1
00:00:00,160 --> 00:00:02,960
Speaker 1: Hey everyone, it's Robert and Joe here. Today we've got

2
00:00:02,960 --> 00:00:05,240
Speaker 1: something a little bit different to share with you. It's

3
00:00:05,280 --> 00:00:09,400
Speaker 1: a new season of the Smart Talks with IBM podcast series.

4
00:00:09,720 --> 00:00:13,520
Speaker 2: This season on Smart Talks with IBM, Malcolm Gladwell is back,

5
00:00:13,520 --> 00:00:15,760
Speaker 2: and this time he's taking the show on the road.

6
00:00:15,880 --> 00:00:19,160
Speaker 2: Malcolm is stepping outside the studio to explore how IBM

7
00:00:19,239 --> 00:00:22,880
Speaker 2: clients are using artificial intelligence to solve real world challenges

8
00:00:23,079 --> 00:00:25,440
Speaker 2: and transform the way they do business.

9
00:00:25,480 --> 00:00:29,560
Speaker 1: From accelerating scientific breakthroughs to reimagining education. It's a fresh

10
00:00:29,600 --> 00:00:33,040
Speaker 1: look at innovation in action, where big ideas meet cutting

11
00:00:33,120 --> 00:00:33,960
Speaker 1: edge solutions.

12
00:00:34,320 --> 00:00:37,160
Speaker 2: You'll hear from industry leaders, creative thinkers, and of course

13
00:00:37,280 --> 00:00:40,920
Speaker 2: Malcolm Gladwell himself as he guides you through each story.

14
00:00:41,440 --> 00:00:44,680
Speaker 1: New episodes of Smart Talks with IBM drop every month

15
00:00:44,680 --> 00:00:48,279
Speaker 1: on the iHeartRadio app, Apple Podcasts, or wherever you get

16
00:00:48,320 --> 00:00:52,240
Speaker 1: your podcasts. Learn more at IBM dot com slash smart Talks.

17
00:00:52,880 --> 00:00:55,400
Speaker 1: This is a paid advertisement from IBM.

18
00:00:57,200 --> 00:01:02,920
Speaker 3: I'm Malcolm Gladwell and you're listening to Smart Talk with IBM.

19
00:01:03,320 --> 00:01:05,680
Speaker 3: Most of what happens in a UFC fight is too

20
00:01:05,680 --> 00:01:08,640
Speaker 3: fast to see. A fighter drops their shoulder for a

21
00:01:08,680 --> 00:01:12,039
Speaker 3: split second. Did you catch it? A shift in position

22
00:01:12,319 --> 00:01:17,920
Speaker 3: that looks insignificant but changes everything? Were you watching alone?

23
00:01:17,959 --> 00:01:20,040
Speaker 3: Cohen is the head of R and D for UFC.

24
00:01:20,400 --> 00:01:23,480
Speaker 3: His job is to help people see those moments, not

25
00:01:23,600 --> 00:01:26,880
Speaker 3: by slowing things down, but by knowing what to point

26
00:01:26,920 --> 00:01:30,200
Speaker 3: to after it happens. By the time you realize something mattered,

27
00:01:30,520 --> 00:01:34,880
Speaker 3: his data systems have already figured out why it's taken

28
00:01:34,959 --> 00:01:37,880
Speaker 3: fifteen years to get here, first with paper scorecards and

29
00:01:37,920 --> 00:01:42,080
Speaker 3: a TVO, now in partnership with IBM, and along the way,

30
00:01:42,480 --> 00:01:46,480
Speaker 3: a loan's learned something that applies far beyond fighting. The

31
00:01:46,520 --> 00:01:49,240
Speaker 3: best technology is the kind you don't.

32
00:01:49,040 --> 00:01:49,680
Speaker 4: Notice at all.

33
00:01:50,240 --> 00:01:54,760
Speaker 3: It just helps you see. I thought you were going

34
00:01:54,800 --> 00:02:00,840
Speaker 3: to be like tattoos muscle shirt. I thought you were

35
00:02:00,840 --> 00:02:02,960
Speaker 3: going to like I thought you were going to represent

36
00:02:03,000 --> 00:02:06,800
Speaker 3: the brand in that respect that you yourself would be.

37
00:02:07,000 --> 00:02:09,040
Speaker 3: It would be a kind of, you know, mixed martial

38
00:02:09,120 --> 00:02:11,360
Speaker 3: arts type. You were not, in fact a mixed martial

39
00:02:11,440 --> 00:02:11,960
Speaker 3: arts type.

40
00:02:12,919 --> 00:02:15,320
Speaker 5: I did a little taekwondo in my past. I am

41
00:02:15,639 --> 00:02:17,560
Speaker 5: I am not in the way that you put that,

42
00:02:17,720 --> 00:02:20,040
Speaker 5: not a mixed martial arts type. But I think if

43
00:02:20,120 --> 00:02:24,720
Speaker 5: you if you come through the building in Vegas. If

44
00:02:24,720 --> 00:02:26,400
Speaker 5: you come through the building when we do a show,

45
00:02:26,760 --> 00:02:28,760
Speaker 5: you're going to see a lot of the mixed martial

46
00:02:28,840 --> 00:02:31,520
Speaker 5: arts type. It even surprised me. I came out of

47
00:02:31,560 --> 00:02:34,040
Speaker 5: the tech world, the young tech world in the two thousands,

48
00:02:34,080 --> 00:02:36,120
Speaker 5: and everybody was talking about, you know, we have to

49
00:02:36,160 --> 00:02:38,520
Speaker 5: have a mix of viewpoints and a mix of you know,

50
00:02:38,560 --> 00:02:40,280
Speaker 5: people from all walks of life, and all of these

51
00:02:40,320 --> 00:02:44,560
Speaker 5: other things that evolved into into something else. But yeah,

52
00:02:44,600 --> 00:02:46,399
Speaker 5: that didn't happen to me until I went to work

53
00:02:46,440 --> 00:02:49,399
Speaker 5: for UFC, which is like a weird way to get there.

54
00:02:49,680 --> 00:02:51,839
Speaker 3: Were you in a tech world before? What were you doing?

55
00:02:51,960 --> 00:02:55,399
Speaker 5: So when I left school, I went to a startup

56
00:02:55,760 --> 00:02:58,600
Speaker 5: that today we would call a big data company, and

57
00:02:58,680 --> 00:03:00,800
Speaker 5: it was there at a moment where everybody was doing

58
00:03:00,880 --> 00:03:05,679
Speaker 5: relational databases and relational databases from micro Strategy and from

59
00:03:05,720 --> 00:03:08,880
Speaker 5: Cognos and from whoever. So we did that two thousand

60
00:03:08,919 --> 00:03:12,400
Speaker 5: and one hits after I graduate nine to eleven and

61
00:03:12,639 --> 00:03:14,840
Speaker 5: the dot com bust and all that kind of stuff,

62
00:03:14,880 --> 00:03:16,079
Speaker 5: and we all looked at each other and we said,

63
00:03:16,080 --> 00:03:18,800
Speaker 5: no one, no one of these institutions that needs this

64
00:03:18,919 --> 00:03:21,200
Speaker 5: kind of information is going to take a risk on

65
00:03:21,200 --> 00:03:23,240
Speaker 5: a small company right now, they're going to retrench.

66
00:03:23,639 --> 00:03:25,200
Speaker 4: And I went to law school.

67
00:03:26,200 --> 00:03:29,400
Speaker 3: Oh, I see, you started tech. You briefly have a

68
00:03:29,400 --> 00:03:31,480
Speaker 3: foot in the twenty first century, and then you decide, no,

69
00:03:31,520 --> 00:03:33,359
Speaker 3: I'm going to go get a law degree.

70
00:03:34,960 --> 00:03:35,280
Speaker 4: I did.

71
00:03:35,320 --> 00:03:38,840
Speaker 5: People said, you feel like somebody who would benefit from this,

72
00:03:39,440 --> 00:03:43,840
Speaker 5: and I wanted to want it. And the first attorney

73
00:03:43,880 --> 00:03:46,280
Speaker 5: I worked for out of law school, he looked at

74
00:03:46,320 --> 00:03:48,760
Speaker 5: me and he goes, you need to go be in business,

75
00:03:49,120 --> 00:03:50,400
Speaker 5: like you're not an attorney.

76
00:03:50,360 --> 00:03:54,280
Speaker 3: Yeah, so you're you're a failed lawyer. They kick you out.

77
00:03:55,600 --> 00:03:56,360
Speaker 3: Where do you go next?

78
00:03:56,880 --> 00:03:57,720
Speaker 4: So that's what happened.

79
00:03:57,760 --> 00:03:59,920
Speaker 5: So in two thousand and eight, there were a bunch

80
00:03:59,920 --> 00:04:01,960
Speaker 5: of lawyers who were doing just fine, but the bottom

81
00:04:02,000 --> 00:04:04,040
Speaker 5: has fallen out of the legal market. I was helping

82
00:04:04,080 --> 00:04:05,960
Speaker 5: a friend off the side of my desk who had

83
00:04:06,040 --> 00:04:08,640
Speaker 5: come to me with a banker's box of paper and said,

84
00:04:08,720 --> 00:04:11,240
Speaker 5: I have been watching The Ultimate Fighter, the reality show

85
00:04:11,280 --> 00:04:15,120
Speaker 5: where they pick new entrance to the UFC, and I

86
00:04:15,160 --> 00:04:17,920
Speaker 5: had been watching. He was a sports writer before, and

87
00:04:18,000 --> 00:04:21,760
Speaker 5: he very much. This gentleman's name is Romick an hour

88
00:04:21,880 --> 00:04:24,840
Speaker 5: romy comes to me and he says, there's no stats

89
00:04:24,839 --> 00:04:25,400
Speaker 5: for this stuff.

90
00:04:25,440 --> 00:04:26,800
Speaker 4: I'm used to writing about baseball.

91
00:04:26,920 --> 00:04:29,920
Speaker 5: So I did a full regression analysis after watching one

92
00:04:29,960 --> 00:04:32,520
Speaker 5: hundred hours of fight time, and now we're at closer

93
00:04:32,600 --> 00:04:34,039
Speaker 5: to two hundred hours of fight time. And I've been

94
00:04:34,040 --> 00:04:38,679
Speaker 5: scoring these fights, and the UFC's talent have been reading

95
00:04:38,720 --> 00:04:40,960
Speaker 5: my blog and they want me to do analysis for them,

96
00:04:40,960 --> 00:04:43,320
Speaker 5: and I can't be doing it on paper. And so

97
00:04:43,360 --> 00:04:45,120
Speaker 5: he came to a technologist and said, what do I do?

98
00:04:45,200 --> 00:04:47,280
Speaker 5: And we talked about double king the data in Indonesia

99
00:04:47,320 --> 00:04:50,919
Speaker 5: and building a database and APIs. A year later, we

100
00:04:50,920 --> 00:04:54,000
Speaker 5: were making small salaries and we had gotten this thing

101
00:04:54,279 --> 00:04:56,760
Speaker 5: up off the ground and going, and the UFC was

102
00:04:56,760 --> 00:04:58,560
Speaker 5: starting to use our statistics, and we had turned it

103
00:04:58,560 --> 00:05:00,920
Speaker 5: into a real data product. You know, we were able

104
00:05:00,920 --> 00:05:04,039
Speaker 5: to score electronically at that point. And the rest is

105
00:05:04,520 --> 00:05:06,320
Speaker 5: at this point seventeen or eighteen years of history.

106
00:05:06,360 --> 00:05:08,800
Speaker 3: So wait, back up, this is really interesting. Pretend I

107
00:05:08,839 --> 00:05:11,880
Speaker 3: know nothing about UFC. Okay, So Romi comes to you.

108
00:05:12,400 --> 00:05:15,640
Speaker 3: And Romy's issue is what that the way in which

109
00:05:16,240 --> 00:05:19,640
Speaker 3: these matches are scored is too subjective.

110
00:05:20,640 --> 00:05:24,640
Speaker 5: The way that these matches are scored is inscrutable to

111
00:05:24,760 --> 00:05:29,680
Speaker 5: the average person. So his analogy, and it's the one

112
00:05:29,680 --> 00:05:31,840
Speaker 5: that he experienced is when you write a summary of

113
00:05:31,920 --> 00:05:33,800
Speaker 5: last night's baseball game, right, what do you do?

114
00:05:33,880 --> 00:05:35,920
Speaker 4: You say, here's my thesis statement.

115
00:05:36,240 --> 00:05:39,320
Speaker 5: So and so had a great outing, and middle relief

116
00:05:39,320 --> 00:05:42,320
Speaker 5: collapsed or whatever it is, and last night at seven

117
00:05:42,360 --> 00:05:44,479
Speaker 5: pm at this stadium, the following happened. Here was the

118
00:05:44,480 --> 00:05:47,839
Speaker 5: final score. And then this middle paragraph is, you know,

119
00:05:47,920 --> 00:05:50,000
Speaker 5: middle relief had this as a cra and so on

120
00:05:50,040 --> 00:05:51,960
Speaker 5: and so forth, and you'd run an analysis and then

121
00:05:52,000 --> 00:05:53,159
Speaker 5: you talk about, you know, so and so has been

122
00:05:53,200 --> 00:05:54,880
Speaker 5: performing well and he had this outing and that outing

123
00:05:54,920 --> 00:05:55,839
Speaker 5: and blah blah blah blah.

124
00:05:55,880 --> 00:05:56,599
Speaker 4: This is part of a trend.

125
00:05:56,640 --> 00:05:59,159
Speaker 3: You tell the story of the game and the context

126
00:05:59,200 --> 00:06:02,479
Speaker 3: of a core of statistical.

127
00:06:01,920 --> 00:06:04,159
Speaker 5: Knowledge exactly right. And this is at least how he wrote.

128
00:06:04,160 --> 00:06:07,120
Speaker 5: And you get to that paragraph in an MMA story

129
00:06:07,600 --> 00:06:10,320
Speaker 5: and it's non existent. It's so and so is a

130
00:06:10,320 --> 00:06:12,720
Speaker 5: strong striker. So and so is coming off of two losses,

131
00:06:12,760 --> 00:06:14,360
Speaker 5: so and so is on a three fight win streak.

132
00:06:14,640 --> 00:06:18,039
Speaker 5: And that is the totality of what's currently available. And

133
00:06:19,080 --> 00:06:25,039
Speaker 5: Romy's approach is we can count everything, but everything's not important.

134
00:06:25,360 --> 00:06:28,520
Speaker 5: And you can't count everything practically because by the time

135
00:06:28,560 --> 00:06:30,960
Speaker 5: he gets to the end of this first iteration, we

136
00:06:31,000 --> 00:06:33,480
Speaker 5: have sixty seven data points per fighter per round.

137
00:06:33,640 --> 00:06:35,040
Speaker 4: So it's a ton of stuff.

138
00:06:35,160 --> 00:06:36,960
Speaker 3: At this point, you're collecting the data.

139
00:06:37,040 --> 00:06:40,320
Speaker 5: How just visually, he has a piece of paper and

140
00:06:40,480 --> 00:06:43,640
Speaker 5: he is on a TVO. Remember TV's. We had TV's

141
00:06:43,680 --> 00:06:45,480
Speaker 5: until long after they were dead because we needed them.

142
00:06:45,520 --> 00:06:48,480
Speaker 5: And he is playing and pausing and rewinding to be able.

143
00:06:48,240 --> 00:06:50,720
Speaker 3: To He's doing it the way people in the olders

144
00:06:50,720 --> 00:06:53,240
Speaker 3: in baseball, remember they would score. Sure, they would take

145
00:06:53,279 --> 00:06:54,720
Speaker 3: the piece of paper and they would score the game

146
00:06:54,760 --> 00:06:56,920
Speaker 3: as they were watching. It's flowing to matches.

147
00:06:57,320 --> 00:07:00,680
Speaker 5: His history is I go to baseball games with dad.

148
00:07:01,040 --> 00:07:03,800
Speaker 5: Dad has a scorecard, and you, as a kid, are

149
00:07:03,800 --> 00:07:05,920
Speaker 5: both drawn to it and averse to it because you're

150
00:07:05,960 --> 00:07:07,479
Speaker 5: like dad, you're missing the game, and at some point

151
00:07:07,520 --> 00:07:09,440
Speaker 5: you find yourself marking down the scorecard.

152
00:07:09,480 --> 00:07:11,560
Speaker 4: And he brings a lot of that to this.

153
00:07:12,240 --> 00:07:15,760
Speaker 3: So you say in his first iteration he had seventy

154
00:07:15,800 --> 00:07:18,640
Speaker 3: two data points, say that again, sixty seven sixty seven

155
00:07:18,720 --> 00:07:23,040
Speaker 3: data points per fighter fighter per round per round.

156
00:07:23,240 --> 00:07:24,400
Speaker 4: Yeah, so that's.

157
00:07:24,240 --> 00:07:28,400
Speaker 5: Consistently there or this as an average of Well, so

158
00:07:28,520 --> 00:07:31,120
Speaker 5: at any points on the sheet, there are sixty seven

159
00:07:31,160 --> 00:07:33,760
Speaker 5: different things that we are tracking. So there may be

160
00:07:33,800 --> 00:07:37,280
Speaker 5: more head strikes landed, powerhead strikes landed, and there may

161
00:07:37,280 --> 00:07:40,000
Speaker 5: be some other head strikes landed, but there's probably not

162
00:07:40,080 --> 00:07:42,800
Speaker 5: necessary tight submission or a camorra or something like that.

163
00:07:42,840 --> 00:07:45,200
Speaker 5: But these are different kinds of submission types that we

164
00:07:45,200 --> 00:07:47,560
Speaker 5: would track, and so all of them are available on

165
00:07:47,600 --> 00:07:49,080
Speaker 5: the sheet so that you can just tick those off

166
00:07:49,120 --> 00:07:51,680
Speaker 5: and move on. Right, So those are the sixty seven

167
00:07:51,680 --> 00:07:54,800
Speaker 5: independent boxes on the sheet that we could mark something

168
00:07:54,840 --> 00:07:55,160
Speaker 5: down on.

169
00:07:55,320 --> 00:07:58,600
Speaker 3: Give me examples of things that non obvious things that

170
00:07:58,640 --> 00:08:01,880
Speaker 3: you would learn if you did a formal statistical analysis

171
00:08:01,960 --> 00:08:06,240
Speaker 3: of a match, like as a viewer or as a fan.

172
00:08:06,400 --> 00:08:08,680
Speaker 3: How is this enhancing his experience of what he's watching.

173
00:08:08,720 --> 00:08:10,680
Speaker 5: A couple of things are happening inside of an MMA

174
00:08:10,800 --> 00:08:15,440
Speaker 5: match that are worthy of note to the average fan.

175
00:08:16,240 --> 00:08:18,640
Speaker 5: It is scored per round. And one of the problems

176
00:08:18,640 --> 00:08:20,200
Speaker 5: that you have when you score per round is that

177
00:08:20,280 --> 00:08:23,600
Speaker 5: people don't think per round right. They don't get to

178
00:08:23,640 --> 00:08:26,120
Speaker 5: the end of round one and say round one is over.

179
00:08:26,240 --> 00:08:28,240
Speaker 5: I will make a decision about round one, let us

180
00:08:28,320 --> 00:08:29,920
Speaker 5: now go to They don't do that. They look at

181
00:08:29,920 --> 00:08:32,560
Speaker 5: the totality of the match. The shorthand I used for

182
00:08:32,600 --> 00:08:34,600
Speaker 5: this for a long time is you make the invisible

183
00:08:34,720 --> 00:08:35,880
Speaker 5: visible to people like.

184
00:08:36,120 --> 00:08:37,320
Speaker 4: What exactly did I see?

185
00:08:37,520 --> 00:08:39,320
Speaker 5: The other thing that the last thing that the numbers

186
00:08:39,320 --> 00:08:44,199
Speaker 5: do for us is they tend to combat recency because

187
00:08:44,559 --> 00:08:46,720
Speaker 5: you just forget what happened in round one or it

188
00:08:46,800 --> 00:08:48,599
Speaker 5: just doesn't leave as strong an impression. And one of

189
00:08:48,640 --> 00:08:51,120
Speaker 5: the and of course fighters fight to that what do

190
00:08:51,160 --> 00:08:51,880
Speaker 5: you do right at.

191
00:08:51,840 --> 00:08:52,480
Speaker 4: The end of the round?

192
00:08:52,600 --> 00:08:55,199
Speaker 5: You take a dude and you take him down because

193
00:08:55,200 --> 00:08:56,720
Speaker 5: you were going to finish the fight now because you

194
00:08:56,760 --> 00:08:58,600
Speaker 5: wanted to leave in the impression and the impression in

195
00:08:58,640 --> 00:09:01,040
Speaker 5: the mind of the judges like that do you controlled

196
00:09:01,040 --> 00:09:03,760
Speaker 5: that round? And so you were trying to score these

197
00:09:03,800 --> 00:09:05,559
Speaker 5: big shots at the end of a round, and the

198
00:09:05,640 --> 00:09:08,320
Speaker 5: numbers allow us to battle some of that for our

199
00:09:08,360 --> 00:09:10,520
Speaker 5: fans to say, like, how well did he actually do?

200
00:09:10,800 --> 00:09:12,560
Speaker 4: What did she actually do in this thing?

201
00:09:12,880 --> 00:09:15,800
Speaker 5: So that's one of the big learnings is you're able

202
00:09:15,800 --> 00:09:17,320
Speaker 5: to go back in time and not have to watch

203
00:09:17,360 --> 00:09:20,200
Speaker 5: the video. You just get a spot to be like, oh, yeah,

204
00:09:20,360 --> 00:09:22,160
Speaker 5: there are two takedowns in that first round.

205
00:09:21,960 --> 00:09:25,840
Speaker 3: And so you're allowing people to construct a much more

206
00:09:25,920 --> 00:09:28,720
Speaker 3: complete and accurate narrative of the of the match.

207
00:09:28,920 --> 00:09:31,680
Speaker 5: I think so yeah, And I think for new fans

208
00:09:31,760 --> 00:09:35,560
Speaker 5: there's a different calculus because new fans don't know how

209
00:09:35,600 --> 00:09:38,720
Speaker 5: to watch the sport. And when you it's I joke

210
00:09:38,800 --> 00:09:40,840
Speaker 5: that when we all watch the Olympics, like, I don't

211
00:09:40,880 --> 00:09:43,760
Speaker 5: know anything about synchronized swimming and I shouldn't pretend to,

212
00:09:44,120 --> 00:09:48,360
Speaker 5: but the Olympic broadcast allows me to pretend to, right,

213
00:09:49,120 --> 00:09:51,360
Speaker 5: And that's a little bit of what's happening here is

214
00:09:51,400 --> 00:09:53,080
Speaker 5: that they can look at this and say, oh, I

215
00:09:53,160 --> 00:09:56,040
Speaker 5: get it right, Like I'm watching all this action, and

216
00:09:56,080 --> 00:09:59,480
Speaker 5: then we condense that down to this is the number

217
00:09:59,520 --> 00:10:01,600
Speaker 5: of like kicks this person landed, and they're like, oh,

218
00:10:01,600 --> 00:10:03,360
Speaker 5: I did see Okay, Now I know to focus on that.

219
00:10:03,400 --> 00:10:06,120
Speaker 5: This is part of the storyline and it is a

220
00:10:06,200 --> 00:10:08,080
Speaker 5: signpost and that's a big deal.

221
00:10:08,160 --> 00:10:11,080
Speaker 3: So in the beginning, you're just a contractor with UFC.

222
00:10:11,600 --> 00:10:13,440
Speaker 4: We are a contractor entity.

223
00:10:13,520 --> 00:10:13,720
Speaker 2: Yeah.

224
00:10:13,720 --> 00:10:16,360
Speaker 5: For the first seven years of our existence, we are

225
00:10:16,360 --> 00:10:19,280
Speaker 5: independent and are the official data feed of the UFC.

226
00:10:19,480 --> 00:10:22,880
Speaker 3: And and at a certain point UFC says, come and

227
00:10:22,960 --> 00:10:23,400
Speaker 3: join the.

228
00:10:25,280 --> 00:10:29,520
Speaker 5: Ari emmanual at wm IMG at the time buys UFC

229
00:10:30,080 --> 00:10:31,559
Speaker 5: and he says, I want to sell the data.

230
00:10:31,600 --> 00:10:32,480
Speaker 4: Where's the data?

231
00:10:32,559 --> 00:10:34,360
Speaker 5: And they said, well, these two nice kids in the

232
00:10:34,360 --> 00:10:36,199
Speaker 5: Georgetown waterfront they have the data, and he goes, what

233
00:10:36,480 --> 00:10:38,640
Speaker 5: do you mean they have the data? Like, it's our data,

234
00:10:38,920 --> 00:10:41,640
Speaker 5: and that started an acquisition. Conver We were a fold

235
00:10:41,800 --> 00:10:43,360
Speaker 5: into that larger acquisition.

236
00:10:43,480 --> 00:10:45,320
Speaker 3: Going back to the fights, the data, how is it?

237
00:10:45,480 --> 00:10:48,079
Speaker 3: How is the scoring working at this point?

238
00:10:48,600 --> 00:10:50,520
Speaker 4: So the tvo I brought up earlier, you're able to

239
00:10:50,520 --> 00:10:53,680
Speaker 4: watch it with slow moo and rewind and we have

240
00:10:53,920 --> 00:10:56,679
Speaker 4: at the time tablet or a touch screen, and we

241
00:10:56,720 --> 00:10:58,920
Speaker 4: have taken our piece of paper and we have split

242
00:10:58,960 --> 00:11:01,200
Speaker 4: it in twain and you're able to mark what you

243
00:11:01,240 --> 00:11:03,480
Speaker 4: see as you see it, save around you score.

244
00:11:03,520 --> 00:11:06,479
Speaker 3: You've got trained scores, yes, yeah.

245
00:11:06,120 --> 00:11:08,120
Speaker 5: They are some of my favorite people. We have had

246
00:11:08,160 --> 00:11:11,160
Speaker 5: trained scorers in our group that go back thirteen years

247
00:11:11,200 --> 00:11:13,760
Speaker 5: with us, folks that we plucked out of the ether. Right,

248
00:11:13,800 --> 00:11:16,200
Speaker 5: you put something online looking for people who are interested

249
00:11:16,200 --> 00:11:18,280
Speaker 5: in learning this stuff, and then you have to whittle

250
00:11:18,280 --> 00:11:20,360
Speaker 5: out the people who can learn how to do it

251
00:11:20,400 --> 00:11:22,200
Speaker 5: and then do it quickly. The first time you score

252
00:11:22,240 --> 00:11:24,040
Speaker 5: something like this and try to make sense of the

253
00:11:25,040 --> 00:11:28,600
Speaker 5: scoring page, it's going to take you fifteen minutes to do.

254
00:11:28,800 --> 00:11:30,840
Speaker 5: A five minute round might take you more. And that

255
00:11:30,880 --> 00:11:32,199
Speaker 5: means that at the end of the fight, you're probably

256
00:11:32,200 --> 00:11:34,600
Speaker 5: thirty five minutes late. And so you have to do

257
00:11:34,640 --> 00:11:37,240
Speaker 5: this for six or seven months on a regular basis,

258
00:11:37,320 --> 00:11:39,640
Speaker 5: paying you make sure that you know you're fairly compensated

259
00:11:40,000 --> 00:11:41,959
Speaker 5: to get you to the point where you're able to

260
00:11:42,000 --> 00:11:45,000
Speaker 5: make decisions rapidly. It just comes as second nature to you.

261
00:11:45,480 --> 00:11:48,720
Speaker 5: And this is the dirty little secret of UFC. But

262
00:11:49,320 --> 00:11:51,120
Speaker 5: we joke, we've been trying to put ourselves out of

263
00:11:51,120 --> 00:11:55,560
Speaker 5: that business for at this point eleven or twelve years

264
00:11:56,000 --> 00:11:57,720
Speaker 5: and we aren't close yet.

265
00:11:57,960 --> 00:11:59,920
Speaker 3: Put yourself out of the business of having to rely

266
00:12:00,080 --> 00:12:02,280
Speaker 3: eye on a human score, a human score.

267
00:12:02,520 --> 00:12:06,440
Speaker 5: Yeah, and we said that joshingly. But there's a sense

268
00:12:06,480 --> 00:12:08,680
Speaker 5: of which there's a sense in which at some point

269
00:12:08,760 --> 00:12:12,280
Speaker 5: you understand that compute ought to catch up with you, right,

270
00:12:12,840 --> 00:12:16,080
Speaker 5: And we said we should be out ahead of this

271
00:12:16,720 --> 00:12:18,640
Speaker 5: and sort of the Netflix model, Right, if DVDs are

272
00:12:18,679 --> 00:12:20,439
Speaker 5: going to be a problem, you shouldigure out what's coming next,

273
00:12:20,480 --> 00:12:23,040
Speaker 5: and you should do that also and disrupt yourself. And

274
00:12:24,040 --> 00:12:26,640
Speaker 5: you know, we wanted initially we have used it as

275
00:12:26,720 --> 00:12:29,600
Speaker 5: additive data, so we wanted to do motion tracking. So

276
00:12:30,000 --> 00:12:32,520
Speaker 5: eleven twelve years ago we bought a motion tracking software

277
00:12:32,920 --> 00:12:34,560
Speaker 5: and it couldn't make sense of any of what it

278
00:12:34,600 --> 00:12:37,240
Speaker 5: was seeing inside the octagon, like it really couldn't understand.

279
00:12:37,280 --> 00:12:39,200
Speaker 5: And this wasn't even striking. This is just like red

280
00:12:39,280 --> 00:12:40,800
Speaker 5: is here, blue is here, and the rough is here.

281
00:12:41,400 --> 00:12:44,400
Speaker 5: And so we taught a human system how to do it.

282
00:12:44,440 --> 00:12:47,720
Speaker 5: But we keep going back to the to the machine well,

283
00:12:48,120 --> 00:12:50,280
Speaker 5: hoping it will get better. And today, in fact we

284
00:12:50,320 --> 00:12:52,880
Speaker 5: are we are closer than ever, but nowhere near close

285
00:12:52,960 --> 00:12:55,720
Speaker 5: to overtaking us as scores.

286
00:12:55,880 --> 00:12:56,679
Speaker 4: It's weird.

287
00:12:56,920 --> 00:12:59,280
Speaker 5: I have been playing, as I said, with my motion

288
00:12:59,640 --> 00:13:04,040
Speaker 5: tracking computer vision experiments. We're twelve years ago. Ten years

289
00:13:04,040 --> 00:13:07,959
Speaker 5: ago I was hanging cameras to attempt to do a

290
00:13:08,080 --> 00:13:10,319
Speaker 5: skeletal tracking to see what that could give us, and

291
00:13:10,360 --> 00:13:13,480
Speaker 5: it didn't work. You know, we have been on a

292
00:13:13,520 --> 00:13:16,880
Speaker 5: ten year journey to discover what it can do and

293
00:13:16,920 --> 00:13:20,640
Speaker 5: being fostering of it. Right, That's what R and D

294
00:13:20,720 --> 00:13:23,080
Speaker 5: is supposed to say when something fails, to say, well

295
00:13:23,160 --> 00:13:25,959
Speaker 5: is there something of this that I can retain? Does

296
00:13:26,000 --> 00:13:28,360
Speaker 5: this add some value enough that it really makes sense

297
00:13:28,400 --> 00:13:31,000
Speaker 5: to just continue to invest in it some so if

298
00:13:31,080 --> 00:13:34,679
Speaker 5: it can ever make good on its promises, we'll get

299
00:13:34,679 --> 00:13:35,920
Speaker 5: to see exponentially.

300
00:13:36,000 --> 00:13:38,640
Speaker 3: What striving you here is what you want to get

301
00:13:38,679 --> 00:13:42,160
Speaker 3: your data quicker? Or do you want more data? Do

302
00:13:42,200 --> 00:13:43,880
Speaker 3: you want qualitatively different data?

303
00:13:43,920 --> 00:13:44,320
Speaker 2: What is it?

304
00:13:44,400 --> 00:13:46,640
Speaker 3: What's what's what's in the back of your head?

305
00:13:46,880 --> 00:13:51,600
Speaker 5: So man, so many motivations are pushing this for us.

306
00:13:52,280 --> 00:13:55,960
Speaker 5: One motivation is faster data. So imagine the following very simple.

307
00:13:56,600 --> 00:13:59,559
Speaker 5: If I see a significant strike landed in the middle

308
00:13:59,600 --> 00:14:01,640
Speaker 5: and so well, let's say the first person to one

309
00:14:01,679 --> 00:14:05,000
Speaker 5: hundred significant strikes landed in this fight, that is potentially

310
00:14:05,040 --> 00:14:07,760
Speaker 5: betting data. Somebody could bet who could get to that first.

311
00:14:07,960 --> 00:14:10,240
Speaker 5: Right now, I can only offer that data unofficially because

312
00:14:10,240 --> 00:14:13,360
Speaker 5: my humans need to wait for my official data to

313
00:14:13,360 --> 00:14:16,080
Speaker 5: come in. If computer vision could really get that right,

314
00:14:16,679 --> 00:14:18,800
Speaker 5: then I could stand behind that number and say that

315
00:14:19,200 --> 00:14:22,240
Speaker 5: is betting ready, and you could resolve that bet at

316
00:14:22,320 --> 00:14:24,040
Speaker 5: the moment it happens. And so you open up all

317
00:14:24,040 --> 00:14:28,040
Speaker 5: of these markets for one like that. That's a big

318
00:14:28,080 --> 00:14:32,240
Speaker 5: motivator for this kind of stuff. The second is I

319
00:14:32,320 --> 00:14:36,240
Speaker 5: am When I'm adding these AI technologies, I'm often looking

320
00:14:36,240 --> 00:14:37,800
Speaker 5: at them and saying, like, what else can I do

321
00:14:37,880 --> 00:14:39,400
Speaker 5: with this? What else is it telling me about the

322
00:14:39,400 --> 00:14:41,520
Speaker 5: fight that I'm not already getting motion was a big

323
00:14:41,560 --> 00:14:44,640
Speaker 5: one of them. Right, where are people in the octagon?

324
00:14:45,760 --> 00:14:49,200
Speaker 5: What I mentioned earlier that one of the one of

325
00:14:49,240 --> 00:14:51,680
Speaker 5: the criteria for judging a fight is control of the

326
00:14:51,680 --> 00:14:54,840
Speaker 5: octagon space. Well, okay, everything I was tracking with my

327
00:14:54,880 --> 00:14:56,520
Speaker 5: scores before doesn't talk about that.

328
00:14:56,640 --> 00:14:56,760
Speaker 4: Right.

329
00:14:56,760 --> 00:14:59,080
Speaker 5: If I'm hitting you more and you're you know, defending

330
00:14:59,120 --> 00:15:02,960
Speaker 5: poorly or whatever it is, maybe that's control. But what

331
00:15:03,000 --> 00:15:05,920
Speaker 5: if you're a counterpuncher, right, famous fighters who are counterpunchers.

332
00:15:05,960 --> 00:15:08,160
Speaker 5: I draw you in and then I hit you. Does

333
00:15:08,200 --> 00:15:09,840
Speaker 5: that mean I'm seating control to be able to take

334
00:15:09,880 --> 00:15:10,320
Speaker 5: back control?

335
00:15:10,360 --> 00:15:10,960
Speaker 4: What does that mean?

336
00:15:11,400 --> 00:15:13,880
Speaker 5: And so we didn't have any any numbers on where

337
00:15:13,880 --> 00:15:16,440
Speaker 5: people are. That's an immediate use of AI that is

338
00:15:16,840 --> 00:15:19,840
Speaker 5: additive to what I am doing. And all of this

339
00:15:19,920 --> 00:15:24,200
Speaker 5: is always in the service of storyline. Right, what is

340
00:15:24,280 --> 00:15:28,320
Speaker 5: not being captured in the fight? And what should the

341
00:15:28,360 --> 00:15:30,640
Speaker 5: commentators be talking about? And how can I support what

342
00:15:30,680 --> 00:15:32,760
Speaker 5: they're talking about? How can I give them something that

343
00:15:32,800 --> 00:15:34,840
Speaker 5: makes them think, Oh, actually, the thing I want to

344
00:15:34,880 --> 00:15:38,560
Speaker 5: be talking about here is this, Right, that's ultimately the

345
00:15:38,600 --> 00:15:41,680
Speaker 5: goal the goal here is I mean, let's go back

346
00:15:41,720 --> 00:15:43,200
Speaker 5: to the first thing you said. You said, I don't

347
00:15:43,240 --> 00:15:44,680
Speaker 5: look like the kind of person who might be an

348
00:15:44,760 --> 00:15:48,680
Speaker 5: MMA fan. There is a point at which if I

349
00:15:48,720 --> 00:15:50,760
Speaker 5: throw up a number on the screen and it's like,

350
00:15:50,800 --> 00:15:53,120
Speaker 5: this person is twenty seven point two four, and people

351
00:15:53,120 --> 00:15:55,520
Speaker 5: are like nerd, like, what are you doing? And there

352
00:15:56,000 --> 00:15:58,920
Speaker 5: is a space for that in other sports, but sometimes

353
00:15:58,960 --> 00:16:01,680
Speaker 5: we're doing it just to seem to technical. I think

354
00:16:01,680 --> 00:16:04,160
Speaker 5: if you're doing the tech right, you should forget about

355
00:16:04,160 --> 00:16:06,080
Speaker 5: the tech. You should forget about the fact that there's

356
00:16:06,080 --> 00:16:07,960
Speaker 5: AI behind that. You should forget about all of this.

357
00:16:08,280 --> 00:16:11,560
Speaker 5: The thing that hits the screen is so and so

358
00:16:11,600 --> 00:16:16,760
Speaker 5: has never lost a fight if they have landed four takedowns.

359
00:16:17,680 --> 00:16:21,320
Speaker 5: We're on number three. Right now, we know what we're watching. Yeah,

360
00:16:21,400 --> 00:16:25,080
Speaker 5: right now you have a binary, simple thing to look at.

361
00:16:25,240 --> 00:16:28,800
Speaker 5: So and so has a takedown accuracy of ninety four percent.

362
00:16:28,960 --> 00:16:32,120
Speaker 5: So and so has a takedown defense of seventy five percent.

363
00:16:32,400 --> 00:16:34,600
Speaker 5: Let's see how they match up against their classic averages.

364
00:16:34,840 --> 00:16:36,400
Speaker 5: And now you're getting into some more tech and number

365
00:16:36,440 --> 00:16:38,120
Speaker 5: you're like, this guy is below his average and that

366
00:16:38,120 --> 00:16:39,920
Speaker 5: guy's above his average. Okay, now we have a way

367
00:16:39,960 --> 00:16:43,400
Speaker 5: of measuring very quickly who's winning and who's losing. Across

368
00:16:43,680 --> 00:16:45,600
Speaker 5: at least this part of the fight, and so the

369
00:16:45,640 --> 00:16:48,000
Speaker 5: goal is to get to storyline as fast as possible

370
00:16:49,080 --> 00:16:52,480
Speaker 5: in human understandable terms. Right, This is in some ways

371
00:16:52,520 --> 00:16:56,040
Speaker 5: the great moment of the LM as well. Right, Who

372
00:16:56,120 --> 00:16:59,840
Speaker 5: cares how many weights and how many parameters and how

373
00:16:59,840 --> 00:17:02,320
Speaker 5: many whatever? When I talk to it, it feels like

374
00:17:02,400 --> 00:17:05,200
Speaker 5: I'm talking to a human. That's the great unlock. The

375
00:17:05,640 --> 00:17:10,639
Speaker 5: technology suddenly disappears and the experience stays. And that's at

376
00:17:11,119 --> 00:17:14,520
Speaker 5: our level, at our level of sophistication, that's what we're doing.

377
00:17:14,600 --> 00:17:17,000
Speaker 3: Yeah, I want to go back to make sure I

378
00:17:17,080 --> 00:17:22,840
Speaker 3: understand when you talk about control of the octagon of

379
00:17:22,840 --> 00:17:26,239
Speaker 3: the octagon conceptually, what is control the octagon mean?

380
00:17:26,560 --> 00:17:29,879
Speaker 5: It is the last of the criteria in the judging rules,

381
00:17:30,080 --> 00:17:35,919
Speaker 5: and it should mean that in a fight one of

382
00:17:35,920 --> 00:17:39,159
Speaker 5: the things you're trying to do is assert dominance the

383
00:17:39,160 --> 00:17:40,040
Speaker 5: most basic level.

384
00:17:40,400 --> 00:17:41,439
Speaker 4: It is left vague.

385
00:17:41,960 --> 00:17:44,720
Speaker 5: Different commissions around the country will tell you slightly different

386
00:17:44,760 --> 00:17:46,960
Speaker 5: things about what it means. Some of them will boil

387
00:17:47,000 --> 00:17:49,760
Speaker 5: it down to if I hit you more, I control

388
00:17:49,800 --> 00:17:51,760
Speaker 5: the fight, but a lot of others of them will

389
00:17:51,760 --> 00:17:55,359
Speaker 5: say it is really there to be that sixth sense

390
00:17:55,760 --> 00:17:59,199
Speaker 5: feel of was I able to put you where I

391
00:17:59,320 --> 00:18:02,160
Speaker 5: wanted you to be when I want you to be there,

392
00:18:02,760 --> 00:18:06,280
Speaker 5: and in our statistics, one of the ways you can

393
00:18:06,280 --> 00:18:10,040
Speaker 5: talk about that is advancing or center control. If I

394
00:18:10,080 --> 00:18:12,399
Speaker 5: always own the center, you always have the wall at

395
00:18:12,440 --> 00:18:14,720
Speaker 5: your back, so I control the space.

396
00:18:15,119 --> 00:18:15,280
Speaker 4: Right.

397
00:18:15,280 --> 00:18:23,159
Speaker 5: There are ways of finding small analogus directional pointers to

398
00:18:23,200 --> 00:18:25,280
Speaker 5: tell you that. But when you talk about what a

399
00:18:25,400 --> 00:18:28,119
Speaker 5: judge is looking for, they're looking for that ephemeral sense

400
00:18:28,760 --> 00:18:31,280
Speaker 5: of you know, who's the big.

401
00:18:31,160 --> 00:18:35,160
Speaker 3: Dog, and that is if you're strictly kind of scoring

402
00:18:35,160 --> 00:18:38,080
Speaker 3: the match, that's one thing you're not picking up because

403
00:18:38,359 --> 00:18:43,120
Speaker 3: you're too focused on position, who's striking whom that kind

404
00:18:43,119 --> 00:18:47,440
Speaker 3: of larger gestal you're talking about a gestalt of right, Sure, yeah.

405
00:18:47,040 --> 00:18:50,400
Speaker 5: Somehow the Novata State Athletic Commission to get to criteria

406
00:18:50,440 --> 00:18:52,000
Speaker 5: five and go gestalt.

407
00:18:52,160 --> 00:18:53,160
Speaker 4: That's what we're looking for.

408
00:18:53,880 --> 00:18:56,199
Speaker 5: We're trying to deal as a as a as a

409
00:18:56,240 --> 00:18:58,480
Speaker 5: matter of course, we want to bring people into our arena.

410
00:18:59,320 --> 00:19:02,560
Speaker 5: Have the six it's be devoid of Weltschmertz and focus

411
00:19:02,600 --> 00:19:03,320
Speaker 5: on the gestalt.

412
00:19:03,440 --> 00:19:05,199
Speaker 4: And it just didn't. It didn't make it, That's what

413
00:19:05,200 --> 00:19:05,680
Speaker 4: I'm saying.

414
00:19:06,680 --> 00:19:11,119
Speaker 3: So you've got your motivations are one I can enhance

415
00:19:11,160 --> 00:19:14,000
Speaker 3: the viewer experience in terms of story. Yep, I can.

416
00:19:14,320 --> 00:19:19,040
Speaker 3: I can facilitate things like betting on the on the match,

417
00:19:19,520 --> 00:19:23,040
Speaker 3: and I can aid the judges in getting a sense

418
00:19:23,119 --> 00:19:24,960
Speaker 3: of who the dominant party is.

419
00:19:25,080 --> 00:19:25,479
Speaker 4: One and two.

420
00:19:25,560 --> 00:19:30,320
Speaker 5: Yes, three, we actually scrub all of our all of

421
00:19:30,359 --> 00:19:33,800
Speaker 5: our video feeds into the arena, a scrubbed of our statistics.

422
00:19:34,359 --> 00:19:37,199
Speaker 5: I do not want the judges. Their job is to

423
00:19:37,520 --> 00:19:39,960
Speaker 5: follow the criteria set by their commission. So we're in

424
00:19:40,000 --> 00:19:44,199
Speaker 5: this weird spot where like my sport has, my fighters

425
00:19:44,200 --> 00:19:47,800
Speaker 5: are independent contractors. It's a regulated sport by the state

426
00:19:47,800 --> 00:19:50,399
Speaker 5: commission that we're dealing with, and my job is to

427
00:19:50,600 --> 00:19:53,240
Speaker 5: is to bring the event together. And as a result,

428
00:19:53,359 --> 00:19:56,399
Speaker 5: I don't want anyone pointing at us saying you said.

429
00:19:56,680 --> 00:19:59,640
Speaker 5: And so I decided we're like, look, you understand your criteria,

430
00:19:59,760 --> 00:20:02,280
Speaker 5: you're fully trained, judge. You need to decide who won

431
00:20:02,280 --> 00:20:05,440
Speaker 5: this fight. Our statistics will tell the story our statistics

432
00:20:05,480 --> 00:20:07,320
Speaker 5: tell and by the way, we can have an occasionally

433
00:20:07,320 --> 00:20:10,960
Speaker 5: antagonistic relationship where we thought that was a poor decision, right.

434
00:20:11,040 --> 00:20:13,520
Speaker 4: Yeah, that makes it interesting exactly. Yeah.

435
00:20:13,600 --> 00:20:15,760
Speaker 3: So okay, let's go back to the AI. So you're

436
00:20:15,960 --> 00:20:19,440
Speaker 3: you start experimenting with this a while ago. Oh yeah,

437
00:20:19,480 --> 00:20:23,840
Speaker 3: and you're not you're not entirely happy with the with the results.

438
00:20:24,080 --> 00:20:27,040
Speaker 3: Is that a fair or you're Oh, you understand it's embryotic.

439
00:20:28,320 --> 00:20:29,000
Speaker 4: Both, right?

440
00:20:29,600 --> 00:20:32,720
Speaker 5: You? Why should I start down the project of making

441
00:20:32,760 --> 00:20:35,080
Speaker 5: this work if I don't believe it's going to work?

442
00:20:35,320 --> 00:20:37,040
Speaker 4: Right, I try?

443
00:20:37,160 --> 00:20:38,960
Speaker 5: If I'm going to bother dedicating the time and the

444
00:20:38,960 --> 00:20:42,359
Speaker 5: effort and potentially a lot of money. I wanted to work. Uh,

445
00:20:42,520 --> 00:20:44,760
Speaker 5: and I was disappointed in time and again, and I

446
00:20:44,800 --> 00:20:48,880
Speaker 5: have found things to bring along right, So early skeletal

447
00:20:48,920 --> 00:20:51,000
Speaker 5: tracking has become the basis for today.

448
00:20:50,880 --> 00:20:53,879
Speaker 3: SAT tracking is defined.

449
00:20:53,920 --> 00:20:57,679
Speaker 5: The term skeletal tracking is when you hang cameras. Nowadays

450
00:20:57,960 --> 00:21:00,240
Speaker 5: you are very quickly able to discern what is is

451
00:21:00,280 --> 00:21:02,480
Speaker 5: not a human inside of your space, and then you're

452
00:21:02,480 --> 00:21:06,680
Speaker 5: able to identify points on their body, nowadays more points

453
00:21:06,680 --> 00:21:07,119
Speaker 5: than we used to.

454
00:21:07,119 --> 00:21:07,760
Speaker 4: It used to be like.

455
00:21:08,080 --> 00:21:13,040
Speaker 5: Shoulders, head, torso, hips, knees, feet, right, and nowadays you

456
00:21:13,080 --> 00:21:14,760
Speaker 5: have two additional points on the feet so you can

457
00:21:14,760 --> 00:21:17,760
Speaker 5: see their orientation. Hands you can see the orientation, and

458
00:21:17,800 --> 00:21:21,000
Speaker 5: face you can see where I'm looking. Those are basic

459
00:21:21,040 --> 00:21:24,639
Speaker 5: skeletal models and you need that to be able to

460
00:21:24,720 --> 00:21:30,000
Speaker 5: understand from an AI perspective, who is striking and was

461
00:21:30,040 --> 00:21:30,560
Speaker 5: that a miss?

462
00:21:30,680 --> 00:21:31,200
Speaker 4: Was that a hit?

463
00:21:31,720 --> 00:21:34,119
Speaker 5: We can get into collision detection, which is very difficult

464
00:21:34,119 --> 00:21:37,520
Speaker 5: because skeletal tracking isn't it's not the outside of your arm,

465
00:21:37,560 --> 00:21:41,520
Speaker 5: it's the inside. And then beyond was there a collision,

466
00:21:41,560 --> 00:21:44,520
Speaker 5: we get into problems of intent. So intent is the

467
00:21:44,520 --> 00:21:46,640
Speaker 5: hardest thing actually to solve. There's a lot of times

468
00:21:46,640 --> 00:21:48,320
Speaker 5: where two guys are fighting and one guy's going like this,

469
00:21:48,560 --> 00:21:51,119
Speaker 5: he's range finding, just putting a hand out there to

470
00:21:51,200 --> 00:21:54,080
Speaker 5: see is you know, how far away are you and

471
00:21:54,119 --> 00:21:58,280
Speaker 5: sort of to bat you away? And that is not

472
00:21:58,440 --> 00:22:00,879
Speaker 5: a strike attempt. How do you teach the computer that

473
00:22:00,880 --> 00:22:03,159
Speaker 5: that's not a strike attempt? Because sometimes it looks like it,

474
00:22:03,359 --> 00:22:06,840
Speaker 5: sometimes it doesn't. And then same thing with kick to

475
00:22:06,880 --> 00:22:10,040
Speaker 5: the head. It makes contact with the outset of my arm,

476
00:22:10,080 --> 00:22:14,639
Speaker 5: which I brought up in our world, that's a blocked strike.

477
00:22:15,960 --> 00:22:19,119
Speaker 5: But teaching a computer what exactly that means and when

478
00:22:19,240 --> 00:22:21,639
Speaker 5: and how like when it's up here, when my arm

479
00:22:21,720 --> 00:22:23,560
Speaker 5: is up, that's a block. When my arm is down

480
00:22:23,640 --> 00:22:26,919
Speaker 5: and hits my shoulder, that's not. It's those nuances that

481
00:22:27,000 --> 00:22:30,320
Speaker 5: proved incredibly difficult for machines to be able to handle

482
00:22:30,359 --> 00:22:34,080
Speaker 5: for a very very long time. And the acceleration in

483
00:22:34,119 --> 00:22:36,160
Speaker 5: the last few years which has allowed us to onboard

484
00:22:36,160 --> 00:22:39,840
Speaker 5: this technology has solved so many of these problems. At

485
00:22:39,840 --> 00:22:42,439
Speaker 5: the same time, we no longer need fixed cameras. I

486
00:22:42,560 --> 00:22:44,600
Speaker 5: now do it off the broadcast cameras, which are on

487
00:22:44,640 --> 00:22:47,680
Speaker 5: people's shoulders and they shake and they rotate and whatever

488
00:22:47,720 --> 00:22:50,000
Speaker 5: doesn't matter. It can still figure out what it's looking at.

489
00:22:50,160 --> 00:22:52,359
Speaker 5: This is the octagon, These are the fighters. Here are

490
00:22:52,359 --> 00:22:55,159
Speaker 5: their bodies, that's the ref He doesn't matter all of

491
00:22:55,200 --> 00:22:57,760
Speaker 5: that stuff. It can now figure out. It can begin

492
00:22:57,800 --> 00:23:00,720
Speaker 5: to understand what real landed myst is, and it can

493
00:23:00,760 --> 00:23:03,880
Speaker 5: do so on much less perfect data than we needed

494
00:23:03,920 --> 00:23:07,159
Speaker 5: the first time. That acceleration has been incredible. If I

495
00:23:07,160 --> 00:23:09,520
Speaker 5: didn't know step one and two and three and four,

496
00:23:10,040 --> 00:23:11,919
Speaker 5: I'm not sure how ready I would have been for

497
00:23:11,960 --> 00:23:12,720
Speaker 5: where we are now.

498
00:23:12,840 --> 00:23:16,560
Speaker 3: Yeah, So let's just dig into where we are now

499
00:23:16,600 --> 00:23:19,040
Speaker 3: a little bit. So you go. I'm interested in the

500
00:23:19,040 --> 00:23:23,239
Speaker 3: partnership with IBM and the kind of transition from this

501
00:23:23,320 --> 00:23:27,719
Speaker 3: period of experimentation to where you are now. So where

502
00:23:28,200 --> 00:23:30,960
Speaker 3: at the time you guys, you're aware of.

503
00:23:30,880 --> 00:23:32,480
Speaker 4: The development of AI yep.

504
00:23:33,119 --> 00:23:34,840
Speaker 3: And it's getting better and better and better and better

505
00:23:34,880 --> 00:23:37,960
Speaker 3: and better, and you clearly have an intuition that, oh,

506
00:23:38,040 --> 00:23:43,600
Speaker 3: this is this could potentially really open some opportunities for us.

507
00:23:43,720 --> 00:23:43,960
Speaker 4: Yep.

508
00:23:44,200 --> 00:23:46,760
Speaker 3: What's the point in time where you start to think, oh,

509
00:23:46,760 --> 00:23:47,480
Speaker 3: this could be real.

510
00:23:49,240 --> 00:23:54,239
Speaker 5: Our relationship with IBM is born of a little bit

511
00:23:54,240 --> 00:23:56,520
Speaker 5: of suffering on our behalf, is the best way to

512
00:23:56,520 --> 00:23:58,240
Speaker 5: put it. We had a partner as part of the

513
00:23:58,320 --> 00:24:01,400
Speaker 5: larger business that really wanted to focus on AI with us,

514
00:24:01,880 --> 00:24:04,840
Speaker 5: and we said, we would really like to do an

515
00:24:04,840 --> 00:24:07,959
Speaker 5: insights engine. We think it's really additive. These storylines in

516
00:24:08,000 --> 00:24:10,760
Speaker 5: real time would change the way that we talk about

517
00:24:10,840 --> 00:24:15,120
Speaker 5: the sport in broadcast. And for two years we went

518
00:24:15,160 --> 00:24:20,080
Speaker 5: down that road and it crashed and burned. We go

519
00:24:20,119 --> 00:24:23,959
Speaker 5: into the reasons, but they didn't understand the sport. They

520
00:24:24,000 --> 00:24:26,480
Speaker 5: tried to brute force the problem. And if you try

521
00:24:26,480 --> 00:24:28,320
Speaker 5: to brute force this problem, it's going to kill you.

522
00:24:28,760 --> 00:24:31,200
Speaker 5: And because there are just so many permutations out there,

523
00:24:32,040 --> 00:24:38,600
Speaker 5: and IBM starts a conversation with us, and very quickly

524
00:24:38,600 --> 00:24:42,480
Speaker 5: as we engage, a couple of things emerge, and they're

525
00:24:42,520 --> 00:24:44,919
Speaker 5: not about the tech. They say to us, we think

526
00:24:44,960 --> 00:24:46,240
Speaker 5: we can do it with the tech. We say, we

527
00:24:46,240 --> 00:24:47,320
Speaker 5: have to believe you that you can do it with

528
00:24:47,359 --> 00:24:49,040
Speaker 5: the tech. We don't know, we're not experts in it.

529
00:24:49,880 --> 00:24:53,480
Speaker 5: They can bring people who understand sports analysis to the table,

530
00:24:53,720 --> 00:24:55,960
Speaker 5: so all of a sudden, when my phone call ends,

531
00:24:56,040 --> 00:24:58,560
Speaker 5: when somebody at IBM has a question, they can talk

532
00:24:58,560 --> 00:25:01,000
Speaker 5: to somebody better. Yet, within the first of weeks, I said,

533
00:25:01,400 --> 00:25:03,240
Speaker 5: it would really be helpful to have somebody who's an

534
00:25:03,400 --> 00:25:06,600
Speaker 5: MMA fan, because some of your questions are specific to

535
00:25:06,640 --> 00:25:08,440
Speaker 5: the sport and we could answer them all the time.

536
00:25:08,480 --> 00:25:11,120
Speaker 5: But I think that the velocity here goes up if

537
00:25:11,160 --> 00:25:15,680
Speaker 5: you can just say, hey, Brad, what's happening here? Can

538
00:25:15,720 --> 00:25:20,040
Speaker 5: you explain what is the culture, what's important, what's not important? Right,

539
00:25:20,119 --> 00:25:21,680
Speaker 5: because you can come up with a lot of ideas

540
00:25:21,960 --> 00:25:23,879
Speaker 5: and fans will be like, I don't care, and it

541
00:25:23,880 --> 00:25:24,720
Speaker 5: gets worse from there.

542
00:25:24,760 --> 00:25:27,320
Speaker 4: So we quickly eliminated that. With IBM.

543
00:25:27,359 --> 00:25:30,159
Speaker 5: They had people on the bench who understood just the

544
00:25:30,640 --> 00:25:33,360
Speaker 5: area and then the specifics of our sport, and they

545
00:25:33,359 --> 00:25:35,040
Speaker 5: were able to fill in some of the gaps for us,

546
00:25:35,040 --> 00:25:37,199
Speaker 5: and so we could move at pace. And it was

547
00:25:37,240 --> 00:25:39,720
Speaker 5: all about the people at the beginning, and obviously beyond that,

548
00:25:39,800 --> 00:25:42,040
Speaker 5: they had to make good on the arctech and actually

549
00:25:42,040 --> 00:25:45,040
Speaker 5: do this. We can use the AI. It's not so

550
00:25:45,160 --> 00:25:47,440
Speaker 5: much that the AI needs to run the final product.

551
00:25:48,359 --> 00:25:50,240
Speaker 5: That's not what we're talking about. It's can the AI

552
00:25:50,840 --> 00:25:55,040
Speaker 5: speed up the creation of the product? Can AI help

553
00:25:55,119 --> 00:25:58,119
Speaker 5: us expand its reach? Can AI suggest to us the

554
00:25:58,160 --> 00:26:00,880
Speaker 5: things around the corner that we didn't see. That's where

555
00:26:00,880 --> 00:26:02,840
Speaker 5: the AI came in. That's really the genius of this.

556
00:26:02,880 --> 00:26:06,280
Speaker 5: Once it's all hard coded down, the AI does take

557
00:26:06,320 --> 00:26:08,520
Speaker 5: sort of a backseat to the code because once you

558
00:26:08,600 --> 00:26:10,400
Speaker 5: have to go fast, once you need to be efficient.

559
00:26:11,280 --> 00:26:11,560
Speaker 4: You know.

560
00:26:12,160 --> 00:26:14,800
Speaker 5: That's the way that you do it is you run

561
00:26:14,840 --> 00:26:17,600
Speaker 5: a whole bunch of traditional code and then you run

562
00:26:17,600 --> 00:26:19,280
Speaker 5: it through an interpretation model to be able to give

563
00:26:19,320 --> 00:26:21,640
Speaker 5: you natural language at the end. And that's super efficient

564
00:26:21,640 --> 00:26:23,960
Speaker 5: in AI. But it's an understanding of how to deploy

565
00:26:24,000 --> 00:26:26,240
Speaker 5: both of those things in concert and have the right

566
00:26:26,320 --> 00:26:26,879
Speaker 5: kind of people.

567
00:26:27,359 --> 00:26:30,080
Speaker 3: So walk me through here, IBM, what are they doing.

568
00:26:30,359 --> 00:26:34,120
Speaker 3: They're taking the video feed, they're taking the data feed

569
00:26:34,520 --> 00:26:35,280
Speaker 3: the data feed.

570
00:26:35,359 --> 00:26:38,160
Speaker 5: Correct, it's instead of trying to have them extract everything

571
00:26:38,200 --> 00:26:40,760
Speaker 5: out of the video feed, We've already done a lot

572
00:26:40,760 --> 00:26:42,359
Speaker 5: of that stuff and we're doing it at speed. So

573
00:26:42,400 --> 00:26:44,880
Speaker 5: we have my data feeds, and then we also have

574
00:26:45,560 --> 00:26:48,560
Speaker 5: any augmented information that we may be getting from external

575
00:26:48,680 --> 00:26:52,680
Speaker 5: AI computer vision systems, and that's coming in as the

576
00:26:52,720 --> 00:26:55,879
Speaker 5: source data to IBM. So now they're getting they're getting

577
00:26:55,880 --> 00:26:59,200
Speaker 5: a data feed, and they're getting it at speed. Their

578
00:26:59,320 --> 00:27:03,479
Speaker 5: job is to bust open that data feed into all

579
00:27:03,560 --> 00:27:06,120
Speaker 5: of the various questions that somebody might ask and think

580
00:27:06,119 --> 00:27:09,320
Speaker 5: about the interrelationships between data. So sometimes it's as simple

581
00:27:09,359 --> 00:27:14,280
Speaker 5: as this person is on pace to set the largest

582
00:27:14,320 --> 00:27:16,680
Speaker 5: number of takedown attempts we've seen the last twelve months

583
00:27:16,720 --> 00:27:18,720
Speaker 5: in this division. It's like a really nice talking point,

584
00:27:19,040 --> 00:27:22,879
Speaker 5: relatively simple all the way too. Here are two comparative

585
00:27:23,080 --> 00:27:25,520
Speaker 5: metrics for these two fighters, one of which is this

586
00:27:25,560 --> 00:27:27,200
Speaker 5: is his strength and this is his strength, and here's

587
00:27:27,200 --> 00:27:30,879
Speaker 5: how they're matching up right now and everything in between.

588
00:27:31,119 --> 00:27:32,159
Speaker 5: And to do that at speed.

589
00:27:32,720 --> 00:27:36,040
Speaker 3: Yeah, so they're taking two things. They're taking the data

590
00:27:36,280 --> 00:27:41,520
Speaker 3: stream from your scores and then also so you have

591
00:27:41,640 --> 00:27:43,480
Speaker 3: these vision cameras.

592
00:27:43,560 --> 00:27:46,520
Speaker 5: Yeah, it's a computer vision AI STAT system that we're

593
00:27:46,600 --> 00:27:48,520
Speaker 5: using to augment our STAT system.

594
00:27:48,600 --> 00:27:49,960
Speaker 3: And how does that work?

595
00:27:50,359 --> 00:27:52,679
Speaker 5: So the way that works is we've finally found somebody

596
00:27:52,680 --> 00:27:56,240
Speaker 5: who could do two things for us, give us accurate counts,

597
00:27:57,400 --> 00:28:00,960
Speaker 5: and be able to do so without us attaching a

598
00:28:00,960 --> 00:28:03,320
Speaker 5: whole bunch of cameras to the octagon, right, So they

599
00:28:03,320 --> 00:28:07,760
Speaker 5: can do it directly off of my handhelds, and they're

600
00:28:07,800 --> 00:28:10,280
Speaker 5: able to do things like give us combo percentage. They

601
00:28:10,280 --> 00:28:14,120
Speaker 5: can do punch and CounterPunch because they can assess that,

602
00:28:14,160 --> 00:28:14,600
Speaker 5: they could.

603
00:28:14,440 --> 00:28:15,280
Speaker 4: Talk about style.

604
00:28:15,520 --> 00:28:18,480
Speaker 5: That's a cross, that's a round house, that's right, those

605
00:28:18,560 --> 00:28:22,560
Speaker 5: kinds of things, and all of that is additive. Plus

606
00:28:22,600 --> 00:28:24,080
Speaker 5: by the way, they brought back all the motion stuff

607
00:28:24,080 --> 00:28:26,720
Speaker 5: because now my AI computer vision can actually do motion

608
00:28:27,000 --> 00:28:29,479
Speaker 5: for real, and so they brought all that back in

609
00:28:29,520 --> 00:28:30,919
Speaker 5: as well into the.

610
00:28:30,880 --> 00:28:33,400
Speaker 3: Way I can now do just because the tech is better.

611
00:28:33,320 --> 00:28:34,080
Speaker 4: Text a lot better.

612
00:28:34,440 --> 00:28:37,080
Speaker 5: The ability to just know there's blue tape on the

613
00:28:37,080 --> 00:28:40,160
Speaker 5: blue fighter's wrists and red tape on the red fighter's wrist,

614
00:28:40,200 --> 00:28:43,200
Speaker 5: and we've gotten to the point now where scaltill tracking

615
00:28:43,200 --> 00:28:46,080
Speaker 5: and computer vision is able to identify people by that

616
00:28:46,120 --> 00:28:49,440
Speaker 5: and also take additional notes that dude's in green shorts

617
00:28:49,480 --> 00:28:51,440
Speaker 5: and that dude's in yellow shorts, right, and it'll do

618
00:28:51,440 --> 00:28:53,120
Speaker 5: those kinds of things to just figure out who's who.

619
00:28:53,560 --> 00:28:56,600
Speaker 5: And the hardest thing for a computer vision system to

620
00:28:56,600 --> 00:28:59,360
Speaker 5: do as it does motion is when we become the

621
00:28:59,400 --> 00:29:03,560
Speaker 5: eight legged B and we come apart. It can know

622
00:29:03,640 --> 00:29:05,320
Speaker 5: that we're purple, right, we're red and blue at the

623
00:29:05,320 --> 00:29:07,280
Speaker 5: same time, but when we come apart, it has to

624
00:29:07,360 --> 00:29:10,560
Speaker 5: rapidly understand who's red and who's blue. Again, that's where

625
00:29:10,560 --> 00:29:12,880
Speaker 5: a lot of these motion systems had a lot of difficulty.

626
00:29:13,080 --> 00:29:15,760
Speaker 5: Refs in black pretty easy to figure out, like somebody's

627
00:29:15,800 --> 00:29:16,320
Speaker 5: dressed in black.

628
00:29:16,520 --> 00:29:16,720
Speaker 4: Head to.

629
00:29:17,120 --> 00:29:20,800
Speaker 3: You've got these data streams, and presumably there is some

630
00:29:20,840 --> 00:29:23,600
Speaker 3: overlap between them, but that's fine. You're just feeding as

631
00:29:23,680 --> 00:29:27,960
Speaker 3: much data as you can from these two different sources

632
00:29:28,520 --> 00:29:34,640
Speaker 3: into the IBM black box, and the black box is

633
00:29:35,600 --> 00:29:40,800
Speaker 3: generating stories for us, ideas, observations, insights.

634
00:29:40,960 --> 00:29:42,479
Speaker 5: I can give you a little more detail on how

635
00:29:42,480 --> 00:29:44,760
Speaker 5: it happens. I think that the beauty of unpacking how

636
00:29:44,960 --> 00:29:48,240
Speaker 5: AI works is there are moments of magic and moments

637
00:29:48,240 --> 00:29:50,200
Speaker 5: of drudgery, and they have to work perfectly together.

638
00:29:50,680 --> 00:29:51,240
Speaker 4: Right AI.

639
00:29:51,800 --> 00:29:55,240
Speaker 5: The data comes in to IBM, and it has a

640
00:29:55,280 --> 00:29:59,440
Speaker 5: storage system called Iceberg, and it runs a classic data

641
00:29:59,480 --> 00:30:03,400
Speaker 5: process called ETL. It's extract transform load. We care about

642
00:30:03,400 --> 00:30:05,880
Speaker 5: our data at the fight level or the round level.

643
00:30:06,480 --> 00:30:08,880
Speaker 5: IBM is trying to tell us things about our fighters,

644
00:30:08,880 --> 00:30:11,800
Speaker 5: so it is consolidating all of its information in these

645
00:30:11,800 --> 00:30:14,320
Speaker 5: big fighter tables that it can then sick the AI on,

646
00:30:15,000 --> 00:30:19,160
Speaker 5: and then the data moves from this etl iceberg level

647
00:30:19,520 --> 00:30:21,719
Speaker 5: up a level. They actually, I think they do it

648
00:30:21,760 --> 00:30:24,040
Speaker 5: by metals, So you start at bronze and you move

649
00:30:24,040 --> 00:30:26,200
Speaker 5: all the way up to gold and you move up

650
00:30:26,240 --> 00:30:29,320
Speaker 5: a level. So now your data is mezzanined basically for

651
00:30:29,440 --> 00:30:32,080
Speaker 5: the for the actual AI system to be able to

652
00:30:32,160 --> 00:30:35,960
Speaker 5: draw on it and efficiently process this data into insights,

653
00:30:36,480 --> 00:30:39,680
Speaker 5: and then it moves up into insight delivery, where we

654
00:30:39,720 --> 00:30:42,080
Speaker 5: have an interface that allows us to not just look

655
00:30:42,120 --> 00:30:44,640
Speaker 5: at insights. But then I have a human factor's problem.

656
00:30:45,040 --> 00:30:47,840
Speaker 5: My human needs to see the insights and he can't

657
00:30:47,920 --> 00:30:50,800
Speaker 5: see or she can't see all of them. Right, that's

658
00:30:51,120 --> 00:30:52,959
Speaker 5: way too many. So you have to have a scoring

659
00:30:52,960 --> 00:30:56,520
Speaker 5: system that says this insight is more relevant than this insight,

660
00:30:56,840 --> 00:31:00,440
Speaker 5: and surface only those so that a human Actually, this

661
00:31:00,560 --> 00:31:02,160
Speaker 5: was true of Watson a long time ago, right when

662
00:31:02,160 --> 00:31:06,600
Speaker 5: it did medical diagnostics. It couldn't give you the answer.

663
00:31:06,720 --> 00:31:09,480
Speaker 5: It could give you seven answers, and a doctor would

664
00:31:09,480 --> 00:31:11,239
Speaker 5: look through them and say it, got it, got it.

665
00:31:11,320 --> 00:31:14,200
Speaker 5: That's silly, that's silly. Hadn't thought of that one? And

666
00:31:14,240 --> 00:31:16,280
Speaker 5: that was the value right here. And we have a

667
00:31:16,320 --> 00:31:18,920
Speaker 5: similar thing right where we'll get a whole list of

668
00:31:18,960 --> 00:31:20,440
Speaker 5: them and you get to pick one of them and

669
00:31:20,440 --> 00:31:23,400
Speaker 5: be like go. And so that's how the data comes in.

670
00:31:23,480 --> 00:31:25,240
Speaker 5: And to everybody else that should be a black box.

671
00:31:25,720 --> 00:31:27,800
Speaker 5: But I'd prefer it not be a black box to us.

672
00:31:28,280 --> 00:31:30,960
Speaker 3: In the course it is my possibility of question, in

673
00:31:31,040 --> 00:31:36,240
Speaker 3: the course of a fight, how many insights are being generated?

674
00:31:36,880 --> 00:31:38,560
Speaker 4: Yeah, that's an excellent question.

675
00:31:38,680 --> 00:31:41,560
Speaker 5: So in an active let's say we'll pick what I

676
00:31:41,560 --> 00:31:43,800
Speaker 5: would consider to be the Cadillac of fights. Right, So

677
00:31:43,840 --> 00:31:48,160
Speaker 5: it's five rounds, it's a championship fight. So you have

678
00:31:48,280 --> 00:31:52,360
Speaker 5: two fighters with long history, good records in that fight.

679
00:31:53,160 --> 00:31:56,040
Speaker 5: I mean it could be it could be a thousand,

680
00:31:56,280 --> 00:32:01,760
Speaker 5: a thousand, because because every permutation is possible. Yeah, right,

681
00:32:01,880 --> 00:32:04,920
Speaker 5: So the system is generating just an enormous amount, and

682
00:32:05,120 --> 00:32:08,200
Speaker 5: the real the intelligence is in filtering it out so

683
00:32:08,240 --> 00:32:11,240
Speaker 5: that my humans don't see a thousand, My humans see

684
00:32:11,280 --> 00:32:17,000
Speaker 5: fifteen fifteen. Yeah, because remember the moment that in a fight,

685
00:32:17,040 --> 00:32:19,320
Speaker 5: there's not a lot of chances, right, I'm not going

686
00:32:19,400 --> 00:32:22,840
Speaker 5: to spit out all fifteen onto the screen. In traditional statistics,

687
00:32:22,920 --> 00:32:25,640
Speaker 5: I've got two stats around. Maybe we don't have any

688
00:32:25,680 --> 00:32:27,760
Speaker 5: stats at the first half of the first round. If

689
00:32:27,800 --> 00:32:29,920
Speaker 5: something as amazing as happening, my job is to move

690
00:32:29,960 --> 00:32:32,200
Speaker 5: out of the way and just let that sit on

691
00:32:32,240 --> 00:32:35,280
Speaker 5: the screen, And so my insights are there to be

692
00:32:35,360 --> 00:32:37,480
Speaker 5: the most valuable thing I can put on screen at

693
00:32:37,560 --> 00:32:39,960
Speaker 5: that moment. And that's going to happen in a fight

694
00:32:41,120 --> 00:32:44,520
Speaker 5: twice and over the course of the night ten times.

695
00:32:44,960 --> 00:32:48,400
Speaker 5: But when it happens, everybody should have been talking about

696
00:32:48,400 --> 00:32:50,800
Speaker 5: this fight like this, and now they should be talking

697
00:32:50,800 --> 00:32:52,000
Speaker 5: about this fight like that.

698
00:32:52,120 --> 00:32:54,720
Speaker 3: Give me an example of the kind of insight you

699
00:32:54,840 --> 00:32:58,720
Speaker 3: get that you probably would never have gotten in the

700
00:32:58,880 --> 00:33:01,120
Speaker 3: kind of previous sure universe.

701
00:33:01,320 --> 00:33:04,080
Speaker 5: Sure, I think compound insights are the most complicated right

702
00:33:04,480 --> 00:33:08,880
Speaker 5: where it's we used to get basic, basic, but difficult

703
00:33:08,920 --> 00:33:13,320
Speaker 5: to answer questions when was the last time someone want

704
00:33:13,320 --> 00:33:16,080
Speaker 5: to fight after being down one hundred strikes? Like that's

705
00:33:16,120 --> 00:33:17,920
Speaker 5: a hard thing to throw into a sequel database at

706
00:33:17,920 --> 00:33:20,360
Speaker 5: the Latin a moment's notice and get that answer. But

707
00:33:20,440 --> 00:33:22,680
Speaker 5: you can do that kind of thing with INSIGHT's engine.

708
00:33:22,720 --> 00:33:25,560
Speaker 5: Right was like, You're like, oh, it is looking for

709
00:33:25,600 --> 00:33:27,760
Speaker 5: it's one of its pipelines. It's looking for those kinds

710
00:33:27,880 --> 00:33:30,880
Speaker 5: of disconnects between the data and the reality.

711
00:33:30,920 --> 00:33:32,600
Speaker 4: This is the.

712
00:33:32,160 --> 00:33:36,800
Speaker 5: Largest differential striking differential between two fighters, you know, in

713
00:33:36,840 --> 00:33:40,320
Speaker 5: a middleweight title fight since and here's this other iconic

714
00:33:40,360 --> 00:33:43,760
Speaker 5: fight that everybody remembers. Right, It's those kinds of things

715
00:33:43,840 --> 00:33:47,640
Speaker 5: that you might in the back of your head think

716
00:33:47,640 --> 00:33:49,640
Speaker 5: to yourself like, huh, this feels like I don't think

717
00:33:49,640 --> 00:33:50,600
Speaker 5: I've seen this in a while.

718
00:33:50,840 --> 00:33:51,479
Speaker 4: But you can't.

719
00:33:51,640 --> 00:33:54,280
Speaker 5: You can't give voice to it. And in a live fight,

720
00:33:54,560 --> 00:33:58,320
Speaker 5: there's very little time. So what will happen is two fighters,

721
00:33:58,400 --> 00:34:00,959
Speaker 5: Let's say it's a striking statistic. Two fighters have not

722
00:34:01,000 --> 00:34:02,920
Speaker 5: had this much of a striking differential in a fight

723
00:34:03,040 --> 00:34:07,320
Speaker 5: since X point. It is currently three minutes into the

724
00:34:07,400 --> 00:34:12,160
Speaker 5: round of round three, let's say, and the commentators are talking,

725
00:34:12,200 --> 00:34:15,160
Speaker 5: we're watching the action, everything's happening. You get this sense

726
00:34:15,200 --> 00:34:18,640
Speaker 5: in your head, I need to get in the next

727
00:34:18,640 --> 00:34:21,480
Speaker 5: two minutes. The last thirty seconds are already we're going

728
00:34:21,480 --> 00:34:24,040
Speaker 5: out to commercial like we're not interrupting that the fight

729
00:34:24,040 --> 00:34:26,719
Speaker 5: could end at any moment, So there's time pressure I've

730
00:34:26,760 --> 00:34:28,680
Speaker 5: got to hit a promo at some point in the

731
00:34:28,719 --> 00:34:31,160
Speaker 5: middle of this thing, So you see how my time

732
00:34:31,520 --> 00:34:33,759
Speaker 5: is getting shorter and shorter. To be able to deliver this,

733
00:34:33,960 --> 00:34:36,440
Speaker 5: I've got to have my producer be able to communicate it.

734
00:34:36,440 --> 00:34:39,040
Speaker 5: To the producer, it's got to communicate it to commentators,

735
00:34:39,080 --> 00:34:40,160
Speaker 5: who's got to be able to get it out of

736
00:34:40,200 --> 00:34:42,200
Speaker 5: a system into a graphic and then sold in to

737
00:34:42,239 --> 00:34:43,160
Speaker 5: be able to put on screen.

738
00:34:43,200 --> 00:34:46,640
Speaker 4: All of that has to happen in call it between thirty.

739
00:34:47,120 --> 00:34:48,600
Speaker 5: I said to you, like, I have three minutes, so

740
00:34:48,600 --> 00:34:50,640
Speaker 5: I have two minutes left in my round, probably only

741
00:34:50,640 --> 00:34:54,279
Speaker 5: have about forty seconds window, and I'm going to lose

742
00:34:54,320 --> 00:34:56,680
Speaker 5: it starting the moment I think of it until the

743
00:34:56,680 --> 00:34:58,240
Speaker 5: moment it hits screen. So I've got to have about

744
00:34:58,239 --> 00:35:03,040
Speaker 5: a fifteen second turnaround. And so it's not you don't

745
00:35:03,080 --> 00:35:04,799
Speaker 5: have to ask a very complex question for you to

746
00:35:04,840 --> 00:35:07,759
Speaker 5: exceed the fifteen second turnaround unless it's at your fingertips.

747
00:35:08,080 --> 00:35:10,279
Speaker 5: And that's what insights engine is for. It's it's to

748
00:35:10,360 --> 00:35:11,640
Speaker 5: kind of come in and be like, I know you

749
00:35:11,680 --> 00:35:13,920
Speaker 5: have some cool ideas, but we're never going to be

750
00:35:13,920 --> 00:35:16,719
Speaker 5: able to get him into fight time unless something is

751
00:35:17,120 --> 00:35:18,600
Speaker 5: pushing not pulling.

752
00:35:19,520 --> 00:35:21,680
Speaker 3: You need to have buy in from the commentators.

753
00:35:22,160 --> 00:35:23,960
Speaker 4: Yes, we do so.

754
00:35:24,000 --> 00:35:26,160
Speaker 5: First of all, being a commentator is a very sophisticated

755
00:35:26,480 --> 00:35:28,799
Speaker 5: undertaking almost anything you see. And it's one of those

756
00:35:28,840 --> 00:35:30,279
Speaker 5: things that you watch happen and you say, well, that

757
00:35:30,280 --> 00:35:32,680
Speaker 5: doesn't look that difficult. But a commentator has an IFB

758
00:35:32,719 --> 00:35:36,040
Speaker 5: a little ear piece in and I've watched them do

759
00:35:36,080 --> 00:35:37,759
Speaker 5: this and I can't wrap my head around doing it,

760
00:35:37,800 --> 00:35:40,359
Speaker 5: which is you must be looking at the camera and

761
00:35:40,400 --> 00:35:45,359
Speaker 5: presenting information in coherent pros while the producer is talking

762
00:35:45,360 --> 00:35:47,640
Speaker 5: in your ear, and then when you stop talking, he

763
00:35:47,680 --> 00:35:49,799
Speaker 5: starts talking. No, no, no, it's talking in your ear while

764
00:35:49,800 --> 00:35:51,680
Speaker 5: you're talking. And you've got to be able to process

765
00:35:51,719 --> 00:35:54,680
Speaker 5: that information and two sentences hence make a part of

766
00:35:54,719 --> 00:36:00,279
Speaker 5: your delivery. And that information processing capability is by in

767
00:36:00,400 --> 00:36:04,760
Speaker 5: large sophisticated. Now, people are people. Some people are more

768
00:36:04,840 --> 00:36:08,120
Speaker 5: pro stat some people are more you know, anti static.

769
00:36:08,120 --> 00:36:10,120
Speaker 5: It doesn't really matter. You just have to know who

770
00:36:10,120 --> 00:36:13,360
Speaker 5: that is. But they have a relationship of trust with

771
00:36:13,560 --> 00:36:17,520
Speaker 5: their producer, and if the producer says I want to

772
00:36:17,560 --> 00:36:18,160
Speaker 5: go here.

773
00:36:19,080 --> 00:36:19,919
Speaker 4: They can push back.

774
00:36:20,400 --> 00:36:22,720
Speaker 5: But those two people work together on the regular producer's

775
00:36:22,760 --> 00:36:24,160
Speaker 5: going to sell you the things he thinks you're going

776
00:36:24,239 --> 00:36:26,960
Speaker 5: to buy, right, or at least the things he thinks.

777
00:36:27,280 --> 00:36:28,560
Speaker 5: This is as far as I can push it, and

778
00:36:28,600 --> 00:36:30,439
Speaker 5: you'll really go there because he wants you to sound

779
00:36:30,480 --> 00:36:32,800
Speaker 5: authentic and he certainly doesn't want you to be fighting

780
00:36:32,840 --> 00:36:37,440
Speaker 5: with him on air, right, So that very human process

781
00:36:37,520 --> 00:36:40,600
Speaker 5: is taking place. And by and large, one of the

782
00:36:40,600 --> 00:36:45,200
Speaker 5: reasons Insights Engine produces human readable stories is because I

783
00:36:45,239 --> 00:36:49,280
Speaker 5: would have pushed back if somebody said, IBM Insights Engine

784
00:36:49,360 --> 00:36:52,000
Speaker 5: says that so and so is fifty seven point two

785
00:36:52,080 --> 00:36:56,240
Speaker 5: four six percent likely to win the fight, Like okay,

786
00:36:57,239 --> 00:37:01,600
Speaker 5: and that doesn't feel authentic. But so like we've noticed

787
00:37:01,640 --> 00:37:04,960
Speaker 5: that he's advancing and he's a counterpuncher. This is off

788
00:37:05,320 --> 00:37:07,600
Speaker 5: and so now hopefully the commentator is going like, wow,

789
00:37:07,719 --> 00:37:11,520
Speaker 5: the movement in this fight is different than we would

790
00:37:11,560 --> 00:37:13,960
Speaker 5: expect to see, and let me give you some things

791
00:37:13,960 --> 00:37:18,000
Speaker 5: to bolster that that point. Or and people have done this,

792
00:37:18,080 --> 00:37:20,799
Speaker 5: We have given them an insight and they're like, oh, yeah,

793
00:37:20,840 --> 00:37:24,279
Speaker 5: you're right, Like that's what's weird yeah about this fight?

794
00:37:24,440 --> 00:37:30,399
Speaker 3: Yeah, it's fascinating because genuine engagement with the insights. He's

795
00:37:30,440 --> 00:37:34,400
Speaker 3: actually from the viewers standpoint, the most interesting thing, the

796
00:37:34,440 --> 00:37:37,000
Speaker 3: idea that they would be there'll be some tension between

797
00:37:37,040 --> 00:37:39,839
Speaker 3: what we're seeing and what's actually happening. That is one

798
00:37:39,840 --> 00:37:43,520
Speaker 3: of the most interesting things in sport, right Oh, I thought,

799
00:37:44,600 --> 00:37:47,640
Speaker 3: I thought, you know, Steph Curry had a great game,

800
00:37:47,719 --> 00:37:49,760
Speaker 3: and then I look after at the stat line afterwards

801
00:37:49,840 --> 00:37:53,880
Speaker 3: and I'm like, or the opposite is even better. He

802
00:37:53,920 --> 00:37:55,840
Speaker 3: looked terrible out there. I look at the statline, He's like,

803
00:37:56,080 --> 00:37:58,120
Speaker 3: oh my god. He was like, I find my shot

804
00:37:58,280 --> 00:37:59,399
Speaker 3: seventy percent from the field.

805
00:37:59,520 --> 00:38:01,400
Speaker 5: I feel like I do that more often than not,

806
00:38:01,440 --> 00:38:03,719
Speaker 5: where you're like, that looks terrible. So a couple of

807
00:38:03,760 --> 00:38:05,319
Speaker 5: things that are true in fighting that are I think

808
00:38:05,360 --> 00:38:09,080
Speaker 5: true elsewhere is if people have a favorite, they tend

809
00:38:09,080 --> 00:38:11,360
Speaker 5: to watch their guy, and so their guy always, unless

810
00:38:11,360 --> 00:38:13,960
Speaker 5: he looks terrible, he looks great because you're not watching

811
00:38:13,960 --> 00:38:16,960
Speaker 5: the other guy. You're just watching your guy, and that

812
00:38:17,360 --> 00:38:20,239
Speaker 5: stats help bridge that gap a lot. Our commentators in

813
00:38:20,560 --> 00:38:22,520
Speaker 5: general don't fall into that trap, and so when people

814
00:38:22,600 --> 00:38:25,279
Speaker 5: fight with the commentators. When when fans fight with the commentators,

815
00:38:25,280 --> 00:38:27,920
Speaker 5: they're watching one guy, They're not watching both guys. I

816
00:38:27,920 --> 00:38:29,520
Speaker 5: will watch fights and I will say, you know, I'm

817
00:38:29,520 --> 00:38:32,200
Speaker 5: not watching I'm not seeing the fight you're seeing. Some

818
00:38:32,239 --> 00:38:35,319
Speaker 5: of that is because a lot of our commentators are practitioners.

819
00:38:35,400 --> 00:38:37,360
Speaker 5: They're seeing something I'm not seeing and they but it

820
00:38:37,440 --> 00:38:39,160
Speaker 5: is then their job to explain to me what I'm

821
00:38:39,160 --> 00:38:42,879
Speaker 5: not seeing, and that's fine. I think that The other

822
00:38:42,880 --> 00:38:44,520
Speaker 5: place that I would want to take this, and where

823
00:38:44,520 --> 00:38:48,160
Speaker 5: insights I think are very useful, is our fans watching

824
00:38:48,200 --> 00:38:51,680
Speaker 5: a lot of contexts a lot and how many times

825
00:38:51,719 --> 00:38:53,040
Speaker 5: have you walked by a game?

826
00:38:53,440 --> 00:38:53,600
Speaker 2: Right?

827
00:38:53,719 --> 00:38:56,839
Speaker 5: The classic is I'm at a sports bar and maybe

828
00:38:56,840 --> 00:38:58,600
Speaker 5: the audios on, but maybe the audio is not. Maybe

829
00:38:58,640 --> 00:39:00,680
Speaker 5: it doesn't matter because it's loud. I want to be

830
00:39:00,719 --> 00:39:03,080
Speaker 5: able to provide an experience for that person that is

831
00:39:03,160 --> 00:39:07,440
Speaker 5: meaningful and potentially more meaningful. So, yes, my insights are

832
00:39:07,440 --> 00:39:09,839
Speaker 5: going to interact with my commentators. But if I can't

833
00:39:09,840 --> 00:39:12,040
Speaker 5: hear the commentators, how nice to be able to put

834
00:39:12,040 --> 00:39:14,800
Speaker 5: a storyline, not just a number, but a storyline directly

835
00:39:14,840 --> 00:39:17,359
Speaker 5: on screen. Right, we'll have this with milestone so and

836
00:39:17,400 --> 00:39:20,000
Speaker 5: so has exceeded their best ever whatever, and it's just

837
00:39:20,040 --> 00:39:21,799
Speaker 5: a moment where you're watching it like that's cool, and

838
00:39:21,840 --> 00:39:22,760
Speaker 5: you'll go back to watching.

839
00:39:22,920 --> 00:39:24,640
Speaker 4: And that's really that's what we're going for.

840
00:39:25,239 --> 00:39:28,080
Speaker 3: Are we fundamentally changing the relationship between the fan and

841
00:39:28,120 --> 00:39:28,520
Speaker 3: the sport.

842
00:39:29,400 --> 00:39:32,440
Speaker 4: Interesting. I I didn't set.

843
00:39:32,200 --> 00:39:35,440
Speaker 5: Out to have insights engine fundamentally change the relationship of

844
00:39:35,480 --> 00:39:39,880
Speaker 5: the fan and the sport. I wanted. I wanted something

845
00:39:39,880 --> 00:39:43,480
Speaker 5: to fit like a glove. Right if if you were

846
00:39:43,520 --> 00:39:47,400
Speaker 5: sitting next to someone who had an encyclopedic understanding of

847
00:39:47,440 --> 00:39:52,279
Speaker 5: the sport and they were in and they were fun, right,

848
00:39:52,360 --> 00:39:54,360
Speaker 5: that was a big part of it. What are the

849
00:39:54,400 --> 00:39:57,960
Speaker 5: kinds of things they could tell you where you're like, oh, cool,

850
00:39:58,840 --> 00:40:02,080
Speaker 5: and it's more fun to watch because of that and

851
00:40:02,120 --> 00:40:04,719
Speaker 5: not I think where a lot of the push was

852
00:40:04,800 --> 00:40:07,600
Speaker 5: in adding tech to the sport, which is we're gonna

853
00:40:07,600 --> 00:40:10,759
Speaker 5: make it nerdier like that was the thing I don't

854
00:40:10,800 --> 00:40:15,080
Speaker 5: want to do. I'm interested in leaning in to what

855
00:40:15,239 --> 00:40:20,920
Speaker 5: makes this great and letting you see writ in the

856
00:40:20,960 --> 00:40:24,080
Speaker 5: classic the stats to make the invisible visible, letting you

857
00:40:24,160 --> 00:40:26,880
Speaker 5: see those things but in a narrative arc so that

858
00:40:26,960 --> 00:40:30,240
Speaker 5: you can contextualize them. You can say something more intelligent

859
00:40:30,280 --> 00:40:33,080
Speaker 5: about sport, but like it. I wasn't here to make

860
00:40:33,080 --> 00:40:35,879
Speaker 5: you feel like I changed the sport. I was here

861
00:40:35,880 --> 00:40:40,080
Speaker 5: to make you feel like you were spoken to more

862
00:40:40,120 --> 00:40:42,400
Speaker 5: crisply about it. You were able to speak more crisply

863
00:40:42,440 --> 00:40:47,120
Speaker 5: about it, and the fight just looked a little shinier,

864
00:40:47,160 --> 00:40:49,799
Speaker 5: a little brighter because of this kind of thing. And

865
00:40:49,840 --> 00:40:53,160
Speaker 5: like I said, when this tech does its best work,

866
00:40:53,960 --> 00:40:55,760
Speaker 5: it just disappears.

867
00:40:55,840 --> 00:40:58,160
Speaker 3: Hearing you describe it, it strikes me that what you're

868
00:40:58,160 --> 00:41:00,960
Speaker 3: doing is you're moving people up a level.

869
00:41:01,080 --> 00:41:03,560
Speaker 5: You're right, we are bringing it to a place where

870
00:41:03,680 --> 00:41:06,040
Speaker 5: you feel like you can you could even have an

871
00:41:06,040 --> 00:41:08,640
Speaker 5: opinion because you understand enough of what's going on. You

872
00:41:08,680 --> 00:41:11,719
Speaker 5: are one hundred percent right. We're moving everybody right right

873
00:41:11,840 --> 00:41:14,640
Speaker 5: up the scale on a daily basis. I want to

874
00:41:14,640 --> 00:41:18,480
Speaker 5: go back to INSIGHT's engine is not here to feel technical,

875
00:41:19,000 --> 00:41:23,120
Speaker 5: and that's the genius of it. It's simplicity. It's narrative

876
00:41:23,440 --> 00:41:25,360
Speaker 5: is the thing that allows you to bring people forward

877
00:41:25,400 --> 00:41:28,440
Speaker 5: because you're not using jargon, because you're not just into

878
00:41:28,440 --> 00:41:32,160
Speaker 5: the numbers, right, That's really that's really the key here.

879
00:41:32,560 --> 00:41:34,920
Speaker 3: When you look back at your experience with fight data,

880
00:41:35,000 --> 00:41:38,760
Speaker 3: are there a few categories that are essential to winning?

881
00:41:40,760 --> 00:41:43,200
Speaker 5: I think the easiest way to answer that is the

882
00:41:43,360 --> 00:41:47,640
Speaker 5: entire statistical system that we built is based on only

883
00:41:47,640 --> 00:41:48,240
Speaker 5: one question.

884
00:41:49,320 --> 00:41:53,400
Speaker 4: We don't track style.

885
00:41:53,440 --> 00:41:57,640
Speaker 5: For example, I don't track whether somebody hit you with

886
00:41:58,360 --> 00:42:01,680
Speaker 5: a cross or a jab. Even more ridiculous, in the

887
00:42:01,680 --> 00:42:04,440
Speaker 5: official statistics, we don't track whether we hit you with

888
00:42:04,440 --> 00:42:07,040
Speaker 5: an arm or a leg. And that's because the entire

889
00:42:07,040 --> 00:42:11,759
Speaker 5: statistical system is built on this question what moved you

890
00:42:11,800 --> 00:42:14,759
Speaker 5: closer to winning the fight? And it turns out that

891
00:42:14,800 --> 00:42:16,760
Speaker 5: what moves you closer to winning a fight is where

892
00:42:16,840 --> 00:42:19,960
Speaker 5: you take damage, right, and how much damage you take.

893
00:42:20,160 --> 00:42:22,719
Speaker 5: So if I hit you in the head, arm, leg

894
00:42:22,760 --> 00:42:26,400
Speaker 5: doesn't really matter, style doesn't really matter, it's did it land?

895
00:42:26,560 --> 00:42:30,240
Speaker 5: So the whole statistical system is built on that question

896
00:42:30,480 --> 00:42:33,200
Speaker 5: of how much of each of these things did you do?

897
00:42:33,560 --> 00:42:37,120
Speaker 5: These things which we have directly correlated to victory, And

898
00:42:37,560 --> 00:42:40,600
Speaker 5: with live statistics and the augments, we can do lefts

899
00:42:40,600 --> 00:42:42,239
Speaker 5: and rights and things like that people want to see.

900
00:42:43,160 --> 00:42:46,400
Speaker 5: But that's the fundamental question. So for the question is

901
00:42:46,760 --> 00:42:48,319
Speaker 5: what's the most important thing you could do?

902
00:42:49,400 --> 00:42:50,080
Speaker 4: Land strikes?

903
00:42:50,160 --> 00:42:53,239
Speaker 5: Land significant strikes to the head that's going to end

904
00:42:53,239 --> 00:42:56,040
Speaker 5: most fights. Can you end a fight with significant strikes

905
00:42:56,080 --> 00:42:58,040
Speaker 5: to the body that massively outpace your opponent?

906
00:42:58,200 --> 00:42:58,560
Speaker 4: You can?

907
00:42:58,920 --> 00:43:01,560
Speaker 5: There at a second level, down can And then there

908
00:43:01,560 --> 00:43:03,800
Speaker 5: are things that are a little harder to talk about

909
00:43:03,800 --> 00:43:08,320
Speaker 5: that way, if you have more ground control, that's putting

910
00:43:08,400 --> 00:43:12,600
Speaker 5: you closer to completing a submission. So that again is

911
00:43:12,920 --> 00:43:14,440
Speaker 5: the kind of thing. And obviously once we get to

912
00:43:14,560 --> 00:43:17,040
Speaker 5: the number of submission attempts you attempt, are going to

913
00:43:17,080 --> 00:43:19,279
Speaker 5: make it more likely that one of those will be successful.

914
00:43:19,719 --> 00:43:22,200
Speaker 5: But it is all built from this basic question of

915
00:43:22,320 --> 00:43:25,040
Speaker 5: all of our stats, does it get you closer to

916
00:43:25,080 --> 00:43:25,640
Speaker 5: winning the fight?

917
00:43:26,000 --> 00:43:26,920
Speaker 2: Yeah?

918
00:43:27,080 --> 00:43:29,360
Speaker 3: One last question. I know you've thought about this a lot,

919
00:43:29,920 --> 00:43:34,000
Speaker 3: which is next. What's like if I, IF I, if

920
00:43:34,040 --> 00:43:36,839
Speaker 3: I invite you back. We have this conversation five years soon. Now,

921
00:43:37,760 --> 00:43:39,600
Speaker 3: what is the fun wrinkle that you're working on.

922
00:43:40,000 --> 00:43:44,160
Speaker 4: We have a couple of things that are in the works.

923
00:43:46,080 --> 00:43:49,840
Speaker 4: Some of them are very sideways to this, but there so.

924
00:43:49,960 --> 00:43:54,080
Speaker 5: One of them, which we're prototyping is the ability to

925
00:43:54,160 --> 00:43:59,120
Speaker 5: bring additional voices into the mix seamlessly. So I'll give

926
00:43:59,120 --> 00:44:02,040
Speaker 5: you a perfect example. If the coaches who are on

927
00:44:02,080 --> 00:44:04,680
Speaker 5: the side and they're yelling in whatever language they are yelling,

928
00:44:05,880 --> 00:44:09,520
Speaker 5: they are attempting to influence the fight, and it is

929
00:44:09,600 --> 00:44:11,280
Speaker 5: and we do this occasionally. We'll cut to the coaches

930
00:44:11,320 --> 00:44:13,360
Speaker 5: camp and we'll listen to them for a moment, but

931
00:44:13,400 --> 00:44:18,040
Speaker 5: they're doing it constantly, wouldn't it be interesting to understand

932
00:44:18,440 --> 00:44:21,960
Speaker 5: what they're trying to do and what it means. So,

933
00:44:22,000 --> 00:44:23,879
Speaker 5: for example, if I'm saying you're doing great, but you're

934
00:44:23,920 --> 00:44:26,600
Speaker 5: not doing great by being encouraging. If I'm saying you

935
00:44:26,600 --> 00:44:28,080
Speaker 5: are doing great and you are doing great, what does

936
00:44:28,080 --> 00:44:29,600
Speaker 5: that mean? What is my tone of voice, what's my

937
00:44:29,680 --> 00:44:31,799
Speaker 5: level of stress? All of these kinds of things, so

938
00:44:31,880 --> 00:44:37,759
Speaker 5: that you could very subtly be able to distill that

939
00:44:37,800 --> 00:44:39,799
Speaker 5: information and bring it in and say, like, by the way,

940
00:44:39,840 --> 00:44:40,520
Speaker 5: if you want to know what.

941
00:44:40,480 --> 00:44:44,160
Speaker 4: The coaches think, feels interesting? So that's like a really

942
00:44:44,239 --> 00:44:45,279
Speaker 4: nice one that's relatively close.

943
00:44:45,320 --> 00:44:47,560
Speaker 3: Now I have a sense of the psychological context of

944
00:44:47,600 --> 00:44:49,080
Speaker 3: the coach and the athlete.

945
00:44:49,280 --> 00:44:52,080
Speaker 5: Let's go further away, right, You said five years in

946
00:44:52,080 --> 00:44:55,040
Speaker 5: the future. I think five years in the future. I

947
00:44:55,120 --> 00:44:58,439
Speaker 5: have two beliefs. I don't believe people want to lean

948
00:44:58,440 --> 00:45:00,680
Speaker 5: forward experience, as much as everybody would love them to

949
00:45:00,760 --> 00:45:03,040
Speaker 5: want to lean forward experience. If you could watch a

950
00:45:03,040 --> 00:45:08,200
Speaker 5: Steph Curry performance and I could let you controlled the camera.

951
00:45:08,520 --> 00:45:09,560
Speaker 3: No, I don't want to get OK.

952
00:45:09,960 --> 00:45:10,960
Speaker 4: I don't believe that people do.

953
00:45:11,239 --> 00:45:14,000
Speaker 5: Yeah, but that doesn't mean that those technologies don't have

954
00:45:14,040 --> 00:45:17,600
Speaker 5: a place. So people are talking about vogumetric capture or

955
00:45:17,600 --> 00:45:19,319
Speaker 5: GAUSSI and splatting, all of these kinds of things that

956
00:45:19,360 --> 00:45:21,520
Speaker 5: allow you to create three dimensional space and move through it.

957
00:45:22,360 --> 00:45:23,440
Speaker 4: I am working on.

958
00:45:23,520 --> 00:45:26,279
Speaker 5: I have a belief that one of the places I

959
00:45:26,280 --> 00:45:28,640
Speaker 5: can't put a camera imagine the two of us were fighting,

960
00:45:28,920 --> 00:45:31,239
Speaker 5: is right here, right in front of my face and

961
00:45:31,280 --> 00:45:33,840
Speaker 5: watching your face. And yet the explanation you want to

962
00:45:33,880 --> 00:45:36,279
Speaker 5: give of I hit Malcolm in the shoulder because he

963
00:45:36,360 --> 00:45:37,960
Speaker 5: had dropped his hand and it was open.

964
00:45:38,560 --> 00:45:40,840
Speaker 4: Show me, make it available to me.

965
00:45:41,160 --> 00:45:43,880
Speaker 5: Speed up that process fast enough, create the telestration tools

966
00:45:43,920 --> 00:45:46,160
Speaker 5: fast enough. And I think that in five years those

967
00:45:46,160 --> 00:45:49,400
Speaker 5: things are going to become rote and we will expect

968
00:45:49,600 --> 00:45:53,560
Speaker 5: impossible angles, all of these kinds of things to become

969
00:45:53,920 --> 00:45:56,200
Speaker 5: part of our world. We are going to have to

970
00:45:56,239 --> 00:45:59,160
Speaker 5: rethink what it means to watch to be a fan

971
00:45:59,200 --> 00:46:03,800
Speaker 5: of something. If you only watch a fifteen minute cutdown

972
00:46:04,040 --> 00:46:06,480
Speaker 5: of a NASCAR race, are you a fan or not

973
00:46:06,520 --> 00:46:06,839
Speaker 5: a fan?

974
00:46:07,400 --> 00:46:07,759
Speaker 4: I don't know.

975
00:46:07,800 --> 00:46:10,480
Speaker 5: Do you watch fifteen minute cutdown of every race? And

976
00:46:10,600 --> 00:46:12,640
Speaker 5: how do I service that person and how do I

977
00:46:12,719 --> 00:46:15,279
Speaker 5: engage with that person? If you do have this very

978
00:46:15,280 --> 00:46:17,200
Speaker 5: for you said earlier that people only watch a half

979
00:46:17,200 --> 00:46:19,359
Speaker 5: a show. For example, if you have a fragmented set

980
00:46:19,400 --> 00:46:21,920
Speaker 5: of attention, I can either fight it and rage against

981
00:46:21,960 --> 00:46:24,480
Speaker 5: the dying of the light, or I could lean into

982
00:46:24,480 --> 00:46:26,759
Speaker 5: it and try to service that person and make them

983
00:46:26,760 --> 00:46:28,160
Speaker 5: a real fan and bring them into the rest of

984
00:46:28,200 --> 00:46:30,279
Speaker 5: our ecosystem. And I think that that is going to

985
00:46:30,320 --> 00:46:32,520
Speaker 5: continue to happen. I think we're going to continue to

986
00:46:32,560 --> 00:46:34,840
Speaker 5: make an argument for live events, and the pull of

987
00:46:34,880 --> 00:46:38,439
Speaker 5: that one only needs to go. As I'm sure you've seen,

988
00:46:38,480 --> 00:46:42,239
Speaker 5: when you're in the room, there's nowhere you'd rather be,

989
00:46:42,480 --> 00:46:45,520
Speaker 5: and it is so interesting, even though it is so

990
00:46:45,600 --> 00:46:47,719
Speaker 5: much less good than the angles you see on television,

991
00:46:48,040 --> 00:46:50,360
Speaker 5: you can't help but look away. We're bought in on

992
00:46:50,360 --> 00:46:53,200
Speaker 5: that thesis. But the moment you step out of that room,

993
00:46:53,360 --> 00:46:55,640
Speaker 5: how do I give you the greatest experience possible?

994
00:46:56,239 --> 00:46:56,520
Speaker 2: Yeah?

995
00:46:56,640 --> 00:46:59,919
Speaker 3: Yeah, alone, Thank you very much.

996
00:47:00,200 --> 00:47:00,560
Speaker 4: Thank you.

997
00:47:01,040 --> 00:47:03,360
Speaker 3: You maybe want to become a UFC fan, which is

998
00:47:03,400 --> 00:47:05,120
Speaker 3: a statement I never thought I would ever say.

999
00:47:05,239 --> 00:47:07,279
Speaker 4: I'll tell you what, Come watch one with us, Come

1000
00:47:07,320 --> 00:47:07,960
Speaker 4: watch one with me?

1001
00:47:08,480 --> 00:47:10,719
Speaker 5: Yeah, I bet you walk away and you're like, I'm

1002
00:47:10,760 --> 00:47:12,920
Speaker 5: not sure I watching every week, But holy moly, is

1003
00:47:12,960 --> 00:47:13,480
Speaker 5: this amazing?

1004
00:47:15,280 --> 00:47:19,320
Speaker 3: Smart talks with IBM is produced by Matt Romano, Amy Gaines, McQuaid,

1005
00:47:19,719 --> 00:47:23,560
Speaker 3: Trina Menino and Jake Harper. Were edited by Lacy Roberts.

1006
00:47:23,960 --> 00:47:27,560
Speaker 3: Engineering by Nina Bird Lawrence, mastering by Sarah Buguire, music

1007
00:47:27,560 --> 00:47:33,160
Speaker 3: by Gramoscope, Strategy by Cassidy Meyer and Sophia Derlin. Smart

1008
00:47:33,200 --> 00:47:36,400
Speaker 3: Talks with IBM is a production of Pushkin Industries and

1009
00:47:36,560 --> 00:47:41,120
Speaker 3: Ruby Studio at iHeartMedia. To find more Pushkin podcasts, listen

1010
00:47:41,120 --> 00:47:45,040
Speaker 3: on the iHeartRadio app, Apple Podcasts, or wherever you listen

1011
00:47:45,080 --> 00:47:49,839
Speaker 3: to podcasts. I'm Malcolm Glappo. This is a paid advertisement

1012
00:47:49,920 --> 00:47:54,240
Speaker 3: from IBM. The conversations on this podcast don't necessarily represent

1013
00:47:54,360 --> 00:47:59,040
Speaker 3: IBM's positions, strategies, or opinions

1014
00:48:02,440 --> 00:48:04,000
Speaker 5: And STI