yeah for the user they can have

1072
00:39:07,619 --> 00:39:11,220
confidence and trust in these models and

1073
00:39:09,660 --> 00:39:13,380
this is in turn encourages further

1074
00:39:11,220 --> 00:39:15,119
uptake for developers we get a better

1075
00:39:13,380 --> 00:39:17,700
understanding of why things go wrong

1076
00:39:15,119 --> 00:39:19,140
when they inevitably do and that makes

1077
00:39:17,700 --> 00:39:21,480
sure that there's less harm being put

1078
00:39:19,140 --> 00:39:22,560
out by this thing we create so I've just

1079
00:39:21,480 --> 00:39:24,180
scratched the surface of these

1080
00:39:22,560 --> 00:39:26,940
techniques and these categories are

1081
00:39:24,180 --> 00:39:28,200
actually broken down even further but I

1082
00:39:26,940 --> 00:39:29,640
literally just finished writing this

1083
00:39:28,200 --> 00:39:33,000
script and I have no idea how long I've

1084
00:39:29,640 --> 00:39:35,700
got left so I want to reiterate that we

1085
00:39:33,000 --> 00:39:37,260
absolutely do not hate complexity it is

1086
00:39:35,700 --> 00:39:39,000
amazing but my point is it's kind of

1087
00:39:37,260 --> 00:39:41,099
like hopping in an Uber and your Uber

1088
00:39:39,000 --> 00:39:43,260
driver goes yeah I'll take you home but

1089
00:39:41,099 --> 00:39:44,820
I want you to cover your eyes and ears

1090
00:39:43,260 --> 00:39:47,099
the whole way and you'll probably end up

1091
00:39:44,820 --> 00:39:49,140
on your front porch this goes beyond

1092
00:39:47,099 --> 00:39:51,000
just machine learning side of things and

1093
00:39:49,140 --> 00:39:53,579
it really relies on the expertise of

1094
00:39:51,000 --> 00:39:55,320
many different sectors from data

1095
00:39:53,579 --> 00:39:58,460
analytics to social science to ethics

1096
00:39:55,320 --> 00:39:58,460
and so on thank you

1097
00:40:01,200 --> 00:40:06,420
yet another first time presenter we uh

1098
00:40:03,780 --> 00:40:08,880
we had three times as many proposals as

1099
00:40:06,420 --> 00:40:10,859
we had space for today including more

1100
00:40:08,880 --> 00:40:13,020
first-time presenters who than people

1101
00:40:10,859 --> 00:40:15,720
who propose anything at all yesterday so

1102
00:40:13,020 --> 00:40:17,220
uh uh yeah thank you all for submitting

1103
00:40:15,720 --> 00:40:19,140
lightning talks especially all of our

1104
00:40:17,220 --> 00:40:20,280
first time proposers hopefully we'll see

1105
00:40:19,140 --> 00:40:24,240
some of the ones that we couldn't take

1106
00:40:20,280 --> 00:40:27,180
in this year uh at next year's uh tangdo

1107
00:40:24,240 --> 00:40:29,700
is up on this side but first Jack

1108
00:40:27,180 --> 00:40:31,200
reichelt who is

1109
00:40:29,700 --> 00:40:33,660
going to do a lot of things with the

1110
00:40:31,200 --> 00:40:35,280
next five minutes yeah so we're going to

1111
00:40:33,660 --> 00:40:36,960
form a techian we're going for an 80

1112
00:40:35,280 --> 00:40:39,060
Speed Run world record attempt

1113
00:40:36,960 --> 00:40:40,560
glitchless as in uh I don't want to

1114
00:40:39,060 --> 00:40:42,480
break the law I don't really know if I'm

1115
00:40:40,560 --> 00:40:44,040
going to and yes this is actually a real

1116
00:40:42,480 --> 00:40:46,200
attempt yes I'm actually studying Union

1117
00:40:44,040 --> 00:40:47,280
yes you can actually sign up so the

1118
00:40:46,200 --> 00:40:48,480
first thing that we need to do is we

1119
00:40:47,280 --> 00:40:50,579
need to have our form where people can

1120
00:40:48,480 --> 00:40:52,500
register input and so that's really good

1121
00:40:50,579 --> 00:40:56,040
here but this link is really put on your

1122
00:40:52,500 --> 00:40:59,460
screen is it not only it should be

1123
00:40:56,040 --> 00:41:01,520
um how do I mirror real quick oh Jesus

1124
00:40:59,460 --> 00:41:01,520
um

1125
00:41:02,579 --> 00:41:05,579
well

1126
00:41:07,800 --> 00:41:09,859
um

1127
00:41:16,520 --> 00:41:22,980
world record speed run attempt make a

1128
00:41:19,260 --> 00:41:23,579
mac oh boy this isn't good Okay cool so

1129
00:41:22,980 --> 00:41:25,800
um

1130
00:41:23,579 --> 00:41:26,880
we need a link uh but that's real long

1131
00:41:25,800 --> 00:41:28,800
that's really bad so I'm going to stick

1132
00:41:26,880 --> 00:41:31,260
this in the URL shorter so that you can

1133
00:41:28,800 --> 00:41:33,240
all get that real nice and quick

1134
00:41:31,260 --> 00:41:34,800
um and we're going to go over here and

1135
00:41:33,240 --> 00:41:36,480
I'm going to skip that because I lost it

1136
00:41:34,800 --> 00:41:38,280
this will go on the Discord later and so

1137
00:41:36,480 --> 00:41:39,900
you can also see that so here we have to

1138
00:41:38,280 --> 00:41:42,119
write a whole bunch of rules um these

1139
00:41:39,900 --> 00:41:46,020
are the rules that a union needs and so

1140
00:41:42,119 --> 00:41:49,200
uh I've written a bunch of them that

1141
00:41:46,020 --> 00:41:52,320
should autofill and aren't oh it's all

1142
00:41:49,200 --> 00:41:53,700
gone terribly wrong okay so in that case

1143
00:41:52,320 --> 00:41:55,680
I'll slow down and I'll talk about other

1144
00:41:53,700 --> 00:41:57,480
things and we're not going to do the

1145
00:41:55,680 --> 00:41:58,740
speed Runway record but

1146
00:41:57,480 --> 00:42:01,380
um I am actually going to start a union

1147
00:41:58,740 --> 00:42:04,200
I don't really know how I have a friend

1148
00:42:01,380 --> 00:42:06,000
of mine who has started a union for PhD

1149
00:42:04,200 --> 00:42:08,520
students or something similar I don't

1150
00:42:06,000 --> 00:42:12,180
know exactly sorry Jess

1151
00:42:08,520 --> 00:42:13,380
um but uh as I I've been saying we

1152
00:42:12,180 --> 00:42:14,280
should have eaten for years and then I

1153
00:42:13,380 --> 00:42:15,900
was kind of inspired by the talk

1154
00:42:14,280 --> 00:42:17,099
yesterday and I thought well if I've

1155
00:42:15,900 --> 00:42:18,900
been saying it for years and nobody's

1156
00:42:17,099 --> 00:42:21,839
done it maybe that just means that I

1157
00:42:18,900 --> 00:42:23,280
should do it and so I'm gonna do it and

1158
00:42:21,839 --> 00:42:25,980
um

1159
00:42:23,280 --> 00:42:27,060
stop and so a couple of things with that

1160
00:42:25,980 --> 00:42:27,720
is

1161
00:42:27,060 --> 00:42:30,119
um

1162
00:42:27,720 --> 00:42:33,180
I don't know if Australians of our Union

1163
00:42:30,119 --> 00:42:34,859
is the best name for it I just thought a

1164
00:42:33,180 --> 00:42:37,380
sue sounded fun

1165
00:42:34,859 --> 00:42:39,540
um so in this form here uh there's a

1166
00:42:37,380 --> 00:42:40,980
thing of uh what do you think the name

1167
00:42:39,540 --> 00:42:42,599
of the Union should be

1168
00:42:40,980 --> 00:42:43,500
um so fill that out and we'll work that

1169
00:42:42,599 --> 00:42:45,000
out

1170
00:42:43,500 --> 00:42:47,220
um in these rules there's a whole bunch

1171
00:42:45,000 --> 00:42:48,540
of stuff here including really

1172
00:42:47,220 --> 00:42:50,700
critically

1173
00:42:48,540 --> 00:42:51,839
um was there property and funds we don't

1174
00:42:50,700 --> 00:42:54,599
have any

1175
00:42:51,839 --> 00:42:56,099
um it's it's just now

1176
00:42:54,599 --> 00:42:58,079
um alteration of rules these are

1177
00:42:56,099 --> 00:43:00,119
probably bad rules I'm going to say that

1178
00:42:58,079 --> 00:43:01,859
right now these are definitely bad rules

1179
00:43:00,119 --> 00:43:03,900
because I wrote them starting at about

1180
00:43:01,859 --> 00:43:06,480
lunchtime yesterday

1181
00:43:03,900 --> 00:43:08,280
um and so these will all be posted

1182
00:43:06,480 --> 00:43:09,839
they're good enough to get started

1183
00:43:08,280 --> 00:43:11,280
because they're better than nothing but

1184
00:43:09,839 --> 00:43:13,020
I made sure to make it so that it's not

1185
00:43:11,280 --> 00:43:14,640
too hard to alter the rules

1186
00:43:13,020 --> 00:43:17,220
um the way that I got this list by the

1187
00:43:14,640 --> 00:43:19,260
way is that the government said the

1188
00:43:17,220 --> 00:43:22,200
government says that every the rules for

1189
00:43:19,260 --> 00:43:24,420
a union have to include these things and

1190
00:43:22,200 --> 00:43:25,319
so it includes these things

1191
00:43:24,420 --> 00:43:27,060
um

1192
00:43:25,319 --> 00:43:28,500
there is also a form that needs to be

1193
00:43:27,060 --> 00:43:29,579
filled out I haven't filled that out

1194
00:43:28,500 --> 00:43:31,740
because I didn't want to swipe across to

1195
00:43:29,579 --> 00:43:33,960
this end docs myself

1196
00:43:31,740 --> 00:43:36,300
um if it's in my notes do not dock

1197
00:43:33,960 --> 00:43:37,859
yourself also don't dox other people a

1198
00:43:36,300 --> 00:43:42,300
critical thing

1199
00:43:37,859 --> 00:43:43,740
um we need 50 responses to have the

1200
00:43:42,300 --> 00:43:45,599
union be made

1201
00:43:43,740 --> 00:43:47,940
um you can't submit to be a union unless

1202
00:43:45,599 --> 00:43:49,859
you have 50 members which seems kind of

1203
00:43:47,940 --> 00:43:52,380
um backwards to me how do you get 50

1204
00:43:49,859 --> 00:43:53,940
members if you're not a union anyway

1205
00:43:52,380 --> 00:43:55,500
um so that's why I've just called this

1206
00:43:53,940 --> 00:43:58,200
an expression of Interest form rather

1207
00:43:55,500 --> 00:43:58,859
than a sign up to the union form

1208
00:43:58,200 --> 00:44:01,619
um

1209
00:43:58,859 --> 00:44:03,480
oh my other big point I don't think that

1210
00:44:01,619 --> 00:44:04,859
software is actually the best title for

1211
00:44:03,480 --> 00:44:07,380
this because

1212
00:44:04,859 --> 00:44:08,640
um that feels like it kind of excludes a

1213
00:44:07,380 --> 00:44:09,720
lot of people and I didn't actually want

1214
00:44:08,640 --> 00:44:12,240
that

1215
00:44:09,720 --> 00:44:13,920
um in the purpose and the who it's for

1216
00:44:12,240 --> 00:44:17,040
and such the eligibility for membership

1217
00:44:13,920 --> 00:44:20,040
it basically I left it super open-ended

1218
00:44:17,040 --> 00:44:22,800
and it's uh do you now have you ever or

1219
00:44:20,040 --> 00:44:25,920
do you intend to work as like a software

1220
00:44:22,800 --> 00:44:28,319
developer or ux designer or a data

1221
00:44:25,920 --> 00:44:29,640
analyst or kind of anything in Tech at

1222
00:44:28,319 --> 00:44:31,500
all

1223
00:44:29,640 --> 00:44:33,599
um or are you a student of that I guess

1224
00:44:31,500 --> 00:44:36,660
that's in you intend to

1225
00:44:33,599 --> 00:44:38,400
um because I think that they're all very

1226
00:44:36,660 --> 00:44:40,800
related fields

1227
00:44:38,400 --> 00:44:43,500
um one thing I wanted to mention was the

1228
00:44:40,800 --> 00:44:45,060
professionals Australia group

1229
00:44:43,500 --> 00:44:47,339
um I think that's really good and they

1230
00:44:45,060 --> 00:44:50,400
they're very large there's uh and and

1231
00:44:47,339 --> 00:44:52,020
many of you may be signed up to them

1232
00:44:50,400 --> 00:44:53,819
um I think that's good but I also think

1233
00:44:52,020 --> 00:44:56,760
that Tech is a kind of a different field

1234
00:44:53,819 --> 00:44:58,800
in how we interact with it how he how it

1235
00:44:56,760 --> 00:45:01,440
evolves very quickly and how Paul

1236
00:44:58,800 --> 00:45:03,000
policies need to be made about that you

1237
00:45:01,440 --> 00:45:04,560
know things around Architects that

1238
00:45:03,000 --> 00:45:07,560
haven't changed as fast as things around

1239
00:45:04,560 --> 00:45:08,579
technology do so I thought a specific

1240
00:45:07,560 --> 00:45:10,200
Union was good rather than just

1241
00:45:08,579 --> 00:45:12,960
encouraging people to join professionals

1242
00:45:10,200 --> 00:45:15,660
Australia that's why I'm doing this I

1243
00:45:12,960 --> 00:45:17,640
will post this tiny URL I don't normally

1244
00:45:15,660 --> 00:45:18,900
trust those because you never know what

1245
00:45:17,640 --> 00:45:20,819
they're actually going to expand to but

1246
00:45:18,900 --> 00:45:22,020
you also may make it so I'm going to

1247
00:45:20,819 --> 00:45:26,339
post that in the Discord after this

1248
00:45:22,020 --> 00:45:27,359
please sign up and I will I'll keep

1249
00:45:26,339 --> 00:45:28,560
doing this I'll keep in touch with

1250
00:45:27,359 --> 00:45:29,520
everybody who signs up thank you very

1251
00:45:28,560 --> 00:45:32,480
much

1252
00:45:29,520 --> 00:45:32,480
thank you

1253
00:45:33,300 --> 00:45:38,640
I feel like this talk encapsulates all

1254
00:45:36,720 --> 00:45:42,180
of the energy of lightning talks which

1255
00:45:38,640 --> 00:45:44,819
is just severe chaotic good energy

1256
00:45:42,180 --> 00:45:47,099
getting up on stage everything going

1257
00:45:44,819 --> 00:45:49,380
wrong and flying by the city of pants to

1258
00:45:47,099 --> 00:45:52,740
convey a brilliant message at the end so

1259
00:45:49,380 --> 00:45:53,940
thank you also join your union yes

1260
00:45:52,740 --> 00:45:56,460
[Applause]

1261
00:45:53,940 --> 00:45:57,900
hawk Is On Deck and first they talk

1262
00:45:56,460 --> 00:45:59,940
about go

1263
00:45:57,900 --> 00:46:01,680
like g'day P icon

1264
00:45:59,940 --> 00:46:04,680
um I was inspired by yesterday lightning

1265
00:46:01,680 --> 00:46:06,599
round I was so inspired and Ben was talk

1266
00:46:04,680 --> 00:46:08,339
this morning so I decided I'm gonna

1267
00:46:06,599 --> 00:46:10,680
prepare something and get up here today

1268
00:46:08,339 --> 00:46:12,660
do my first lighting talk

1269
00:46:10,680 --> 00:46:16,079
oh thank you thank you

1270
00:46:12,660 --> 00:46:18,839
um so AI became a hot popular Topic in

1271
00:46:16,079 --> 00:46:21,660
recent years and this is my story of my

1272
00:46:18,839 --> 00:46:23,579
relationship with AI I started playing

1273
00:46:21,660 --> 00:46:24,960
go about like two years ago so pretty

1274
00:46:23,579 --> 00:46:27,599
new Noob

1275
00:46:24,960 --> 00:46:30,540
um it has since consumed most if not all

1276
00:46:27,599 --> 00:46:32,339
my free time I buy books I study games I

1277
00:46:30,540 --> 00:46:34,980
get lessons

1278
00:46:32,339 --> 00:46:37,319
all of that to be bang on average the

1279
00:46:34,980 --> 00:46:38,400
pig of the curve as they say it and I'm

1280
00:46:37,319 --> 00:46:40,740
proud of it

1281
00:46:38,400 --> 00:46:44,280
the game itself is phenomenal as

1282
00:46:40,740 --> 00:46:45,839
aesthetically pleasing every move feels

1283
00:46:44,280 --> 00:46:48,540
like a conversation between me and my

1284
00:46:45,839 --> 00:46:51,300
opponent my moves usually say that I

1285
00:46:48,540 --> 00:46:54,660
really want to win please make mistakes

1286
00:46:51,300 --> 00:46:57,180
thank you but it wasn't just the game

1287
00:46:54,660 --> 00:46:59,339
itself that kind of got me hooked

1288
00:46:57,180 --> 00:47:01,560
it was the documentary called Alpha girl

1289
00:46:59,339 --> 00:47:03,680
I'm sure some of you even if you don't

1290
00:47:01,560 --> 00:47:06,980
play the game have heard of it

1291
00:47:03,680 --> 00:47:09,720
developed by the team at Google did mine

1292
00:47:06,980 --> 00:47:12,300
AI opponents in games like this are new

1293
00:47:09,720 --> 00:47:14,700
but a strong opponent like Alpha girl

1294
00:47:12,300 --> 00:47:18,480
was just not a thing back then

1295
00:47:14,700 --> 00:47:19,980
the game girl came from China about 3000

1296
00:47:18,480 --> 00:47:22,380
years ago and it has always been

1297
00:47:19,980 --> 00:47:25,079
regarded as the most challenging

1298
00:47:22,380 --> 00:47:27,000
classical game for AI because of his

1299
00:47:25,079 --> 00:47:30,800
complexity

1300
00:47:27,000 --> 00:47:34,140
goal is profoundly complex there are an

1301
00:47:30,800 --> 00:47:37,920
astonishing 10 to the power of 170

1302
00:47:34,140 --> 00:47:40,980
possible board configurations from the

1303
00:47:37,920 --> 00:47:42,720
get-go this makes the game of girl far

1304
00:47:40,980 --> 00:47:45,000
more complex than the game of chess 10

1305
00:47:42,720 --> 00:47:46,859
to the power of 40.

1306
00:47:45,000 --> 00:47:49,859
I'll spell you the details but long

1307
00:47:46,859 --> 00:47:51,839
story short is used neural networks to

1308
00:47:49,859 --> 00:47:54,480
evaluate and pick the best move and play

1309
00:47:51,839 --> 00:47:57,240
them out and give you the best one on

1310
00:47:54,480 --> 00:47:58,500
the surface this isn't anything new or

1311
00:47:57,240 --> 00:48:03,240
mind-blowing

1312
00:47:58,500 --> 00:48:06,660
but in 2016 this program defeated the

1313
00:48:03,240 --> 00:48:09,619
Roger Federer the samka of the wolf girl

1314
00:48:06,660 --> 00:48:12,599
in a spectacular fashion this event

1315
00:48:09,619 --> 00:48:15,420
shocked the whole community

1316
00:48:12,599 --> 00:48:17,220
many fear that AI is going to come along

1317
00:48:15,420 --> 00:48:19,980
and sort of ruin the game

1318
00:48:17,220 --> 00:48:22,800
however in reality it actually led us

1319
00:48:19,980 --> 00:48:24,060
into the modern era of goal of the AI

1320
00:48:22,800 --> 00:48:26,940
era

1321
00:48:24,060 --> 00:48:29,940
change the game forever

1322
00:48:26,940 --> 00:48:32,520
this is an AI move AI would consider

1323
00:48:29,940 --> 00:48:35,400
this move to be this or that so these

1324
00:48:32,520 --> 00:48:38,099
phrases sort of crop up in professional

1325
00:48:35,400 --> 00:48:40,380
games a lot recently

1326
00:48:38,099 --> 00:48:43,200
there are countless of books written

1327
00:48:40,380 --> 00:48:46,680
about AI play style people started to

1328
00:48:43,200 --> 00:48:49,859
analyze games played by AI to find new

1329
00:48:46,680 --> 00:48:51,780
creative ways to play this 3000 years

1330
00:48:49,859 --> 00:48:54,300
old board game

1331
00:48:51,780 --> 00:48:57,300
as if these AI program was like some

1332
00:48:54,300 --> 00:48:58,680
legendary plays back in the past to be

1333
00:48:57,300 --> 00:49:01,859
admired

1334
00:48:58,680 --> 00:49:05,099
to alphago and other similar AI program

1335
00:49:01,859 --> 00:49:07,980
have now become the most influential

1336
00:49:05,099 --> 00:49:11,220
tool used in go

1337
00:49:07,980 --> 00:49:13,020
it is absolutely fascinating to me that

1338
00:49:11,220 --> 00:49:15,780
these programs learn how to play the

1339
00:49:13,020 --> 00:49:18,420
game by training on the data sets or

1340
00:49:15,780 --> 00:49:20,760
data sets of existing human games then

1341
00:49:18,420 --> 00:49:23,160
it played itself to get stronger

1342
00:49:20,760 --> 00:49:26,700
and then it became so strong that is now

1343
00:49:23,160 --> 00:49:29,220
teaching us how to play our own game

1344
00:49:26,700 --> 00:49:32,099
so we use tools like this one here to

1345
00:49:29,220 --> 00:49:35,940
discover new techno new strategies new

1346
00:49:32,099 --> 00:49:38,819
techniques to learn I use it a lot to

1347
00:49:35,940 --> 00:49:41,520
review my games where I lost

1348
00:49:38,819 --> 00:49:44,040
as a teacher it tells me the reason I

1349
00:49:41,520 --> 00:49:46,440
lost and simply because I suck

1350
00:49:44,040 --> 00:49:48,660
it tells me because it tells me where my

1351
00:49:46,440 --> 00:49:51,359
mistakes are it tells me what the

1352
00:49:48,660 --> 00:49:54,060
alternatives are by playing out X number

1353
00:49:51,359 --> 00:49:55,440
of players which is pretty cool as you

1354
00:49:54,060 --> 00:49:58,079
can see

1355
00:49:55,440 --> 00:50:01,740
does that mean AIS is a threat to All

1356
00:49:58,079 --> 00:50:05,640
Gold professionals or someone who teach

1357
00:50:01,740 --> 00:50:08,099
absolutely not right the Gap in strength

1358
00:50:05,640 --> 00:50:11,640
between AI programs like alphagow and

1359
00:50:08,099 --> 00:50:13,740
the average jokes like me it's like vast

1360
00:50:11,640 --> 00:50:15,720
so I still need someone to understand

1361
00:50:13,740 --> 00:50:18,780
the game better

1362
00:50:15,720 --> 00:50:20,940
to study these AI games extensively and

1363
00:50:18,780 --> 00:50:23,700
pass on the knowledge in their own ways

1364
00:50:20,940 --> 00:50:25,920
it's a tool it's a powerful tools gold

1365
00:50:23,700 --> 00:50:27,960
players have been wielding it expertly

1366
00:50:25,920 --> 00:50:29,940
in the last couple of years to further

1367
00:50:27,960 --> 00:50:31,800
Advance the fill and I would like to

1368
00:50:29,940 --> 00:50:35,280
believe that so have we

1369
00:50:31,800 --> 00:50:36,660
since the tools like chat GPT and all

1370
00:50:35,280 --> 00:50:40,079
that came along

1371
00:50:36,660 --> 00:50:43,940
so Skynet still fictions

1372
00:50:40,079 --> 00:50:43,940
um that's it for me thank you

1373
00:50:45,180 --> 00:50:50,099
thanks tongue

1374
00:50:47,339 --> 00:50:54,420
get another first time speaker

1375
00:50:50,099 --> 00:50:56,760
um fantastic uh jakov is on Deck

1376
00:50:54,420 --> 00:50:59,240
and now here's hockey with a talk about

1377
00:50:56,760 --> 00:50:59,240
go

1378
00:50:59,940 --> 00:51:03,480
hello

1379
00:51:01,020 --> 00:51:06,359
um I I don't have any I didn't have any

1380
00:51:03,480 --> 00:51:08,280
analog fancy slides like Tom or space

1381
00:51:06,359 --> 00:51:10,500
pictures or traffic lights or anything

1382
00:51:08,280 --> 00:51:13,500
cool but but I have a hat

1383
00:51:10,500 --> 00:51:15,480
um so that's how you know it's me

1384
00:51:13,500 --> 00:51:18,680
um you can find me on the internet in in

1385
00:51:15,480 --> 00:51:20,640
places I've got the dowel avatars anyway

1386
00:51:18,680 --> 00:51:24,200
ding dong

1387
00:51:20,640 --> 00:51:24,200
have you heard about Moon Acts

1388
00:51:24,780 --> 00:51:29,160
get on with it you might have but you've

1389
00:51:27,300 --> 00:51:30,119
probably heard of you know go right no

1390
00:51:29,160 --> 00:51:32,160
sorry

1391
00:51:30,119 --> 00:51:34,980
go

1392
00:51:32,160 --> 00:51:36,300
um sorry Elena that's still your joke

1393
00:51:34,980 --> 00:51:39,960
um

1394
00:51:36,300 --> 00:51:42,119
it's been a while since the whole we're

1395
00:51:39,960 --> 00:51:44,940
rewriting all of our Python and go thing

1396
00:51:42,119 --> 00:51:47,520
has been current it was a very uh

1397
00:51:44,940 --> 00:51:49,920
rounded in like 2013 2014 that sort of

1398
00:51:47,520 --> 00:51:51,420
thing it's been a running uh sort of

1399
00:51:49,920 --> 00:51:53,520
theme

1400
00:51:51,420 --> 00:51:55,619
and that's that's been forever so I

1401
00:51:53,520 --> 00:51:57,180
don't I don't expect everyone to keep up

1402
00:51:55,619 --> 00:51:59,940
to date with programming languages they

1403
00:51:57,180 --> 00:52:01,740
might not use so here's some things that

1404
00:51:59,940 --> 00:52:03,359
have changed and go since you probably

1405
00:52:01,740 --> 00:52:06,480
haven't looked

1406
00:52:03,359 --> 00:52:07,920
uh go modules now

1407
00:52:06,480 --> 00:52:09,900
the old way where you had to have

1408
00:52:07,920 --> 00:52:12,359
everything in a certain directory in a

1409
00:52:09,900 --> 00:52:13,740
certain place and you couldn't have

1410
00:52:12,359 --> 00:52:16,859
multiple versions of that in different

1411
00:52:13,740 --> 00:52:21,240
different areas it sucked a lot it was

1412
00:52:16,859 --> 00:52:24,720
very bad but there is new uh the new

1413
00:52:21,240 --> 00:52:28,520
bonus which is using uh go go modules

1414
00:52:24,720 --> 00:52:30,960
it's a lot closer to what you might open

1415
00:52:28,520 --> 00:52:32,700
uh what you might expect when you start

1416
00:52:30,960 --> 00:52:34,680
using a programming language instead of

1417
00:52:32,700 --> 00:52:35,760
trash

1418
00:52:34,680 --> 00:52:38,819
um

1419
00:52:35,760 --> 00:52:40,559
when you have a package such as some

1420
00:52:38,819 --> 00:52:42,599
some code and you've made a package and

1421
00:52:40,559 --> 00:52:45,720
you import something you can then tell

1422
00:52:42,599 --> 00:52:47,099
go hey just to tidy up the go mod and it

1423
00:52:45,720 --> 00:52:49,680
will look for the dependencies and work

1424
00:52:47,099 --> 00:52:52,319
it out and then make this nice file that

1425
00:52:49,680 --> 00:52:55,140
says what Go version it requires what

1426
00:52:52,319 --> 00:52:57,359
the module is uh what it requires

1427
00:52:55,140 --> 00:52:59,940
directly as well as what it requires

1428
00:52:57,359 --> 00:53:02,640
indirectly so it can work out the big

1429
00:52:59,940 --> 00:53:05,099
old tree of dependencies and work out

1430
00:53:02,640 --> 00:53:08,700
exactly what you need and the minimum

1431
00:53:05,099 --> 00:53:11,400
version specification for it

1432
00:53:08,700 --> 00:53:13,440
the other big thing is generics and this

1433
00:53:11,400 --> 00:53:16,680
is something that I have been wanting

1434
00:53:13,440 --> 00:53:18,900
for a very long time and now it's been

1435
00:53:16,680 --> 00:53:20,819
since go 118 which has been around for a

1436
00:53:18,900 --> 00:53:22,260
little a little while and there's been

1437
00:53:20,819 --> 00:53:27,000
some improvements in concurrent versions

1438
00:53:22,260 --> 00:53:29,040
we're up to go 121 and I found that it's

1439
00:53:27,000 --> 00:53:31,619
really helpful in even my own personal

1440
00:53:29,040 --> 00:53:34,079
stuff where I've been writing a thing

1441
00:53:31,619 --> 00:53:35,760
where I've been making a risk five CPU

1442
00:53:34,079 --> 00:53:38,280
interpreter because that's what I do

1443
00:53:35,760 --> 00:53:42,540
with my Christmas holidays

1444
00:53:38,280 --> 00:53:46,140
um so if you have a CPU that you want to

1445
00:53:42,540 --> 00:53:48,720
be 16-bit or 32-bit and but you want to

1446
00:53:46,140 --> 00:53:50,700
work in terms of say your bit width you

1447
00:53:48,720 --> 00:53:56,160
can just make a interface a type it face

1448
00:53:50,700 --> 00:53:59,520
which is in 16 or in 32. the squiggly

1449
00:53:56,160 --> 00:54:02,819
line means that it will accept things

1450
00:53:59,520 --> 00:54:06,960
that have that as the base instead so in

1451
00:54:02,819 --> 00:54:09,599
go for example times are often just an

1452
00:54:06,960 --> 00:54:12,540
INT that means seconds in some sort of

1453
00:54:09,599 --> 00:54:13,859
thing so it sort of subclasses but not

1454
00:54:12,540 --> 00:54:15,420
subclass because it's not classes

1455
00:54:13,859 --> 00:54:17,460
because they're not objects but it's

1456
00:54:15,420 --> 00:54:19,800
anyway

1457
00:54:17,460 --> 00:54:21,180
so when you're actually using this it

1458
00:54:19,800 --> 00:54:23,520
works similar to other programming

1459
00:54:21,180 --> 00:54:26,400
languages so you've got you know you

1460
00:54:23,520 --> 00:54:28,980
define it in terms of this W is my

1461
00:54:26,400 --> 00:54:31,500
register width type sort of thing and I

1462
00:54:28,980 --> 00:54:33,180
can Define things inside my structs that

1463
00:54:31,500 --> 00:54:36,059
are that type and it will map and

1464
00:54:33,180 --> 00:54:39,720
everything works and you just keep using

1465
00:54:36,059 --> 00:54:43,440
that as you go and you end up with

1466
00:54:39,720 --> 00:54:45,359
the code being able to understand at uh

1467
00:54:43,440 --> 00:54:47,280
runtime basically what type it is

1468
00:54:45,359 --> 00:54:48,599
without having to copy and paste for

1469
00:54:47,280 --> 00:54:52,500
every different type that you want to

1470
00:54:48,599 --> 00:54:55,140
accept which is massively better

1471
00:54:52,500 --> 00:54:57,780
there's also a bunch more platforms so

1472
00:54:55,140 --> 00:55:00,720
AMD 64 is finally optimizations windows

1473
00:54:57,780 --> 00:55:03,359
and arm 64. risk Clive has quite a few

1474
00:55:00,720 --> 00:55:05,400
implementations and updates now long

1475
00:55:03,359 --> 00:55:07,260
Arch uh wessie they're for targeting

1476
00:55:05,400 --> 00:55:08,880
webassembly

1477
00:55:07,260 --> 00:55:11,280
and a whole bunch of other small

1478
00:55:08,880 --> 00:55:12,540
features that just make life nice like

1479
00:55:11,280 --> 00:55:15,200
Min and Max

1480
00:55:12,540 --> 00:55:18,119
in 2023

1481
00:55:15,200 --> 00:55:20,220
a profile guided optimization is also a

1482
00:55:18,119 --> 00:55:23,660
big one that can be very good for things

1483
00:55:20,220 --> 00:55:28,380
like kubernetes that run using go

1484
00:55:23,660 --> 00:55:30,960
atomics for Atomic purposes compressor

1485
00:55:28,380 --> 00:55:32,460
bug symbols all sorts of things fuzzing

1486
00:55:30,960 --> 00:55:35,160
address sanitizers

1487
00:55:32,460 --> 00:55:37,140
all very nice to have and it has gone a

1488
00:55:35,160 --> 00:55:38,400
long way since you might have looked so

1489
00:55:37,140 --> 00:55:40,619
I recommend

1490
00:55:38,400 --> 00:55:42,660
checking it out seeing if your gripes

1491
00:55:40,619 --> 00:55:44,640
with it in the past are still current

1492
00:55:42,660 --> 00:55:45,440
and if there are keep complaining about

1493
00:55:44,640 --> 00:55:48,480
them

1494
00:55:45,440 --> 00:55:52,440
thank you very much

1495
00:55:48,480 --> 00:55:54,540
thanks thank you hockey I'm

1496
00:55:52,440 --> 00:55:57,300
I'm excited to hear the next go talk at

1497
00:55:54,540 --> 00:56:01,319
pycon AU in 2030.

1498
00:55:57,300 --> 00:56:04,680
on Deck Peter Hall but first a talk

1499
00:56:01,319 --> 00:56:06,780
about somewhere to go with yaakov

1500
00:56:04,680 --> 00:56:08,339
uh you're putting going to everything

1501
00:56:06,780 --> 00:56:09,480
aren't you hello everyone

1502
00:56:08,339 --> 00:56:10,680
um just as a prefer some different

1503
00:56:09,480 --> 00:56:12,300
websites have some slightly different

1504
00:56:10,680 --> 00:56:13,859
numbers and slightly different Tales but

1505
00:56:12,300 --> 00:56:16,859
I put the story together as best I can

1506
00:56:13,859 --> 00:56:18,000
so Hello friends pycon this year is my

1507
00:56:16,859 --> 00:56:19,619
first time to Adelaide and I've been

1508
00:56:18,000 --> 00:56:21,839
using this time to research something

1509
00:56:19,619 --> 00:56:23,339
I'd heard rumors and whispers about

1510
00:56:21,839 --> 00:56:24,660
so put your hands up if you like Jewish

1511
00:56:23,339 --> 00:56:26,339
history

1512
00:56:24,660 --> 00:56:28,980
that's more than I expected put your

1513
00:56:26,339 --> 00:56:30,480
hands up even like nightclubs and bars

1514
00:56:28,980 --> 00:56:31,559
that's more of what I expected so

1515
00:56:30,480 --> 00:56:34,819
hopefully there's something in here for

1516
00:56:31,559 --> 00:56:37,920
everyone so in 1836 just

1517
00:56:34,819 --> 00:56:39,660
187 years ago Adelaide was colonized by

1518
00:56:37,920 --> 00:56:40,920
Europeans and as with much of the

1519
00:56:39,660 --> 00:56:43,740
history of Australia there have been

1520
00:56:40,920 --> 00:56:45,540
Jews there since the beginning in 1844

1521
00:56:43,740 --> 00:56:47,339
though there were just 25 Jews in the

1522
00:56:45,540 --> 00:56:48,420
colony here in Adelaide and as that

1523
00:56:47,339 --> 00:56:50,400
number grew they built the first

1524
00:56:48,420 --> 00:56:52,920
synagogue here in Adelaide and the

1525
00:56:50,400 --> 00:56:54,480
community still kept growing so this I

1526
00:56:52,920 --> 00:56:56,579
believe is the original synagogue scene

1527
00:56:54,480 --> 00:56:58,079
a number of years later I don't actually

1528
00:56:56,579 --> 00:57:01,140
know when this particular photo was

1529
00:56:58,079 --> 00:57:03,420
taken and originally it was 35 feet by

1530
00:57:01,140 --> 00:57:05,280
25 feet which is about 10 meters by

1531
00:57:03,420 --> 00:57:07,200
seven and a half give or take and it

1532
00:57:05,280 --> 00:57:09,780
could fit 150 people but that wasn't big

1533
00:57:07,200 --> 00:57:11,819
enough and it needed to be extended by

1534
00:57:09,780 --> 00:57:15,480
1860 just 10 years later the community

1535
00:57:11,819 --> 00:57:17,400
had grown to 360 members and by 1871 the

1536
00:57:15,480 --> 00:57:19,680
community built a new larger synagogue

1537
00:57:17,400 --> 00:57:21,780
right next door

1538
00:57:19,680 --> 00:57:23,760
the new one on the left was apparently

1539
00:57:21,780 --> 00:57:25,500
built in Italian style the old one in

1540
00:57:23,760 --> 00:57:26,579
Egyptian style I'm not an expert in this

1541
00:57:25,500 --> 00:57:28,440
I have no idea if this is what the

1542
00:57:26,579 --> 00:57:31,140
internet said and according to some

1543
00:57:28,440 --> 00:57:33,059
sources had a capacity for 370 people

1544
00:57:31,140 --> 00:57:34,500
which doesn't sound like good capacity

1545
00:57:33,059 --> 00:57:37,260
planning to me

1546
00:57:34,500 --> 00:57:40,200
20 years later in 1891 the community had

1547
00:57:37,260 --> 00:57:41,339
grown to 1 840 people and it was

1548
00:57:40,200 --> 00:57:43,079
suggested that the synagogue should be

1549
00:57:41,339 --> 00:57:45,000
redeveloped to provide an income for the

1550
00:57:43,079 --> 00:57:46,619
community so to do this they developed

1551
00:57:45,000 --> 00:57:48,540
five shops on Rundle Street which I

1552
00:57:46,619 --> 00:57:50,460
believe are still there as the Rundle

1553
00:57:48,540 --> 00:57:52,500
buildings on the corner but I didn't

1554
00:57:50,460 --> 00:57:54,300
look too closely when I was there

1555
00:57:52,500 --> 00:57:56,520
but as part of this Redevelopment in

1556
00:57:54,300 --> 00:57:58,859
1896 they moved the synagogue entrance

1557
00:57:56,520 --> 00:58:00,839
around to the side street and that side

1558
00:57:58,859 --> 00:58:02,700
feed Bears the name synagogue place to

1559
00:58:00,839 --> 00:58:03,720
this very day which is quite confusing

1560
00:58:02,700 --> 00:58:04,859
when you're researching coming to

1561
00:58:03,720 --> 00:58:06,359
Adelaide you're like oh is there a

1562
00:58:04,859 --> 00:58:07,859
synagogue and Google gives you results

1563
00:58:06,359 --> 00:58:09,420
for a street

1564
00:58:07,859 --> 00:58:11,700
and so let's reach there along with some

1565
00:58:09,420 --> 00:58:13,920
street art there's some graffiti and

1566
00:58:11,700 --> 00:58:16,200
then in 1938 the buildings received a

1567
00:58:13,920 --> 00:58:17,460
new art deco cement facade which was

1568
00:58:16,200 --> 00:58:19,380
probably this photo that I found

1569
00:58:17,460 --> 00:58:21,180
courtesy of Heritage of the city of

1570
00:58:19,380 --> 00:58:23,460
Adelaide you can see them again David

1571
00:58:21,180 --> 00:58:25,559
the Star of David in a circle at the

1572
00:58:23,460 --> 00:58:27,720
very top of the building above the

1573
00:58:25,559 --> 00:58:29,700
balcony and the words bait filler

1574
00:58:27,720 --> 00:58:32,400
engraved on the balcony itself which is

1575
00:58:29,700 --> 00:58:34,319
Hebrew for House of Prayer what's above

1576
00:58:32,400 --> 00:58:36,180
the main entrance and on the 11th of

1577
00:58:34,319 --> 00:58:38,400
September in 1986 this building was

1578
00:58:36,180 --> 00:58:42,540
granted Heritage status and is listed as

1579
00:58:38,400 --> 00:58:44,280
sa Heritage Place number 1345.

1580
00:58:42,540 --> 00:58:47,400
but why is this relevant to most of you

1581
00:58:44,280 --> 00:58:49,559
here well in 1990 the committee moved to

1582
00:58:47,400 --> 00:58:51,780
a new facility in Glenside and sold the

1583
00:58:49,559 --> 00:58:53,640
old freshly Heritage listed building

1584
00:58:51,780 --> 00:58:56,880
and it became a nightclub known as the

1585
00:58:53,640 --> 00:58:58,920
synagogue or synagogue nightclub

1586
00:58:56,880 --> 00:59:02,460
or for some strange reason in 2000

1587
00:58:58,920 --> 00:59:03,780
Church nightclub

1588
00:59:02,460 --> 00:59:05,220
I actually found this picture on a

1589
00:59:03,780 --> 00:59:06,960
Facebook group back from when it was the

1590
00:59:05,220 --> 00:59:08,760
church

1591
00:59:06,960 --> 00:59:10,859
and you could see like this elevated

1592
00:59:08,760 --> 00:59:12,599
forward Central area there's on the

1593
00:59:10,859 --> 00:59:14,099
western side of the main hall this would

1594
00:59:12,599 --> 00:59:16,380
have been the Anna karesh The Arc in

1595
00:59:14,099 --> 00:59:18,240
which the Torah scrolls are kept and

1596
00:59:16,380 --> 00:59:21,000
this is now at least when the photo was

1597
00:59:18,240 --> 00:59:22,799
isn't the DJ desk now

1598
00:59:21,000 --> 00:59:24,839
after that it became the Apple bar in

1599
00:59:22,799 --> 00:59:26,280
about 2009 which was not a place where

1600
00:59:24,839 --> 00:59:28,500
you go to get your computers fixed by

1601
00:59:26,280 --> 00:59:30,900
Geniuses but another interior

1602
00:59:28,500 --> 00:59:32,579
Redevelopment of the same site you can

1603
00:59:30,900 --> 00:59:34,619
see in this photo again that elaborate

1604
00:59:32,579 --> 00:59:36,359
front area very pixelated because these

1605
00:59:34,619 --> 00:59:39,119
are old photos and there's also

1606
00:59:36,359 --> 00:59:40,799
mezzanine level in a ring above which

1607
00:59:39,119 --> 00:59:42,599
was likely the original women's Gallery

1608
00:59:40,799 --> 00:59:44,160
a fairly common layout for synagogues

1609
00:59:42,599 --> 00:59:46,859
like this

1610
00:59:44,160 --> 00:59:50,220
since 2016 however it is now the Mary's

1611
00:59:46,859 --> 00:59:52,380
pop and gay bar open three nights a week

1612
00:59:50,220 --> 00:59:54,420
to be fair it is a gay bar with a very

1613
00:59:52,380 --> 00:59:55,680
friendly and curious security guard who

1614
00:59:54,420 --> 00:59:57,660
is more than happy to answer some of my

1615
00:59:55,680 --> 00:59:59,339
questions earlier this week

1616
00:59:57,660 --> 01:00:01,440
but while the text above the entrance

1617
00:59:59,339 --> 01:00:03,000
that once read House of Prayer is now

1618
01:00:01,440 --> 01:00:05,880
completely gone

1619
01:00:03,000 --> 01:00:07,920
the original Star of David remains

1620
01:00:05,880 --> 01:00:09,599
it's obscured slightly by the paintwork

1621
01:00:07,920 --> 01:00:11,760
but if you know what you're looking for

1622
01:00:09,599 --> 01:00:14,640
the site still does have telltale signs

1623
01:00:11,760 --> 01:00:16,200
of its Origins as the Adelaide synagogue

1624
01:00:14,640 --> 01:00:18,059
so if any of you are going to go hit up

1625
01:00:16,200 --> 01:00:19,440
this queer bar after the con please be

1626
01:00:18,059 --> 01:00:21,240
at least a little respectful of its

1627
01:00:19,440 --> 01:00:22,619
Heritage and just in general I would

1628
01:00:21,240 --> 01:00:24,720
suggest try and keep in mind and

1629
01:00:22,619 --> 01:00:28,760
appreciate history and the value of

1630
01:00:24,720 --> 01:00:28,760
whatever place you're in thank you

1631
01:00:29,400 --> 01:00:39,960
thank you yaakov thank you

1632
01:00:33,480 --> 01:00:43,260
hey uh on Deck we have uh Daisy and Lily

1633
01:00:39,960 --> 01:00:46,079
one talk to go

1634
01:00:43,260 --> 01:00:48,240
how many go puns can I no more uh it's a

1635
01:00:46,079 --> 01:00:50,180
Peter Hall

1636
01:00:48,240 --> 01:00:54,240
all right thank you Chris hi everyone

1637
01:00:50,180 --> 01:00:56,880
you have heard of this movie

1638
01:00:54,240 --> 01:00:58,260
it's the Fast and the Furious and you

1639
01:00:56,880 --> 01:01:01,680
have heard that this movie has weirdly

1640
01:00:58,260 --> 01:01:04,740
named sequels like Too Fast Too Furious

1641
01:01:01,680 --> 01:01:07,880
The Fast and the Furious Tokyo Drift and

1642
01:01:04,740 --> 01:01:11,520
so on and so forth for several movies

1643
01:01:07,880 --> 01:01:13,859
would you like to know why

1644
01:01:11,520 --> 01:01:15,420
this question will be answered we need a

1645
01:01:13,859 --> 01:01:16,799
history lesson about the development of

1646
01:01:15,420 --> 01:01:20,460
the Fast and the Furious

1647
01:01:16,799 --> 01:01:22,260
so in 1998 Rob Cohen read an article in

1648
01:01:20,460 --> 01:01:24,480
a magazine at about an illegal street

1649
01:01:22,260 --> 01:01:27,000
race in New York and thought I should

1650
01:01:24,480 --> 01:01:29,460
make a movie about street racing

1651
01:01:27,000 --> 01:01:30,480
what should we call that movie he asked

1652
01:01:29,460 --> 01:01:32,520
himself

1653
01:01:30,480 --> 01:01:34,559
why not just use the title from the

1654
01:01:32,520 --> 01:01:38,040
magazine article we can make an amazing

1655
01:01:34,559 --> 01:01:39,540
movie named Racer X

1656
01:01:38,040 --> 01:01:40,920
and that became the working title for

1657
01:01:39,540 --> 01:01:42,359
the movie

1658
01:01:40,920 --> 01:01:43,980
and they stayed with that title for

1659
01:01:42,359 --> 01:01:45,480
quite a while and they but eventually

1660
01:01:43,980 --> 01:01:47,700
they decided no this this title is not

1661
01:01:45,480 --> 01:01:50,760
plenty enough we need to change it a bit

1662
01:01:47,700 --> 01:01:52,440
so what was the title of the race that

1663
01:01:50,760 --> 01:01:55,200
happened in New York that's mentioned at

1664
01:01:52,440 --> 01:01:56,640
that magazine article it was called drag

1665
01:01:55,200 --> 01:01:58,680
wars

1666
01:01:56,640 --> 01:02:01,500
they decided not to use that particular

1667
01:01:58,680 --> 01:02:06,119
title but it did Inspire their next

1668
01:02:01,500 --> 01:02:07,740
title they went drag wars almost race

1669
01:02:06,119 --> 01:02:09,240
Wars

1670
01:02:07,740 --> 01:02:10,559
oh

1671
01:02:09,240 --> 01:02:13,140
now

1672
01:02:10,559 --> 01:02:15,720
I feel like this title has some really

1673
01:02:13,140 --> 01:02:17,700
obvious drawbacks

1674
01:02:15,720 --> 01:02:19,440
and so did they so they eventually

1675
01:02:17,700 --> 01:02:22,680
changed it again they came up with a new

1676
01:02:19,440 --> 01:02:24,000
title so race Wars connotations of

1677
01:02:22,680 --> 01:02:27,180
racial violence

1678
01:02:24,000 --> 01:02:29,520
the new title is Red Line

1679
01:02:27,180 --> 01:02:32,760
which if you're familiar with the uh

1680
01:02:29,520 --> 01:02:34,440
history of the US real estate also has

1681
01:02:32,760 --> 01:02:36,780
racial problems

1682
01:02:34,440 --> 01:02:39,660
but eventually

1683
01:02:36,780 --> 01:02:40,740
Neil Moritz the producer of The Fast and

1684
01:02:39,660 --> 01:02:45,180
the Furious

1685
01:02:40,740 --> 01:02:46,559
saw a documentary about Roger Corman the

1686
01:02:45,180 --> 01:02:49,280
director of

1687
01:02:46,559 --> 01:02:54,240
The Fast and the Furious

1688
01:02:49,280 --> 01:02:57,000
1954 and he went that that is the title

1689
01:02:54,240 --> 01:02:59,640
that I want for my movie

1690
01:02:57,000 --> 01:03:03,720
shame it's already a movie though

1691
01:02:59,640 --> 01:03:06,720
nonetheless let's um so just for context

1692
01:03:03,720 --> 01:03:09,900
the 1954 movie was described by variety

1693
01:03:06,720 --> 01:03:13,020
as a high-priced sports car bombs phone

1694
01:03:09,900 --> 01:03:14,700
furnish most of the action and racing

1695
01:03:13,020 --> 01:03:16,740
footage is interesting but becomes

1696
01:03:14,700 --> 01:03:19,020
repetitious and helps to string out the

1697
01:03:16,740 --> 01:03:20,400
running time to an unnecessary 73

1698
01:03:19,020 --> 01:03:22,799
minutes

1699
01:03:20,400 --> 01:03:25,220
an unhandy length for supporting play

1700
01:03:22,799 --> 01:03:25,220
dates

1701
01:03:25,859 --> 01:03:31,140
other reviews for similar

1702
01:03:28,740 --> 01:03:33,000
it wasn't a spectacular box office

1703
01:03:31,140 --> 01:03:35,040
success but they did manage to pay back

1704
01:03:33,000 --> 01:03:36,780
their investors

1705
01:03:35,040 --> 01:03:37,799
so Neil Moritz gets on the phone to

1706
01:03:36,780 --> 01:03:40,619
Universal

1707
01:03:37,799 --> 01:03:41,760
and says can we use this title I want my

1708
01:03:40,619 --> 01:03:45,200
movie to be called The Fast and the

1709
01:03:41,760 --> 01:03:45,200
Furious and Universal said

1710
01:03:45,780 --> 01:03:50,460
nothing at all they just went silent the

1711
01:03:48,359 --> 01:03:52,260
phone line went silent

1712
01:03:50,460 --> 01:03:54,119
and eventually I said look we'll get

1713
01:03:52,260 --> 01:03:55,859
back to you on that and after a few

1714
01:03:54,119 --> 01:03:57,540
hours they called back and said

1715
01:03:55,859 --> 01:03:59,280
all right let's do it

1716
01:03:57,540 --> 01:04:01,319
we will get on the phone with Roger

1717
01:03:59,280 --> 01:04:03,180
Corman and make arrangements so

1718
01:04:01,319 --> 01:04:04,980
Universal calls Roger Corman and says

1719
01:04:03,180 --> 01:04:05,940
hey can we use your tile and Roger was

1720
01:04:04,980 --> 01:04:07,920
like

1721
01:04:05,940 --> 01:04:09,540
well that movie is from 1954 it never

1722
01:04:07,920 --> 01:04:11,520
made much money no one ever really cared

1723
01:04:09,540 --> 01:04:12,720
about it I would be very happy to sell

1724
01:04:11,520 --> 01:04:14,400
you the rights to those targets your

1725
01:04:12,720 --> 01:04:17,579
Universal you're a huge company you have

1726
01:04:14,400 --> 01:04:18,900
a lot of money what can you give me and

1727
01:04:17,579 --> 01:04:20,700
Universal said

1728
01:04:18,900 --> 01:04:22,500
what if we gave you the use of some of

1729
01:04:20,700 --> 01:04:23,940
our stock footage archive and he said

1730
01:04:22,500 --> 01:04:25,820
well it's more than anyone else was

1731
01:04:23,940 --> 01:04:29,149
paying all right I'll take it

1732
01:04:25,820 --> 01:04:29,149
[Applause]

1733
01:04:29,819 --> 01:04:35,160
what if I want to make

1734
01:04:32,099 --> 01:04:38,040
49 years later the Fast and the Furious

1735
01:04:35,160 --> 01:04:40,740
2. so I'm keeping the rights to

1736
01:04:38,040 --> 01:04:42,599
numerical sequels

1737
01:04:40,740 --> 01:04:44,040
you can have the Fast and the Furious

1738
01:04:42,599 --> 01:04:45,420
but if you ever make a sequel you're

1739
01:04:44,040 --> 01:04:47,280
going to have to come up with your own

1740
01:04:45,420 --> 01:04:50,640
names

1741
01:04:47,280 --> 01:04:52,799
and that brings us here

1742
01:04:50,640 --> 01:04:56,160
so

1743
01:04:52,799 --> 01:04:59,940
Universal goes on to make seven billion

1744
01:04:56,160 --> 01:05:02,339
dollars in Gross like box office takings

1745
01:04:59,940 --> 01:05:04,859
from the combined entire Fast and the

1746
01:05:02,339 --> 01:05:06,420
Furious sorry the Fast and the Furious

1747
01:05:04,859 --> 01:05:08,400
franchise

1748
01:05:06,420 --> 01:05:10,200
Roger Corman received some stock footage

1749
01:05:08,400 --> 01:05:12,960
Which is far as I'm aware he has never

1750
01:05:10,200 --> 01:05:15,859
used for anything

1751
01:05:12,960 --> 01:05:15,859
thank you very much

1752
01:05:17,400 --> 01:05:22,940
Peter

1753
01:05:18,839 --> 01:05:22,940
Peter Hall Tokyo Drift everyone

1754
01:05:23,160 --> 01:05:27,780
get off the stage

1755
01:05:25,559 --> 01:05:30,180
right that is the end of the lightning

1756
01:05:27,780 --> 01:05:34,099
talks thank you to the AV people for

1757
01:05:30,180 --> 01:05:34,099
making us all look great again

1758
01:05:35,339 --> 01:05:38,520
thank you to Russell for waving his

1759
01:05:37,020 --> 01:05:41,000
laptop at people and encouraging

1760
01:05:38,520 --> 01:05:41,000
Applause

1761
01:05:42,660 --> 01:05:47,810
thanks thank you to today's elf and

1762
01:05:44,819 --> 01:05:51,720
occasional interpretive dancer Bano

1763
01:05:47,810 --> 01:05:55,319
[Applause]

1764
01:05:51,720 --> 01:05:57,920
to the voice of Bob uh sadly demiked

1765
01:05:55,319 --> 01:05:57,920
Libby Berry

1766
01:05:58,579 --> 01:06:03,359
and of course thank you to each and

1767
01:06:01,079 --> 01:06:04,859
every one of our 13 speakers

1768
01:06:03,359 --> 01:06:07,140
if you and

1769
01:06:04,859 --> 01:06:08,640
if you enjoyed today's lightning talks

1770
01:06:07,140 --> 01:06:11,000
you can't submit any tomorrow because

1771
01:06:08,640 --> 01:06:13,319
there are no lightning talks

1772
01:06:11,000 --> 01:06:14,819
but maybe there will be a pike on EU

1773
01:06:13,319 --> 01:06:16,920
next year and you can submit a lightning

1774
01:06:14,819 --> 01:06:18,900
talk then and until then

1775
01:06:16,920 --> 01:06:22,020
see you later and please give a

1776
01:06:18,900 --> 01:06:22,770
rapturous round of applause to a daisy

1777
01:06:22,020 --> 01:06:28,000
and Nick

1778
01:06:22,770 --> 01:06:28,000
[Applause]