1
00:00:01,550 --> 00:00:06,089
I'm David Farmer, from the

2
00:00:04,080 --> 00:00:07,740
American Institute of Mathematics.

3
00:00:06,089 --> 00:00:09,599
Today we're going to talk about the role

4
00:00:07,740 --> 00:00:11,490
of the job talk in the hiring process for a

5
00:00:09,599 --> 00:00:14,460
tenure-track position in mathematics.

6
00:00:11,490 --> 00:00:15,990
As we will see, the purpose of the job talk

7
00:00:14,460 --> 00:00:18,180
is to demonstrate that you have the

8
00:00:15,990 --> 00:00:20,250
potential to be a good teacher.

9
00:00:18,180 --> 00:00:25,320
It's not for you to blow everyone away with your

10
00:00:20,250 --> 00:00:25,520
wonderful research.

11
00:00:31,320 --> 00:00:35,640
Koutsoliotas: You've been invited to campus, and one of the

12
00:00:32,520 --> 00:00:38,700
the things you'll be asked to do is present a talk about your work.

13
00:00:35,640 --> 00:00:40,320
The job talk serves many different purposes,

14
00:00:38,700 --> 00:00:43,469
depending on what kind of school you're at.

15
00:00:40,320 --> 00:00:46,260
But one of the important things

16
00:00:43,469 --> 00:00:48,360
that is relevant to many schools

17
00:00:46,260 --> 00:00:50,850
is that it's the first time

18
00:00:48,360 --> 00:00:53,940
people will get to see you at the board

19
00:00:50,850 --> 00:00:56,640
or at the screen presenting your work

20
00:00:53,940 --> 00:00:58,469
That is a good way for people to tell

21
00:00:56,640 --> 00:01:00,420
how you are going to be

22
00:00:58,469 --> 00:01:03,329
in the classroom.  One of the things that

23
00:01:00,420 --> 00:01:05,580
you want to find out about is: who is

24
00:01:03,329 --> 00:01:07,380
your audience for this talk? 

25
00:01:05,580 --> 00:01:09,810
Is your talk going to be to the rest of the

26
00:01:07,380 --> 00:01:12,330
faculty, or is it going to be a talk

27
00:01:09,810 --> 00:01:13,409
primarily to undergraduate students?

28
00:01:12,330 --> 00:01:15,000
You want to find out a little bit about

29
00:01:13,409 --> 00:01:18,320
this, and that happens in the exchange as

30
00:01:15,000 --> 00:01:17,320
you are preparing for the interview.

31
00:01:18,390 --> 00:01:24,750
Farmer: Know the first sentence you're going to

32
00:01:21,310 --> 00:01:28,180
say and particularly the first word and

33
00:01:24,750 --> 00:01:30,430
exactly say that sentence.  It will calm you

34
00:01:28,180 --> 00:01:32,979
down and it'll be a good sentence.

35
00:01:30,430 --> 00:01:37,750
The most common first two words of talks is

36
00:01:32,979 --> 00:01:40,179
"Okay, so"  Okay?  Know your first word in your

37
00:01:37,750 --> 00:01:42,070
first sentence, and say it, and don't

38
00:01:40,179 --> 00:01:44,259
start with "Okay, so".  What I mean

39
00:01:42,070 --> 00:01:46,960
is the first sentence after the 'Thanks',

40
00:01:44,259 --> 00:01:49,420
if you give thanks.  Some people give

41
00:01:46,960 --> 00:01:50,770
thanks and some don't.  I happen to don't,

42
00:01:49,420 --> 00:01:53,049
and that's a reflection of my personality,

43
00:01:50,770 --> 00:01:56,170
not a reflection of best practice.

44
00:01:53,049 --> 00:01:58,600
Either way is fine; I'm guessing

45
00:01:56,170 --> 00:02:00,250
that at a job talk, thanking them

46
00:01:58,600 --> 00:02:02,350
for the opportunity to speak is

47
00:02:00,250 --> 00:02:05,170
probably a good thing, so you might want

48
00:02:02,350 --> 00:02:07,149
to err towards thanking, but it can be

49
00:02:05,170 --> 00:02:10,330
very brief.

50
00:02:07,149 --> 00:02:12,910
That 'thanks' should start with the

51
00:02:10,330 --> 00:02:14,319
word "thanks", not "okay so", and then the

52
00:02:12,910 --> 00:02:17,910
sentence after that should be the

53
00:02:14,319 --> 00:02:17,910
sentence you've prepared in advance.

54
00:02:18,260 --> 00:02:26,360
What do these five things have in common:

55
00:02:22,069 --> 00:02:30,500
hosting a talk, playing golf,

56
00:02:26,360 --> 00:02:33,340
playing poker, writing LaTeX, and giving a

57
00:02:30,500 --> 00:02:36,650
giving a colloquium or job talk.

58
00:02:33,340 --> 00:02:39,280
I've identified three things that

59
00:02:36,650 --> 00:02:42,049
they have in common:

60
00:02:39,280 --> 00:02:44,239
most people receive little formal training

61
00:02:42,049 --> 00:02:46,459
before embarking upon these endeavors.

62
00:02:44,239 --> 00:02:48,409
Everybody can play poker,

63
00:02:46,459 --> 00:02:51,019
everybody can play golf.

64
00:02:48,409 --> 00:02:52,069
So what you end up doing, since you have

65
00:02:51,019 --> 00:02:55,099
no formal training, is you either

66
00:02:52,069 --> 00:02:57,829
invent out of thin air what to do,

67
00:02:55,099 --> 00:02:59,150
or you copy the bad habits of other

68
00:02:57,829 --> 00:03:02,000
people.  As a consequence,

69
00:02:59,150 --> 00:03:05,450
most people do these things poorly.

70
00:03:02,000 --> 00:03:08,690
But also, for all of these

71
00:03:05,450 --> 00:03:12,829
just a few basic principles can make you

72
00:03:08,690 --> 00:03:13,489
far better than average.

73
00:03:12,829 --> 00:03:15,019
For the first one

74
00:03:13,489 --> 00:03:16,700
Sally explained the principles of

75
00:03:15,019 --> 00:03:19,400
hosting a talk:  you're already above

76
00:03:16,700 --> 00:03:21,170
average at doing that.  You need help

77
00:03:19,400 --> 00:03:23,840
from other people to improve your golf game

78
00:03:21,170 --> 00:03:25,400
game or to play better poker.  I'm happy

79
00:03:23,840 --> 00:03:27,919
to talk to you about writing LaTeX

80
00:03:25,400 --> 00:03:29,660
during your off time, but what I do want

81
00:03:27,919 --> 00:03:32,959
to tell you is some principles about

82
00:03:29,660 --> 00:03:36,440
giving a colloquium or job talk.

83
00:03:32,959 --> 00:03:41,629
Let me explain why I group colloquium

84
00:03:36,440 --> 00:03:43,639
and job talk together.  Lots of you have taught,

85
00:03:41,629 --> 00:03:45,709
maybe all of you have taught, and

86
00:03:43,639 --> 00:03:48,169
you've also written papers and you've

87
00:03:45,709 --> 00:03:50,109
either written or read textbooks.

88
00:03:48,169 --> 00:03:54,349
All of these things are types of mathematical

89
00:03:50,109 --> 00:03:56,810
exposition, and what I've shown here is

90
00:03:54,349 --> 00:03:59,810
how I see the spectrum of different

91
00:03:56,810 --> 00:04:01,699
types of mathematical exposition.

92
00:03:59,810 --> 00:04:03,889
The traditional, what we think of as

93
00:04:01,699 --> 00:04:06,440
"real mathematics", is very formal, and that's

94
00:04:03,889 --> 00:04:08,540
at the top.  It goes definition-theorem-

95
00:04:06,440 --> 00:04:10,250
proof-theorem-proof-theorem-proof.

96
00:04:08,540 --> 00:04:12,859
That's what your research papers look like, and

97
00:04:10,250 --> 00:04:17,239
that's what seminar talks tend to look like.

98
00:04:12,859 --> 00:04:19,400
At the other end you have things like teaching,

99
00:04:17,239 --> 00:04:21,769
and the job talk, and the colloquium,

100
00:04:19,400 --> 00:04:25,510
which are very informal.

101
00:04:21,769 --> 00:04:28,390
They're driven by examples.

102
00:04:25,510 --> 00:04:30,730
You have terminology more than you have

103
00:04:28,390 --> 00:04:33,010
formal definitions, and while there are

104
00:04:30,730 --> 00:04:34,720
theorems and proofs - somebody once said

105
00:04:33,010 --> 00:04:37,690
every good math talk contains a proof -

106
00:04:34,720 --> 00:04:39,400
that's probably true, but often it's a

107
00:04:37,690 --> 00:04:40,810
"proof" and I've written in quotes because

108
00:04:39,400 --> 00:04:43,180
it's not like you're nailing down

109
00:04:40,810 --> 00:04:45,040
every single detail.

110
00:04:43,180 --> 00:04:47,910
If I explain the idea and you get, that counts

111
00:04:45,040 --> 00:04:52,030
as a proof in the context of the talk.

112
00:04:47,910 --> 00:04:54,310
In this spectrum, at the top,

113
00:04:52,030 --> 00:04:57,310
who's the audience of a research paper?

114
00:04:54,310 --> 00:05:01,210
One person reading it repeatedly for a long time.

115
00:04:57,310 --> 00:05:03,460
In seminar the audience is a small

116
00:05:01,210 --> 00:05:05,350
group of homogeneous people whom you can

117
00:05:03,460 --> 00:05:07,540
assume have something in common.

118
00:05:05,350 --> 00:05:09,220
By the time you get down to the bottom,

119
00:05:07,540 --> 00:05:11,530
there are more people and there's a great

120
00:05:09,220 --> 00:05:13,870
diversity among those people.

121
00:05:11,530 --> 00:05:18,460
That has to affect how you do your exposition.

122
00:05:13,870 --> 00:05:20,230
Let me elaborate a little bit

123
00:05:18,460 --> 00:05:22,420
more about what I mean by examples and terminology.

124
00:05:20,230 --> 00:05:25,680
This is a math talk, so

???????
125
00:05:22,420 --> 00:05:27,640
we need theorems and proofs.  Our first,

126
00:05:25,680 --> 00:05:30,430
well it's a proposition,

127
00:05:27,640 --> 00:05:32,800
the proposition is:  Usually it is not

128
00:05:30,430 --> 00:05:35,020
helpful to state a formal definition in

129
00:05:32,800 --> 00:05:37,900
the colloquium talk.  What's the proof?

130
00:05:35,020 --> 00:05:40,210
The proof breaks into cases

131
00:05:37,900 --> 00:05:43,510
depending on the audience members.

132
00:05:40,210 --> 00:05:44,770
Case 1 is: the people in the audience

133
00:05:43,510 --> 00:05:47,470
who already know the definition.

134
00:05:44,770 --> 00:05:49,780
Obviously it's a waste of time

135
00:05:47,470 --> 00:05:52,270
to tell them the definition.

136
00:05:49,780 --> 00:05:54,910
Case 2: the people who don't already know

137
00:05:52,270 --> 00:05:57,460
the definition.  Well, unless it's

138
00:05:54,910 --> 00:05:59,620
something really simple, telling them the

139
00:05:57,460 --> 00:06:01,900
definition is probably also not helpful.

140
00:05:59,620 --> 00:06:04,660
They need examples, discussion, and

141
00:06:01,900 --> 00:06:06,700
context to understand it.  If my talk is on

142
00:06:04,660 --> 00:06:09,040
Hopf algebras, writing down a

143
00:06:06,700 --> 00:06:10,600
commutative diagram with seven things in it

144
00:06:09,040 --> 00:06:13,300
(part of the definition of a Hopf algebra)

145
00:06:10,600 --> 00:06:15,640
is not going to tell you what a Hopf algebra

146
00:06:13,300 --> 00:06:18,700
really is.  On the other hand,

147
00:06:15,640 --> 00:06:20,590
you are all aware of objects which can

148
00:06:18,700 --> 00:06:23,260
be given a Hopf algebra structure.

149
00:06:20,590 --> 00:06:25,600
I could talk about those objects, describe

150
00:06:23,260 --> 00:06:27,250
which aspects of them are relevant, and

151
00:06:25,600 --> 00:06:29,470
then say, "the things with these properties

152
00:06:27,250 --> 00:06:31,600
are called Hopf algebras". 

153
00:06:29,470 --> 00:06:33,760
I'm not giving a definition, but I am

154
00:06:31,600 --> 00:06:37,870
giving terminology, and I'm also

155
00:06:33,760 --> 00:06:39,520
advocating that the examples come first.

156
00:06:37,870 --> 00:06:41,110
So it's not definition followed by example.

157
00:06:39,520 --> 00:06:43,330
It's example followed by definition,

158
00:06:41,110 --> 00:06:46,600
or better yet, example followed by terminology.

159
00:06:43,330 --> 00:06:48,940
We have already discussed the outline.

160
00:06:46,600 --> 00:06:53,949
The next proposition is:

161
00:06:48,940 --> 00:06:56,380
Never use an outline slide.

162
00:06:53,949 --> 00:06:58,139
The proof also breaks into two cases,

163
00:06:56,380 --> 00:07:01,360
depending upon the nature of the outline.

164
00:06:58,139 --> 00:07:03,850
If your outline is completely

165
00:07:01,360 --> 00:07:06,669
devoid of relevant content to your talk

166
00:07:03,850 --> 00:07:09,900
and could apply to any other talk,

167
00:07:06,669 --> 00:07:12,790
of course it's waste of time.

168
00:07:09,900 --> 00:07:15,280
On the other hand, maybe your outline

169
00:07:12,790 --> 00:07:17,110
does mention some key topics.  The one from

170
00:07:15,280 --> 00:07:19,270
my bad talk did: it had "other symmetry types".

171
00:07:17,110 --> 00:07:21,550
What was I forced to do?

172
00:07:19,270 --> 00:07:24,310
I was forced to try to explain some

173
00:07:21,550 --> 00:07:26,139
complicated thing from the end of my talk

174
00:07:24,310 --> 00:07:29,260
in the middle of explaining the outline,

175
00:07:26,139 --> 00:07:30,910
which satisfied no one.

176
00:07:29,260 --> 00:07:32,080
The alternative is don't even try to

177
00:07:30,910 --> 00:07:32,710
explain it, and then it's even more

178
00:07:32,080 --> 00:07:34,930
confusing.  So, such outlines

179
00:07:32,710 --> 00:07:37,060
also are a waste of time.  Since that

180
00:07:34,930 --> 00:07:41,260
covers all cases, that completes the proof.

181
00:07:37,060 --> 00:07:43,450
I know you've seen

182
00:07:41,260 --> 00:07:46,389
lots of talks with outlines, so let me

183
00:07:43,450 --> 00:07:48,210
say what to do instead.  This was

184
00:07:46,389 --> 00:07:50,889
already mentioned in the discussion.

185
00:07:48,210 --> 00:07:54,070
You want to think of giving the talk as

186
00:07:50,889 --> 00:07:56,320
storytelling.  Start just by beginning to

187
00:07:54,070 --> 00:07:58,330
tell the story. I was greatly

188
00:07:56,320 --> 00:08:01,620
influenced in this by Henryk Iwanice,

189
00:07:58,330 --> 00:08:04,210
who is at Rutgers.  In his graduate course

190
00:08:01,620 --> 00:08:05,740
every semester, all the students have to

191
00:08:04,210 --> 00:08:07,780
go to the board and present some result.

192
00:08:05,740 --> 00:08:10,060
What do they do?  They walk to the

193
00:08:07,780 --> 00:08:12,580
board and they start writing.  Henrik says,

194
00:08:10,060 --> 00:08:14,590
"Say some words".  So they turn around

195
00:08:12,580 --> 00:08:17,830
confused, like, "What do you mean?".  He

196
00:08:14,590 --> 00:08:20,260
repeats, "Say some words", at which time

197
00:08:17,830 --> 00:08:21,639
they start talking about the result

198
00:08:20,260 --> 00:08:24,340
they're going to present and it's proof.

199
00:08:21,639 --> 00:08:27,070
It's almost always extremely clear,

200
00:08:24,340 --> 00:08:29,289
a great introduction.  That, I think,

201
00:08:27,070 --> 00:08:31,570
is a great way to start any colloquium talk.

202
00:08:29,289 --> 00:08:34,479
Just "say some words".  You may actually

203
00:08:31,570 --> 00:08:36,909
mention quite a bit or just part

204
00:08:34,479 --> 00:08:39,279
of your outline.  The outline still

205
00:08:36,909 --> 00:08:42,190
exists as an abstract object describing

206
00:08:39,279 --> 00:08:45,250
your talk.  I'm just telling you

207
00:08:42,190 --> 00:08:46,959
not to put it up on the screen.

208
00:08:45,250 --> 00:08:49,360
The outline will inform your exposition

209
00:08:46,959 --> 00:08:49,960
and you may in fact mention things about

210
00:08:49,360 --> 00:08:51,430
it in your introduction, but it's not

211
00:08:49,960 --> 00:08:56,080
going to look like an outline.

212
00:08:51,430 --> 00:08:57,670
I don't want to

213
00:08:56,080 --> 00:09:00,190
overwhelm you with lots of do's and

214
00:08:57,670 --> 00:09:02,980
don'ts, but let me just get to a

215
00:09:00,190 --> 00:09:05,080
few more principles.  The first one we

216
00:09:02,980 --> 00:09:07,240
already mentioned: know the first

217
00:09:05,080 --> 00:09:09,400
sentence you're going to say, know whether

218
00:09:07,240 --> 00:09:11,230
or not you're going to thank the person

219
00:09:09,400 --> 00:09:15,160
who introduced you or thank the

220
00:09:11,230 --> 00:09:16,630
institution, before you begin.

221
00:09:15,160 --> 00:09:19,240
Second we've just started talking about the

222
00:09:16,630 --> 00:09:21,130
idea of storytelling.  You're telling a story.

223
00:09:19,240 --> 00:09:24,010
What does it mean to tell a story?

224
00:09:21,130 --> 00:09:26,580
You set the scene, you introduce the

225
00:09:24,010 --> 00:09:30,400
characters, you do character development.

226
00:09:26,580 --> 00:09:32,290
There is a plot, you foreshadow things

227
00:09:30,400 --> 00:09:36,450
which are going to come;

228
00:09:32,290 --> 00:09:39,280
there's connections, and there's excitement.

229
00:09:36,450 --> 00:09:41,380
Part of the idea of a short story is

230
00:09:39,280 --> 00:09:45,100
that it's meant to work toward one thing.

231
00:09:41,380 --> 00:09:47,620
As you're planning your job talk decide

232
00:09:45,100 --> 00:09:49,360
first what of your results is going to be

233
00:09:47,620 --> 00:09:50,770
the culmination of your talk.

234
00:09:49,360 --> 00:09:53,740
What exciting thing are you going to tell them about?

235
00:09:50,770 --> 00:09:56,350
Once you know that, you want to

236
00:09:53,740 --> 00:09:58,540
work backward to the beginning.

237
00:09:56,350 --> 00:10:00,670
Don't include anything that's not relevant to

238
00:09:58,540 --> 00:10:03,570
the thing you want to tell them about.

239
00:10:00,670 --> 00:10:06,580
It's not a novel, it's a short story.

240
00:10:03,570 --> 00:10:09,250
There's two ways in which the material

241
00:10:06,580 --> 00:10:12,040
you present can be useful:  you want

242
00:10:09,250 --> 00:10:14,490
people to understand and appreciate your work.

243
00:10:12,040 --> 00:10:17,110
Those are two different things.

244
00:10:14,490 --> 00:10:19,900
Understanding requires some definitions

245
00:10:17,110 --> 00:10:22,390
or terminology.

246
00:10:19,900 --> 00:10:24,220
Appreciate involves some amount of context about

247
00:10:22,390 --> 00:10:25,960
what has been done before, or what are the

248
00:10:24,220 --> 00:10:29,260
consequences of what you are doing.

249
00:10:25,960 --> 00:10:30,790
Keep in mind that, people

250
00:10:29,260 --> 00:10:33,280
don't really understand your results unless

251
00:10:30,790 --> 00:10:34,870
they understand [laughs] and appreciate them.

252
00:10:33,280 --> 00:10:37,570
So you're doing two things.

253
00:10:34,870 --> 00:10:39,430
Eradicate everything that's

254
00:10:37,570 --> 00:10:42,400
not relevant to that in your talk,

255
00:10:39,430 --> 00:10:46,120
and start by working backward.

256
00:10:42,400 --> 00:10:48,490
The third main point here is transitions.

257
00:10:46,120 --> 00:10:50,550
The outline which you are not showing is kind

258
00:10:48,490 --> 00:10:55,010
of like chapters or sections of your story.

259
00:10:50,550 --> 00:10:57,170
Between chapters,

260
00:10:55,010 --> 00:11:00,079
things shouldn't change abruptly.  Things

261
00:10:57,170 --> 00:11:02,810
should flow in a natural way.  When you've

262
00:11:00,079 --> 00:11:05,480
ended one part of your talk and starting

263
00:11:02,810 --> 00:11:07,930
another part, you can do a transition by

264
00:11:05,480 --> 00:11:10,100
wrapping up what just happened and

265
00:11:07,930 --> 00:11:11,779
introducing what's going to happen next.

266
00:11:10,100 --> 00:11:13,190
You are doing little mini wrap-ups

267
00:11:11,779 --> 00:11:16,010
and introductions throughout your talk.

268
00:11:13,190 --> 00:11:17,899
This is a very effective way of

269
00:11:16,010 --> 00:11:20,750
giving the impression that you have a

270
00:11:17,899 --> 00:11:22,279
clear and organized talk.

271
00:11:20,750 --> 00:11:24,279
You want people to leave saying,

272
00:11:22,279 --> 00:11:29,779
"I followed that",  "that was pretty good."

273
00:11:24,279 --> 00:11:31,790
A really great transition:  if someone

274
00:11:29,779 --> 00:11:33,709
woke up right before you did it,

275
00:11:31,790 --> 00:11:35,540
they'll be on board ready to pay

276
00:11:33,709 --> 00:11:37,639
attention again.  Or if they arrive

277
00:11:35,540 --> 00:11:40,970
late, they'll be on board.

278
00:11:37,639 --> 00:11:43,730
Let me do a transition,  way in the

279
00:11:40,970 --> 00:11:48,649
middle of a talk I was just giving.

280
00:11:43,730 --> 00:11:52,370
... so that's the

281
00:11:48,649 --> 00:11:54,410
conjecture we made about the maximum

282
00:11:52,370 --> 00:11:56,540
size of that Z function.  Recall from the

283
00:11:54,410 --> 00:11:58,430
first slide: it got bigger when T was larger,

284
00:11:56,540 --> 00:12:01,279
and that's our conjecture for how

285
00:11:58,430 --> 00:12:03,199
big it gets.  We got this conjecture

286
00:12:01,279 --> 00:12:05,750
in multiple ways by thinking

287
00:12:03,199 --> 00:12:08,180
probabilisticly.  One of the tools used

288
00:12:05,750 --> 00:12:11,120
is what's known as a moment of the zeta

289
00:12:08,180 --> 00:12:13,370
function.  The 2kth moment of the

290
00:12:11,120 --> 00:12:16,610
zeta function is defined as the average

291
00:12:13,370 --> 00:12:20,269
of the 2kth absolute value of the

292
00:12:16,610 --> 00:12:23,269
zeta function on the half-line.

293
00:12:20,269 --> 00:12:25,639
Conjectures for how big this moment is, or

294
00:12:23,269 --> 00:12:28,220
the size of this moment, will be one of the ways

295
00:12:25,639 --> 00:12:31,600
that we arrive at that conjecture for

296
00:12:28,220 --> 00:12:36,620
how big the function itself gets.

297
00:12:31,600 --> 00:12:39,079
That's my transition, okay?  So the idea is

298
00:12:36,620 --> 00:12:41,269
you're supposed to be willing to start

299
00:12:39,079 --> 00:12:44,149
paying attention again if you

300
00:12:41,269 --> 00:12:46,610
dozed or stopped paying attention.

301
00:12:44,149 --> 00:12:49,819
Punctuating these throughout your

302
00:12:46,610 --> 00:12:51,350
talk is helpful and it can't hurt to be

303
00:12:49,819 --> 00:12:52,819
really blatant about it.  If you're

304
00:12:51,350 --> 00:12:55,579
standing here while you're explaining

305
00:12:52,819 --> 00:12:57,860
something, step away.  It will draw their

306
00:12:55,579 --> 00:12:59,480
attention.  Summarize what you just did;

307
00:12:57,860 --> 00:13:01,399
introduce what's going to happen next.

308
00:12:59,480 --> 00:13:03,390
People will get the impression that

309
00:13:01,399 --> 00:13:05,910
they're following what's going on.

310
00:13:03,390 --> 00:13:09,060
You rarely want to read your slides,

311
00:13:05,910 --> 00:13:12,510
so I don't think you need a slide for the

312
00:13:09,060 --> 00:13:14,100
transition.  One of

313
00:13:12,510 --> 00:13:16,290
the exercises you want to do once your

314
00:13:14,100 --> 00:13:17,700
talk is written out, is figure out what

315
00:13:16,290 --> 00:13:20,850
are the sections and how to do the transitions.

316
00:13:17,700 --> 00:13:27,780
It's perfectly fine

317
00:13:20,850 --> 00:13:29,250
to end abruptly.  Thank you.

318
00:13:27,780 --> 00:13:31,380
I'll answer but other people may contradict me.

319
00:13:29,250 --> 00:13:32,910
I think you want to give what will be

320
00:13:31,380 --> 00:13:35,340
a good talk of interest to the whole

321
00:13:32,910 --> 00:13:37,440
department.  If there happens to be people

322
00:13:35,340 --> 00:13:39,510
in the department interested in your

323
00:13:37,440 --> 00:13:41,610
particular research, then 

324
00:13:39,510 --> 00:13:43,800
you'll talk to them personally about

325
00:13:41,610 --> 00:13:47,750
these other more important results.  But

326
00:13:43,800 --> 00:13:50,910
the job talk iss more about:

327
00:13:47,750 --> 00:13:53,400
'will you be okay in the classroom' than

328
00:13:50,910 --> 00:13:56,250
it is 'persuade me that you're a great

329
00:13:53,400 --> 00:13:59,880
researcher'.  At some big R1s maybe it's

330
00:13:56,250 --> 00:14:01,950
different, but a place that at all values

331
00:13:59,880 --> 00:14:05,310
teaching, that's really the secret reason.

332
00:14:01,950 --> 00:14:06,660
Many speakers, after one minute

333
00:14:05,310 --> 00:14:08,370
you might as well not pay attention.

334
00:14:06,660 --> 00:14:10,170
It's obvious that they're not going to

335
00:14:08,370 --> 00:14:11,460
be good in the classroom.

336
00:14:10,170 --> 00:14:13,860
That's what I'm trying to help you avoid.

337
00:14:11,460 --> 00:14:17,130
I wouldn't choose something you've moved

338
00:14:13,860 --> 00:14:18,630
on from, because 

339
00:14:17,130 --> 00:14:20,370
one of the obvious questions is:

340
00:14:18,630 --> 00:14:23,040
What are you planning to do next?

341
00:14:20,370 --> 00:14:25,200
Those future hosts in the audience,

342
00:14:23,040 --> 00:14:27,090
you know that 'what things are you planning

343
00:14:25,200 --> 00:14:28,770
to work on next' is an almost always safe question,

344
00:14:27,090 --> 00:14:31,080
if no one else can think of a question.

345
00:14:28,770 --> 00:14:34,680
You don't want to say,

346
00:14:31,080 --> 00:14:37,470
'well, I'm not working on that anymore'.

347
00:14:34,680 --> 00:14:39,630
Don't use your thesis work if you've really

348
00:14:37,470 --> 00:14:41,250
moved on to a totally new area.

349
00:14:39,630 --> 00:14:43,710
I think all of us are telling you that

350
00:14:41,250 --> 00:14:45,660
it's the quality of the talk and the

351
00:14:43,710 --> 00:14:47,490
entertainment value more than the

352
00:14:45,660 --> 00:14:49,620
high-powered mathematics in the talk.

353
00:14:47,490 --> 00:14:51,840
You're going to meet with many of the

354
00:14:49,620 --> 00:14:53,520
people in the department and you'll talk

355
00:14:51,840 --> 00:14:55,890
to them about your current research

356
00:14:53,520 --> 00:14:58,500
interests, and they'll have

357
00:14:55,890 --> 00:15:00,660
one-on-one sessions with you.  The talk

358
00:14:58,500 --> 00:15:02,960
isn't doing everything - it's not doing

359
00:15:00,660 --> 00:15:02,960
all that.

360
00:15:03,370 --> 00:15:07,630
Koutsoliotas: we do have our

361
00:15:05,950 --> 00:15:09,520
undergraduates at the talks as well,

362
00:15:07,630 --> 00:15:12,370
and we expect that the undergraduates

363
00:15:09,520 --> 00:15:14,830
will follow the first 10 or 15 minutes,

364
00:15:12,370 --> 00:15:16,510
and then they're going to be

365
00:15:14,830 --> 00:15:18,640
be dropping off.

366
00:15:16,510 --> 00:15:21,310
The more junior ones first and then

367
00:15:18,640 --> 00:15:23,860
the later ones.   That's a

368
00:15:21,310 --> 00:15:26,350
hard talk to prepare for because

369
00:15:23,860 --> 00:15:28,000
you're really talking to the faculty, but

370
00:15:26,350 --> 00:15:29,170
you want to start at a level where some

371
00:15:28,000 --> 00:15:31,660
students will at least get a sense of it.

372
00:15:29,170 --> 00:15:33,370
A nice way to do it, is to wrap up at

373
00:15:31,660 --> 00:15:34,750
the end with future projects.

374
00:15:33,370 --> 00:15:36,010
You can bring together the questions

375
00:15:34,750 --> 00:15:39,070
that you opened at the

376
00:15:36,010 --> 00:15:41,140
beginning, to bring

377
00:15:39,070 --> 00:15:43,210
the people who did fall

378
00:15:41,140 --> 00:15:46,050
off back together.

379
00:15:43,210 --> 00:15:46,050
It wraps it up.

380
00:15:46,240 --> 00:15:51,160
If you can give a good chalk talk, that

381
00:15:48,910 --> 00:15:55,899
would be good.  There's nothing wrong with

382
00:15:51,160 --> 00:15:58,480
a good chalk talk.  I couldn't give this

383
00:15:55,899 --> 00:16:02,250
talk, because I need to project that image,

384
00:15:58,480 --> 00:16:04,630
so I would have to change my topic.

385
00:16:02,250 --> 00:16:07,360
That doesn't even mean I would use this

386
00:16:04,630 --> 00:16:09,490
as a job talk.  I might not because I

387
00:16:07,360 --> 00:16:12,010
might choose to give a chalk talk.

388
00:16:09,490 --> 00:16:13,779
Koutsoliotas: Even when you are using slides

389
00:16:12,010 --> 00:16:15,910
the blackboard is always within arm's reach

390
00:16:13,779 --> 00:16:18,130
and, especially during question time,

391
00:16:15,910 --> 00:16:21,580
reaching for a piece of chalk and working,

392
00:16:18,130 --> 00:16:24,010
really indicates a

393
00:16:21,580 --> 00:16:25,810
comfortableness about 

394
00:16:24,010 --> 00:16:27,970
conveying information.  I think that's

395
00:16:25,810 --> 00:16:29,260
always at your disposal.  You shouldn't

396
00:16:27,970 --> 00:16:32,589
feel like you're not allowed to pick up

397
00:16:29,260 --> 00:16:35,459
anything except the stupid

398
00:16:32,589 --> 00:16:35,459
laser pointer.

399
00:16:35,710 --> 00:16:41,770
It depends upon why it happened.

400
00:16:38,580 --> 00:16:43,540
You better have practiced and revised

401
00:16:41,770 --> 00:16:45,100
your job talked several times with

402
00:16:43,540 --> 00:16:49,060
several different friends watching it.

403
00:16:45,100 --> 00:16:51,010
I'm more concerned that you're

404
00:16:49,060 --> 00:16:52,240
running short on time and you haven't

405
00:16:51,010 --> 00:16:54,760
planned ahead what you're going to

406
00:16:52,240 --> 00:16:57,340
jettison.  It's very unlikely you want

407
00:16:54,760 --> 00:17:00,220
to jettison the last few slides, so you

408
00:16:57,340 --> 00:17:01,960
need to have a plan.  What if a

409
00:17:00,220 --> 00:17:04,060
great thing happens, and there's loads of

410
00:17:01,960 --> 00:17:05,830
questions in the early part of the talk,

411
00:17:04,060 --> 00:17:07,510
and all of the sudden you're behind.  You

412
00:17:05,830 --> 00:17:10,060
start panicking and you haven't planned

413
00:17:07,510 --> 00:17:11,920
what you can skip.  So you need to

414
00:17:10,060 --> 00:17:16,630
know in advance what you're willing to skip.

415
00:17:11,920 --> 00:17:18,459
My question to you is:

416
00:17:16,630 --> 00:17:20,290
why are you getting flustered at the

417
00:17:18,459 --> 00:17:22,270
beginning?  If it's because you didn't

418
00:17:20,290 --> 00:17:24,220
practice and you don't have it clear in

419
00:17:22,270 --> 00:17:26,770
your mind, then that's totally your

420
00:17:24,220 --> 00:17:28,240
fault and there's no good answer.

421
00:17:26,770 --> 00:17:30,820
The answer is you've already made the

422
00:17:28,240 --> 00:17:32,860
mistake.  Another thing you might want to

423
00:17:30,820 --> 00:17:35,050
do when you give a slide talk:

424
00:17:32,860 --> 00:17:38,140
get a print out that has about a

425
00:17:35,050 --> 00:17:40,390
dozen slides per sheet.  A really

426
00:17:38,140 --> 00:17:43,180
awful thing that happens is the person

427
00:17:40,390 --> 00:17:44,740
goes over, and then, "Oh, that's what the

428
00:17:43,180 --> 00:17:46,240
next slide is about."  The next slide

429
00:17:44,740 --> 00:17:47,920
shouldn't surprise you, but it happens all

430
00:17:46,240 --> 00:17:49,360
the time:  they hit 'next' and then they

431
00:17:47,920 --> 00:17:52,210
find out what the next part of the talk

432
00:17:49,360 --> 00:17:54,460
is about.  That's easy to happen, but

433
00:17:52,210 --> 00:17:56,290
if you have on the table in front of you

434
00:17:54,460 --> 00:17:58,420
the whole talk on three sheets of paper,

435
00:17:56,290 --> 00:18:00,580
which can be done, you're going to know

436
00:17:58,420 --> 00:18:03,040
the next slide.  The hard part about a

437
00:18:00,580 --> 00:18:05,680
transition is you start it before you

438
00:18:03,040 --> 00:18:07,510
get to the next thing.  You can't do a

439
00:18:05,680 --> 00:18:09,730
good transition if you don't know

440
00:18:07,510 --> 00:18:11,740
exactly what the next slide is, and it's

441
00:18:09,730 --> 00:18:13,240
very difficult to do it without notes.

442
00:18:11,740 --> 00:18:15,160
It was a challenge for me, and I only have

443
00:18:13,240 --> 00:18:18,610
four slides, and I practiced my talk

444
00:18:15,160 --> 00:18:22,360
with Sally this morning.

445
00:18:18,610 --> 00:18:27,970
It is a stressful situation.  What I do is,

446
00:18:22,360 --> 00:18:29,789
draw on the sheets of paper which has

447
00:18:27,970 --> 00:18:36,059
each of the slides.

448
00:18:29,789 --> 00:18:38,190
I draw an L to indicate where

449
00:18:36,059 --> 00:18:39,779
is the end of each part.  I need to

450
00:18:38,190 --> 00:18:40,679
know when is a transition.

451
00:18:39,779 --> 00:18:43,409
I can unobtrusively just glance down

452
00:18:40,679 --> 00:18:48,389
at the table and see it.

453
00:18:43,409 --> 00:18:51,210
To try to actually answer your question,

454
00:18:48,389 --> 00:18:54,090
you may want to just leap and

455
00:18:51,210 --> 00:18:55,710
do the transition - maybe a slightly more

456
00:18:54,090 --> 00:18:58,470
elaborate transition, and get to the next

457
00:18:55,710 --> 00:19:02,249
part and get people caught up.

458
00:18:58,470 --> 00:19:04,440
You need preparation in order to do this.

459
00:19:02,249 --> 00:19:06,389
No one just wings talks

460
00:19:04,440 --> 00:19:08,639
(except for Loek).  No one wings good talks.

461
00:19:06,389 --> 00:19:14,429
People prepare them and write them out.

462
00:19:08,639 --> 00:19:17,580
I don't think you can over-prepare a job talk.

463
00:19:14,429 --> 00:19:20,399
You need to have down cold the transitions.

464
00:19:17,580 --> 00:19:21,479
Koutsoliotas:  Transitions - I think of

465
00:19:20,399 --> 00:19:23,129
it as, there's different

466
00:19:21,479 --> 00:19:25,289
sections to the talk and some sections

467
00:19:23,129 --> 00:19:27,119
can get really technical and people can

468
00:19:25,289 --> 00:19:28,559
kind of doze off, especially if

469
00:19:27,119 --> 00:19:31,109
they're on the edge of understanding to

470
00:19:28,559 --> 00:19:33,059
begin with.  So I really like this

471
00:19:31,109 --> 00:19:35,129
idea of, when you get to the end of the

472
00:19:33,059 --> 00:19:37,019
technical part, you just step

473
00:19:35,129 --> 00:19:38,999
aside and say, "so now that we've done the

474
00:19:37,019 --> 00:19:40,409
technical details, what we've shown is

475
00:19:38,999 --> 00:19:42,330
blah blah blah and we're going to be

476
00:19:40,409 --> 00:19:44,519
able to use this piece to be able to

477
00:19:42,330 --> 00:19:46,769
show this...".  All of the sudden

478
00:19:44,519 --> 00:19:48,539
it's like you've been

479
00:19:46,769 --> 00:19:49,889
granted a pardon to say, 'well, I

480
00:19:48,539 --> 00:19:52,349
don't understand that technical stuff,

481
00:19:49,889 --> 00:19:53,879
but now I know the piece that

482
00:19:52,349 --> 00:19:55,769
matters and that's a piece that I need

483
00:19:53,879 --> 00:19:57,119
to carry forth.'  That's a good way of

484
00:19:55,769 --> 00:19:58,919
recovering if you feel like you're

485
00:19:57,119 --> 00:20:01,379
losing people: say "so what we've just

486
00:19:58,919 --> 00:20:03,479
done so what I hope I've shown

487
00:20:01,379 --> 00:20:04,739
you is blah blah blah and so now we're

488
00:20:03,479 --> 00:20:06,869
going to use it to do this...."

489
00:20:04,739 --> 00:20:09,029
You need to be

490
00:20:06,869 --> 00:20:10,799
familiar with the structure of

491
00:20:09,029 --> 00:20:12,690
your talk so that you can gather people up.

492
00:20:10,799 --> 00:20:14,729
It is like this:

493
00:20:12,690 --> 00:20:16,169
as time goes by they

494
00:20:14,729 --> 00:20:17,609
spread out and you've got to gather them

495
00:20:16,169 --> 00:20:18,550
again after the transition and then they

496
00:20:17,609 --> 00:20:20,320
spread out and gather again.

497
00:20:18,550 --> 00:20:22,220
Hopefully everyone ends at the same

498
00:20:20,320 --> 00:20:30,679
point.

499
00:20:22,220 --> 00:20:30,679
[Music]