[Script Info]
Title: 
[Events]
Format: Layer, Start, End, Style, Name, MarginL, MarginR, MarginV, Effect, Text
Dialogue: 0,0:00:00.95,0:00:04.54,Default,,0000,0000,0000,,My colleagues and I are fascinated\Nby the science of moving dots.
Dialogue: 0,0:00:04.93,0:00:06.08,Default,,0000,0000,0000,,So what are these dots?
Dialogue: 0,0:00:06.10,0:00:07.39,Default,,0000,0000,0000,,Well, it's all of us.
Dialogue: 0,0:00:07.41,0:00:12.50,Default,,0000,0000,0000,,And we're moving in our homes,\Nin our offices, as we shop and travel
Dialogue: 0,0:00:12.52,0:00:14.59,Default,,0000,0000,0000,,throughout our cities\Nand around the world.
Dialogue: 0,0:00:14.96,0:00:18.63,Default,,0000,0000,0000,,And wouldn't it be great\Nif we could understand all this movement?
Dialogue: 0,0:00:18.92,0:00:21.81,Default,,0000,0000,0000,,If we could find patterns and meaning\Nand insight in it.
Dialogue: 0,0:00:22.26,0:00:24.04,Default,,0000,0000,0000,,And luckily for us, we live in a time
Dialogue: 0,0:00:24.07,0:00:28.56,Default,,0000,0000,0000,,where we're incredibly good\Nat capturing information about ourselves.
Dialogue: 0,0:00:28.81,0:00:32.47,Default,,0000,0000,0000,,So whether it's through\Nsensors or videos, or apps,
Dialogue: 0,0:00:32.49,0:00:35.30,Default,,0000,0000,0000,,we can track our movement\Nwith incredibly fine detail.
Dialogue: 0,0:00:36.09,0:00:41.12,Default,,0000,0000,0000,,So it turns out one of the places\Nwhere we have the best data about movement
Dialogue: 0,0:00:41.15,0:00:42.36,Default,,0000,0000,0000,,is sports.
Dialogue: 0,0:00:42.68,0:00:48.02,Default,,0000,0000,0000,,So whether it's basketball or baseball,\Nor football or the other football,
Dialogue: 0,0:00:48.04,0:00:52.44,Default,,0000,0000,0000,,we're instrumenting our stadiums\Nand our players to track their movements
Dialogue: 0,0:00:52.46,0:00:53.78,Default,,0000,0000,0000,,every fraction of a second.
Dialogue: 0,0:00:53.80,0:00:58.18,Default,,0000,0000,0000,,So what we're doing\Nis turning our athletes into --
Dialogue: 0,0:00:58.21,0:01:00.17,Default,,0000,0000,0000,,you probably guessed it --
Dialogue: 0,0:01:00.19,0:01:01.59,Default,,0000,0000,0000,,moving dots.
Dialogue: 0,0:01:01.95,0:01:06.88,Default,,0000,0000,0000,,So we've got mountains of moving dots\Nand like most raw data,
Dialogue: 0,0:01:06.90,0:01:09.41,Default,,0000,0000,0000,,it's hard to deal with\Nand not that interesting.
Dialogue: 0,0:01:09.43,0:01:13.20,Default,,0000,0000,0000,,But there are things that, for example,\Nbasketball coaches want to know.
Dialogue: 0,0:01:13.22,0:01:17.03,Default,,0000,0000,0000,,And the problem is they can't know them\Nbecause they'd have to watch every second
Dialogue: 0,0:01:17.06,0:01:19.65,Default,,0000,0000,0000,,of every game, remember it and process it.
Dialogue: 0,0:01:19.80,0:01:21.73,Default,,0000,0000,0000,,And a person can't do that,
Dialogue: 0,0:01:21.76,0:01:23.07,Default,,0000,0000,0000,,but a machine can.
Dialogue: 0,0:01:23.66,0:01:27.07,Default,,0000,0000,0000,,The problem is a machine can't see\Nthe game with the eye of a coach.
Dialogue: 0,0:01:27.36,0:01:29.62,Default,,0000,0000,0000,,At least they couldn't until now.
Dialogue: 0,0:01:30.23,0:01:32.33,Default,,0000,0000,0000,,So what have we taught the machine to see?
Dialogue: 0,0:01:33.57,0:01:35.36,Default,,0000,0000,0000,,So, we started simply.
Dialogue: 0,0:01:35.38,0:01:39.18,Default,,0000,0000,0000,,We taught it things like passes,\Nshots and rebounds.
Dialogue: 0,0:01:39.20,0:01:41.74,Default,,0000,0000,0000,,Things that most casual fans would know.
Dialogue: 0,0:01:41.77,0:01:44.60,Default,,0000,0000,0000,,And then we moved on to things\Nslightly more complicated.
Dialogue: 0,0:01:44.62,0:01:49.21,Default,,0000,0000,0000,,Events like post-ups,\Nand pick-and-rolls, and isolations.
Dialogue: 0,0:01:49.38,0:01:52.92,Default,,0000,0000,0000,,And if you don't know them, that's okay.\NMost casual players probably do.
Dialogue: 0,0:01:53.56,0:01:58.90,Default,,0000,0000,0000,,Now, we've gotten to a point where today,\Nthe machine understands complex events
Dialogue: 0,0:01:58.92,0:02:01.100,Default,,0000,0000,0000,,like down screens and wide pins.
Dialogue: 0,0:02:02.02,0:02:04.75,Default,,0000,0000,0000,,Basically things only professionals know.
Dialogue: 0,0:02:04.77,0:02:09.16,Default,,0000,0000,0000,,So we have taught a machine to see\Nwith the eyes of a coach.
Dialogue: 0,0:02:10.01,0:02:11.87,Default,,0000,0000,0000,,So how have we been able to do this?
Dialogue: 0,0:02:12.51,0:02:15.63,Default,,0000,0000,0000,,If I asked a coach to describe\Nsomething like a pick-and-roll,
Dialogue: 0,0:02:15.65,0:02:17.29,Default,,0000,0000,0000,,they would give me a description,
Dialogue: 0,0:02:17.32,0:02:20.17,Default,,0000,0000,0000,,and if I encoded that as an algorithm,\Nit would be terrible.
Dialogue: 0,0:02:21.03,0:02:25.30,Default,,0000,0000,0000,,The pick-and-roll happens to be this dance\Nin basketball between four players,
Dialogue: 0,0:02:25.33,0:02:27.24,Default,,0000,0000,0000,,two on offense and two on defense.
Dialogue: 0,0:02:27.49,0:02:29.10,Default,,0000,0000,0000,,And here's kind of how it goes.
Dialogue: 0,0:02:29.13,0:02:31.66,Default,,0000,0000,0000,,So there's the guy on offense\Nwithout the ball
Dialogue: 0,0:02:31.68,0:02:34.89,Default,,0000,0000,0000,,the ball and he goes next to the guy\Nguarding the guy with the ball,
Dialogue: 0,0:02:34.92,0:02:36.18,Default,,0000,0000,0000,,and he kind of stays there
Dialogue: 0,0:02:36.20,0:02:39.52,Default,,0000,0000,0000,,and they both move and stuff happens,\Nand ta-da, it's a pick-and-roll.
Dialogue: 0,0:02:39.54,0:02:41.76,Default,,0000,0000,0000,,(Laughter)
Dialogue: 0,0:02:41.78,0:02:44.29,Default,,0000,0000,0000,,So that is also an example\Nof a terrible algorithm.
Dialogue: 0,0:02:44.91,0:02:49.12,Default,,0000,0000,0000,,So, if the player who's the interferer --\Nhe's called the screener --
Dialogue: 0,0:02:49.28,0:02:52.15,Default,,0000,0000,0000,,goes close by, but he doesn't stop,
Dialogue: 0,0:02:52.17,0:02:53.94,Default,,0000,0000,0000,,it's probably not a pick-and-roll.
Dialogue: 0,0:02:54.56,0:02:58.50,Default,,0000,0000,0000,,Or if he does stop,\Nbut he doesn't stop close enough,
Dialogue: 0,0:02:58.53,0:03:00.29,Default,,0000,0000,0000,,it's probably not a pick-and-roll.
Dialogue: 0,0:03:00.64,0:03:03.88,Default,,0000,0000,0000,,Or, if he does go close by\Nand he does stop
Dialogue: 0,0:03:03.90,0:03:07.23,Default,,0000,0000,0000,,but they do it under the basket,\Nit's probably not a pick-and-roll.
Dialogue: 0,0:03:07.46,0:03:09.99,Default,,0000,0000,0000,,Or I could be wrong,\Nthey could all be pick-and-rolls.
Dialogue: 0,0:03:10.01,0:03:14.58,Default,,0000,0000,0000,,It really depends on the exact timing,\Nthe distances, the locations,
Dialogue: 0,0:03:14.60,0:03:16.10,Default,,0000,0000,0000,,and that's what makes it hard.
Dialogue: 0,0:03:16.58,0:03:21.52,Default,,0000,0000,0000,,So, luckily, with machine learning,\Nwe can go beyond our own ability
Dialogue: 0,0:03:21.55,0:03:23.29,Default,,0000,0000,0000,,to describe the things we know.
Dialogue: 0,0:03:23.31,0:03:25.59,Default,,0000,0000,0000,,So how does this work?\NWell, it's by example.
Dialogue: 0,0:03:25.76,0:03:28.59,Default,,0000,0000,0000,,So we go to the machine and say,\N"Good morning, machine.
Dialogue: 0,0:03:29.08,0:03:32.44,Default,,0000,0000,0000,,Here are some pick-and-rolls,\Nand here are some things that are not.
Dialogue: 0,0:03:32.72,0:03:34.97,Default,,0000,0000,0000,,Please find a way to tell the difference."
Dialogue: 0,0:03:35.08,0:03:38.78,Default,,0000,0000,0000,,And the key to all of this is to find\Nfeatures that enable it to separate.
Dialogue: 0,0:03:38.81,0:03:40.92,Default,,0000,0000,0000,,So if I was going\Nto teach it the difference
Dialogue: 0,0:03:40.94,0:03:42.32,Default,,0000,0000,0000,,between an apple and orange,
Dialogue: 0,0:03:42.34,0:03:44.72,Default,,0000,0000,0000,,I might say, "Why don't you\Nuse color or shape?"
Dialogue: 0,0:03:44.74,0:03:47.69,Default,,0000,0000,0000,,And the problem that we're solving is,\Nwhat are those things?
Dialogue: 0,0:03:47.71,0:03:48.96,Default,,0000,0000,0000,,What are the key features
Dialogue: 0,0:03:48.98,0:03:52.48,Default,,0000,0000,0000,,that let a computer navigate\Nthe world of moving dots?
Dialogue: 0,0:03:52.50,0:03:57.33,Default,,0000,0000,0000,,So figuring out all these relationships\Nwith relative and absolute location,
Dialogue: 0,0:03:57.35,0:03:59.26,Default,,0000,0000,0000,,distance, timing, velocities --
Dialogue: 0,0:03:59.44,0:04:04.37,Default,,0000,0000,0000,,that's really the key to the science\Nof moving dots, or as we like to call it,
Dialogue: 0,0:04:04.39,0:04:07.74,Default,,0000,0000,0000,,spatiotemporal pattern recognition,\Nin academic vernacular.
Dialogue: 0,0:04:07.92,0:04:10.82,Default,,0000,0000,0000,,Because the first thing is,\Nyou have to make it sound hard --
Dialogue: 0,0:04:10.85,0:04:12.12,Default,,0000,0000,0000,,because it is.
Dialogue: 0,0:04:12.41,0:04:15.55,Default,,0000,0000,0000,,The key thing is, for NBA coaches,\Nit's not that they want to know
Dialogue: 0,0:04:15.58,0:04:17.50,Default,,0000,0000,0000,,whether a pick-and-roll happened or not.
Dialogue: 0,0:04:17.52,0:04:19.60,Default,,0000,0000,0000,,It's that they want to know\Nhow it happened.
Dialogue: 0,0:04:19.62,0:04:22.61,Default,,0000,0000,0000,,And why is it so important to them?\NSo here's a little insight.
Dialogue: 0,0:04:22.63,0:04:24.40,Default,,0000,0000,0000,,It turns out in modern basketball,
Dialogue: 0,0:04:24.43,0:04:26.96,Default,,0000,0000,0000,,this pick-and-roll is perhaps\Nthe most important play.
Dialogue: 0,0:04:27.06,0:04:29.68,Default,,0000,0000,0000,,And knowing how to run it,\Nand knowing how to defend it,
Dialogue: 0,0:04:29.71,0:04:32.38,Default,,0000,0000,0000,,is basically a key to winning\Nand losing most games.
Dialogue: 0,0:04:32.40,0:04:36.20,Default,,0000,0000,0000,,So it turns out that this dance\Nhas a great many variations
Dialogue: 0,0:04:36.23,0:04:39.88,Default,,0000,0000,0000,,and identifying the variations\Nis really the thing that matters,
Dialogue: 0,0:04:39.90,0:04:42.43,Default,,0000,0000,0000,,and that's why we need this\Nto be really, really good.
Dialogue: 0,0:04:43.23,0:04:44.40,Default,,0000,0000,0000,,So, here's an example.
Dialogue: 0,0:04:44.43,0:04:46.81,Default,,0000,0000,0000,,There are two offensive\Nand two defensive players,
Dialogue: 0,0:04:46.83,0:04:48.98,Default,,0000,0000,0000,,getting ready to do\Nthe pick-and-roll dance.
Dialogue: 0,0:04:49.01,0:04:51.69,Default,,0000,0000,0000,,So the guy with ball\Ncan either take, or he can reject.
Dialogue: 0,0:04:52.09,0:04:55.09,Default,,0000,0000,0000,,His teammate can either roll or pop.
Dialogue: 0,0:04:55.11,0:04:58.10,Default,,0000,0000,0000,,The guy guarding the ball\Ncan either go over or under.
Dialogue: 0,0:04:58.12,0:05:02.69,Default,,0000,0000,0000,,His teammate can either show\Nor play up to touch, or play soft
Dialogue: 0,0:05:02.71,0:05:05.33,Default,,0000,0000,0000,,and together they can\Neither switch or blitz
Dialogue: 0,0:05:05.35,0:05:08.01,Default,,0000,0000,0000,,and I didn't know\Nmost of these things when I started
Dialogue: 0,0:05:08.04,0:05:11.96,Default,,0000,0000,0000,,and it would be lovely if everybody moved\Naccording to those arrows.
Dialogue: 0,0:05:11.98,0:05:15.88,Default,,0000,0000,0000,,It would make our lives a lot easier,\Nbut it turns out movement is very messy.
Dialogue: 0,0:05:16.05,0:05:21.53,Default,,0000,0000,0000,,People wiggle a lot and getting\Nthese variations identified
Dialogue: 0,0:05:21.56,0:05:22.86,Default,,0000,0000,0000,,with very high accuracy,
Dialogue: 0,0:05:22.88,0:05:24.75,Default,,0000,0000,0000,,both in precision and recall, is tough
Dialogue: 0,0:05:24.77,0:05:28.39,Default,,0000,0000,0000,,because that's what it takes to get\Na professional coach to believe in you.
Dialogue: 0,0:05:28.42,0:05:31.80,Default,,0000,0000,0000,,And despite all the difficulties\Nwith the right spatiotemporal features
Dialogue: 0,0:05:31.82,0:05:33.29,Default,,0000,0000,0000,,we have been able to do that.
Dialogue: 0,0:05:33.32,0:05:37.24,Default,,0000,0000,0000,,Coaches trust our ability of our machine\Nto identify these variations.
Dialogue: 0,0:05:37.48,0:05:41.01,Default,,0000,0000,0000,,We're at the point where\Nalmost every single contender
Dialogue: 0,0:05:41.04,0:05:42.66,Default,,0000,0000,0000,,for an NBA championship this year
Dialogue: 0,0:05:42.68,0:05:47.09,Default,,0000,0000,0000,,is using our software, which is built\Non a machine that understands
Dialogue: 0,0:05:47.11,0:05:48.75,Default,,0000,0000,0000,,the moving dots of basketball.
Dialogue: 0,0:05:49.87,0:05:55.02,Default,,0000,0000,0000,,So not only that, we have given advice\Nthat has changed strategies
Dialogue: 0,0:05:55.05,0:05:58.40,Default,,0000,0000,0000,,that have helped teams win\Nvery important games,
Dialogue: 0,0:05:58.42,0:06:02.16,Default,,0000,0000,0000,,and it's very exciting because you have\Ncoaches who've been in the league
Dialogue: 0,0:06:02.18,0:06:05.25,Default,,0000,0000,0000,,for 30 years that are willing to take\Nadvice from a machine.
Dialogue: 0,0:06:05.87,0:06:08.78,Default,,0000,0000,0000,,And it's very exciting,\Nit's much more than the pick-and-roll.
Dialogue: 0,0:06:08.80,0:06:10.88,Default,,0000,0000,0000,,Our computer started out\Nwith simple things
Dialogue: 0,0:06:10.90,0:06:12.97,Default,,0000,0000,0000,,and learned more and more complex things
Dialogue: 0,0:06:12.99,0:06:14.55,Default,,0000,0000,0000,,and now it knows so many things.
Dialogue: 0,0:06:14.58,0:06:17.41,Default,,0000,0000,0000,,Frankly, I don't understand\Nmuch of what it does,
Dialogue: 0,0:06:17.44,0:06:21.15,Default,,0000,0000,0000,,and while it's not that special\Nto be smarter than me,
Dialogue: 0,0:06:21.18,0:06:24.82,Default,,0000,0000,0000,,we were wondering,\Ncan a machine know more than a coach?
Dialogue: 0,0:06:24.84,0:06:26.90,Default,,0000,0000,0000,,Can it know more than person could know?
Dialogue: 0,0:06:26.92,0:06:28.67,Default,,0000,0000,0000,,And it turns out the answer is yes.
Dialogue: 0,0:06:28.69,0:06:31.25,Default,,0000,0000,0000,,The coaches want players\Nto take good shots.
Dialogue: 0,0:06:31.27,0:06:32.92,Default,,0000,0000,0000,,So if I'm standing near the basket
Dialogue: 0,0:06:32.95,0:06:35.11,Default,,0000,0000,0000,,and there's nobody near me,\Nit's a good shot.
Dialogue: 0,0:06:35.14,0:06:39.08,Default,,0000,0000,0000,,If I'm standing far away surrounded\Nby defenders, that's generally a bad shot.
Dialogue: 0,0:06:39.10,0:06:43.98,Default,,0000,0000,0000,,But we never knew how good "good" was,\Nor how bad "bad" was quantitatively.
Dialogue: 0,0:06:44.21,0:06:45.36,Default,,0000,0000,0000,,Until now.
Dialogue: 0,0:06:45.77,0:06:48.83,Default,,0000,0000,0000,,So what we can do, again,\Nusing spatiotemporal features,
Dialogue: 0,0:06:48.85,0:06:50.23,Default,,0000,0000,0000,,we looked at every shot.
Dialogue: 0,0:06:50.25,0:06:53.26,Default,,0000,0000,0000,,We can see: Where is the shot?\NWhat's the angle to the basket?
Dialogue: 0,0:06:53.28,0:06:56.04,Default,,0000,0000,0000,,Where are the defenders standing?\NWhat are their distances?
Dialogue: 0,0:06:56.07,0:06:57.40,Default,,0000,0000,0000,,What are their angles?
Dialogue: 0,0:06:57.42,0:07:00.40,Default,,0000,0000,0000,,For multiple defenders, we can look\Nat how the player's moving
Dialogue: 0,0:07:00.42,0:07:01.86,Default,,0000,0000,0000,,and predict the shot type.
Dialogue: 0,0:07:01.88,0:07:05.95,Default,,0000,0000,0000,,We can look at all their velocities\Nand we can build a model that predicts
Dialogue: 0,0:07:05.98,0:07:10.03,Default,,0000,0000,0000,,what is the likelihood that this shot\Nwould go in under these circumstances?
Dialogue: 0,0:07:10.19,0:07:11.69,Default,,0000,0000,0000,,So why is this important?
Dialogue: 0,0:07:12.10,0:07:14.90,Default,,0000,0000,0000,,We can take something that was shooting,
Dialogue: 0,0:07:14.93,0:07:17.61,Default,,0000,0000,0000,,which was one thing before,\Nand turn it into two things:
Dialogue: 0,0:07:17.63,0:07:20.28,Default,,0000,0000,0000,,the quality of the shot\Nand the quality of the shooter.
Dialogue: 0,0:07:21.68,0:07:24.94,Default,,0000,0000,0000,,So here's a bubble chart,\Nbecause what's TED without a bubble chart?
Dialogue: 0,0:07:24.97,0:07:25.98,Default,,0000,0000,0000,,(Laughter)
Dialogue: 0,0:07:26.00,0:07:27.32,Default,,0000,0000,0000,,Those are NBA players.
Dialogue: 0,0:07:27.34,0:07:30.46,Default,,0000,0000,0000,,The size is the size of the player\Nand the color is the position.
Dialogue: 0,0:07:30.48,0:07:32.62,Default,,0000,0000,0000,,On the x-axis,\Nwe have the shot probability.
Dialogue: 0,0:07:32.64,0:07:34.59,Default,,0000,0000,0000,,People on the left take difficult shots,
Dialogue: 0,0:07:34.62,0:07:36.84,Default,,0000,0000,0000,,on the right, they take easy shots.
Dialogue: 0,0:07:37.19,0:07:39.25,Default,,0000,0000,0000,,On the [y-axis] is their shooting ability.
Dialogue: 0,0:07:39.28,0:07:41.84,Default,,0000,0000,0000,,People who are good are at the top,\Nbad at the bottom.
Dialogue: 0,0:07:41.86,0:07:43.62,Default,,0000,0000,0000,,So for example, if there was a player
Dialogue: 0,0:07:43.62,0:07:45.72,Default,,0000,0000,0000,,who generally made\N47 percent of their shots,
Dialogue: 0,0:07:45.72,0:07:47.11,Default,,0000,0000,0000,,that's all you knew before.
Dialogue: 0,0:07:47.34,0:07:52.20,Default,,0000,0000,0000,,But today, I can tell you that player\Ntakes shots that an average NBA player
Dialogue: 0,0:07:52.22,0:07:54.18,Default,,0000,0000,0000,,would make 49 percent of the time,
Dialogue: 0,0:07:54.20,0:07:55.89,Default,,0000,0000,0000,,and they are two percent worse.
Dialogue: 0,0:07:56.27,0:08:00.78,Default,,0000,0000,0000,,And the reason that's important\Nis that there are lots of 47s out there.
Dialogue: 0,0:08:01.71,0:08:04.26,Default,,0000,0000,0000,,And so it's really important to know
Dialogue: 0,0:08:04.29,0:08:08.24,Default,,0000,0000,0000,,if the 47 that you're considering\Ngiving 100 million dollars to
Dialogue: 0,0:08:08.27,0:08:11.32,Default,,0000,0000,0000,,is a good shooter who takes bad shots
Dialogue: 0,0:08:11.35,0:08:13.74,Default,,0000,0000,0000,,or a bad shooter who takes good shots.
Dialogue: 0,0:08:15.13,0:08:18.46,Default,,0000,0000,0000,,Machine understanding doesn't just change\Nhow we look at players,
Dialogue: 0,0:08:18.49,0:08:20.34,Default,,0000,0000,0000,,it changes how we look at the game.
Dialogue: 0,0:08:20.37,0:08:24.12,Default,,0000,0000,0000,,So there was this very exciting game\Na couple of years ago, in the NBA finals.
Dialogue: 0,0:08:24.15,0:08:27.36,Default,,0000,0000,0000,,Miami was down by three,\Nthere was 20 seconds left.
Dialogue: 0,0:08:27.38,0:08:29.40,Default,,0000,0000,0000,,They were about to lose the championship.
Dialogue: 0,0:08:29.43,0:08:32.77,Default,,0000,0000,0000,,A gentleman named LeBron James\Ncame up and he took a three to tie.
Dialogue: 0,0:08:32.79,0:08:33.99,Default,,0000,0000,0000,,He missed.
Dialogue: 0,0:08:34.02,0:08:35.85,Default,,0000,0000,0000,,His teammate Chris Bosh got a rebound,
Dialogue: 0,0:08:35.88,0:08:38.04,Default,,0000,0000,0000,,passed it to another teammate\Nnamed Ray Allen.
Dialogue: 0,0:08:38.06,0:08:39.98,Default,,0000,0000,0000,,He sank a three. It went into overtime.
Dialogue: 0,0:08:40.00,0:08:42.10,Default,,0000,0000,0000,,They won the game.\NThey won the championship.
Dialogue: 0,0:08:42.12,0:08:44.57,Default,,0000,0000,0000,,It was one of the most exciting\Ngames in basketball.
Dialogue: 0,0:08:45.44,0:08:48.87,Default,,0000,0000,0000,,And our ability to know\Nthe shot probability for every player
Dialogue: 0,0:08:48.89,0:08:50.08,Default,,0000,0000,0000,,at every second,
Dialogue: 0,0:08:50.10,0:08:53.06,Default,,0000,0000,0000,,and the likelihood of them getting\Na rebound at every second
Dialogue: 0,0:08:53.08,0:08:56.53,Default,,0000,0000,0000,,can illuminate this moment in a way\Nthat we never could before.
Dialogue: 0,0:08:57.62,0:09:00.29,Default,,0000,0000,0000,,Now unfortunately,\NI can't show you that video.
Dialogue: 0,0:09:00.31,0:09:04.80,Default,,0000,0000,0000,,But for you, we recreated that moment
Dialogue: 0,0:09:04.83,0:09:07.16,Default,,0000,0000,0000,,at our weekly basketball game\Nabout 3 weeks ago.
Dialogue: 0,0:09:07.28,0:09:09.45,Default,,0000,0000,0000,,(Laughter)
Dialogue: 0,0:09:09.57,0:09:12.98,Default,,0000,0000,0000,,And we recreated the tracking\Nthat led to the insights.
Dialogue: 0,0:09:13.20,0:09:17.45,Default,,0000,0000,0000,,So, here is us.\NThis is Chinatown in Los Angeles,
Dialogue: 0,0:09:17.48,0:09:19.04,Default,,0000,0000,0000,,a park we play at every week,
Dialogue: 0,0:09:19.07,0:09:21.30,Default,,0000,0000,0000,,and that's us recreating\Nthe Ray Allen moment
Dialogue: 0,0:09:21.32,0:09:23.55,Default,,0000,0000,0000,,and all the tracking\Nthat's associated with it.
Dialogue: 0,0:09:24.77,0:09:26.29,Default,,0000,0000,0000,,So, here's the shot.
Dialogue: 0,0:09:26.31,0:09:28.83,Default,,0000,0000,0000,,I'm going to show you that moment
Dialogue: 0,0:09:28.85,0:09:31.44,Default,,0000,0000,0000,,and all the insights of that moment.
Dialogue: 0,0:09:31.46,0:09:35.19,Default,,0000,0000,0000,,The only difference is, instead\Nof the professional players, it's us,
Dialogue: 0,0:09:35.22,0:09:37.84,Default,,0000,0000,0000,,and instead of a professional\Nannouncer, it's me.
Dialogue: 0,0:09:37.86,0:09:39.34,Default,,0000,0000,0000,,So, bear with me.
Dialogue: 0,0:09:41.15,0:09:42.30,Default,,0000,0000,0000,,Miami.
Dialogue: 0,0:09:42.67,0:09:43.82,Default,,0000,0000,0000,,Down three.
Dialogue: 0,0:09:44.11,0:09:45.26,Default,,0000,0000,0000,,Twenty seconds left.
Dialogue: 0,0:09:47.38,0:09:48.58,Default,,0000,0000,0000,,Jeff brings up the ball.
Dialogue: 0,0:09:50.66,0:09:52.19,Default,,0000,0000,0000,,Josh catches, puts up a three!
Dialogue: 0,0:09:52.63,0:09:54.48,Default,,0000,0000,0000,,[Calculating shot probability]
Dialogue: 0,0:09:55.28,0:09:56.43,Default,,0000,0000,0000,,[Shot quality]
Dialogue: 0,0:09:57.05,0:09:58.83,Default,,0000,0000,0000,,[Rebound probability]
Dialogue: 0,0:10:00.37,0:10:01.55,Default,,0000,0000,0000,,Won't go!
Dialogue: 0,0:10:01.57,0:10:03.02,Default,,0000,0000,0000,,[Rebound probability]
Dialogue: 0,0:10:03.78,0:10:05.03,Default,,0000,0000,0000,,Rebound, Noel.
Dialogue: 0,0:10:05.06,0:10:06.21,Default,,0000,0000,0000,,Back to Daria.
Dialogue: 0,0:10:06.51,0:10:09.87,Default,,0000,0000,0000,,[Shot quality]
Dialogue: 0,0:10:10.68,0:10:12.30,Default,,0000,0000,0000,,Her three-pointer -- bang!
Dialogue: 0,0:10:12.32,0:10:14.52,Default,,0000,0000,0000,,Tie game with five seconds left.
Dialogue: 0,0:10:14.88,0:10:16.50,Default,,0000,0000,0000,,The crowd goes wild.
Dialogue: 0,0:10:16.52,0:10:18.18,Default,,0000,0000,0000,,(Laughter)
Dialogue: 0,0:10:18.20,0:10:19.75,Default,,0000,0000,0000,,That's roughly how it happened.
Dialogue: 0,0:10:19.78,0:10:20.93,Default,,0000,0000,0000,,(Applause)
Dialogue: 0,0:10:20.95,0:10:22.13,Default,,0000,0000,0000,,Roughly.
Dialogue: 0,0:10:22.15,0:10:23.68,Default,,0000,0000,0000,,(Applause)
Dialogue: 0,0:10:24.12,0:10:29.60,Default,,0000,0000,0000,,That moment had about a nine percent\Nchance of happening in the NBA
Dialogue: 0,0:10:29.63,0:10:31.89,Default,,0000,0000,0000,,and we know that\Nand a great many other things.
Dialogue: 0,0:10:31.91,0:10:35.40,Default,,0000,0000,0000,,I'm not going to tell you how many times\Nit took us to make that happen.
Dialogue: 0,0:10:35.43,0:10:37.18,Default,,0000,0000,0000,,(Laughter)
Dialogue: 0,0:10:37.20,0:10:39.07,Default,,0000,0000,0000,,Okay, I will! It was four.
Dialogue: 0,0:10:39.10,0:10:40.10,Default,,0000,0000,0000,,(Laughter)
Dialogue: 0,0:10:40.12,0:10:41.29,Default,,0000,0000,0000,,Way to go, Daria.
Dialogue: 0,0:10:41.65,0:10:45.91,Default,,0000,0000,0000,,But the important thing about that video
Dialogue: 0,0:10:45.93,0:10:50.50,Default,,0000,0000,0000,,and the insights we have for every second\Nof every NBA game -- it's not that.
Dialogue: 0,0:10:50.64,0:10:54.57,Default,,0000,0000,0000,,It's the fact you don't have to be\Na professional team to track movement.
Dialogue: 0,0:10:55.08,0:10:58.74,Default,,0000,0000,0000,,You do not have to be a professional\Nplayer to get insights about movement.
Dialogue: 0,0:10:58.76,0:11:02.62,Default,,0000,0000,0000,,In fact, it doesn't even have to be about\Nsports because we're moving everywhere.
Dialogue: 0,0:11:03.65,0:11:06.02,Default,,0000,0000,0000,,We're moving in our homes,
Dialogue: 0,0:11:09.43,0:11:10.63,Default,,0000,0000,0000,,in our offices,
Dialogue: 0,0:11:12.24,0:11:14.93,Default,,0000,0000,0000,,as we shop and we travel
Dialogue: 0,0:11:17.32,0:11:18.57,Default,,0000,0000,0000,,throughout our cities
Dialogue: 0,0:11:20.06,0:11:21.68,Default,,0000,0000,0000,,and around our world.
Dialogue: 0,0:11:23.27,0:11:25.56,Default,,0000,0000,0000,,What will we know? What will we learn?
Dialogue: 0,0:11:25.59,0:11:27.89,Default,,0000,0000,0000,,Perhaps, instead of identifying\Npick-and-rolls,
Dialogue: 0,0:11:27.92,0:11:30.93,Default,,0000,0000,0000,,a machine can identify\Nthe moment and let me know
Dialogue: 0,0:11:30.95,0:11:33.01,Default,,0000,0000,0000,,when my daughter takes her first steps.
Dialogue: 0,0:11:33.04,0:11:35.57,Default,,0000,0000,0000,,Which could literally be happening\Nany second now.
Dialogue: 0,0:11:36.14,0:11:39.84,Default,,0000,0000,0000,,Perhaps we can learn to better use\Nour buildings, better plan our cities.
Dialogue: 0,0:11:40.36,0:11:44.54,Default,,0000,0000,0000,,I believe that with the development\Nof the science of moving dots,
Dialogue: 0,0:11:44.56,0:11:48.20,Default,,0000,0000,0000,,we will move better, we will move smarter,\Nwe will move forward.
Dialogue: 0,0:11:48.61,0:11:49.80,Default,,0000,0000,0000,,Thank you very much.
Dialogue: 0,0:11:49.82,0:11:54.86,Default,,0000,0000,0000,,(Applause)