Pod Insights StarTalk Podcast Cosmic Queries – Rise of the Machines with Matt Ginsberg About this episode Can machine learning predict the outcome of basketball games and March Madness? On this episode of StarTalk Sports Edition, Neil deGrasse Tyson, Gary O’Reilly, and Chuck Nice talk machine learning with computer scientist and author of The Factor Man, Matt Ginsberg.
Transcript Speaker names marked ? are inferred from speaking time and the show’s known hosts — they are not voice-verified.
0:01 Neil deGrasse Tyson? Welcome to StarTalk. Your place in the universe where science and pop culture collide. StarTalk begins right now. This is StarTalk, Sports Edition. A kind of a cosmic queries on AI, sports, March Madness, and what happens if we take sports to other planets? I got Chuck with me, Chuck. Hey Neil. All right, and the person who gives authenticity to the entire concept of Sports Edition, Gary O'Reilly. Gary? Hey Neil. All right, an old footballer from the UK. I was browsing the internet and I found a Wiki page on you.
0:44 Neil deGrasse Tyson? Right there you were on the soccer field and you were looking, looking buff and with- He's got great legs. Sexy legs. Sexy legs, baby. Sexy legs. So, interested in Gary, you can dig him up on the internet. So, we don't have particular expertise in March Madness, although we've done a lot of thinking about it. So we had to go to a man about town who thinks about this kind of stuff, and that's Matt Ginsberg. Matt, welcome back to StarTalk.
1:13 Speaker 01 Thank you, it's great to be back.
1:15 Neil deGrasse Tyson? Yeah, so Matt is like, was a wonder kind. Was getting his PhD in math at age 24 from Oxford. Oxford, excuse me. God, you're such a slacker. Oxford, okay.
1:28 Speaker 01 Well, I took a year off at Caltech.
1:31 Neil deGrasse Tyson? Okay, and researched mathematical physics. My wife's specialty is mathematical physics. You're a scientist and entrepreneur and author. You've got a book out there, a fictional book, The Factor Man. I like fiction that is deeply informed by math and physics and the universe, and that's one such book where you introduced God's algorithm, the secret formula that will solve all the problems of the world and who's, and the fight to take command of it. I love the, maybe we'll talk about it later. Love where that's going. Also, one of your companies- One of? Excuse me, excuse me. I think those are called, what are they called? It's a humble brag. Humble brag. When you hear someone say, "Well, one of my companies."
2:16 Neil deGrasse Tyson? Not all of the companies. Right. Just one of them.
2:20 Speaker 01 No, it's just, I get, I get bored. So I start a company and it works, and then it's like, "Oh, I gotta go do something else," and I start another company. That's why I have this- You know, trail of companies.
2:31 Neil deGrasse Tyson? So, so the, the, the.
2:33 Chuck Nice? Another humble brag.
2:36 Neil deGrasse Tyson? Yeah, Matt, just shut up there. Just, just stop there.
2:40 Chuck Nice? Like, you know what? Some people, when they get bored, they turn on television. When I get bored, I start a company.
2:51 Neil deGrasse Tyson? I get bored, we're the potato chips.
2:56 Neil deGrasse Tyson? Wait, so you provide statistical support for professional sports teams. In fact, we delved into that topic the last time you were a guest. I'll invite people to dig up that episode. And what I learned in that episode is you wrote a computer program that can enter crossword puzzle competitions. And it's called, get this, Dr. Phil.
3:16 Chuck Nice? Nice.
3:17 Neil deGrasse Tyson? F I L L. Very, very clever. So anyhow, Matt, let's get, let's get straight to March Madness. There are brackets, you know, teams win, it's winner go home. Is that right? It's not, it's like single elimination, last I checked. And is there any way to predict the whole thing that you know of, that you've invented, that you can share with us? And if so, how, how do, how do you use it? And what do you do? Do we all need a PhD in math to make that happen?
3:45 Chuck Nice? Can we get Warren Buffett's money?
3:48 Speaker 01 There you go.
3:49 Chuck Nice? Can we really get Warren Buffett's money?
3:53 Neil deGrasse Tyson? Get right down to the basic question.
3:55 Speaker 01 So, my understanding is that in order to get Warren Buffett's money, you have to predict every single game in March Madness.
4:03 Chuck Nice? Yep.
4:04 Speaker 01 And so I actually looked at this. I knew we were going to talk about this. I looked at this last night. And I think people look at this as, you know, what are my chances of winning? And the answer is they're basically zero. And I looked at it the other way around. And I said, let's say.
4:20 Neil deGrasse Tyson? But the chance of winning would be the chance of getting every single game predicted correctly. Okay.
4:25 Speaker 01 So if you're, if you're even money on any particular game, then you have, you know, you have one chance, one time in two to the 64, will you actually get them all right? And two to 64 is some giant number, so you have no chance. But computers are helping us make better predictions. So I thought, okay, what if I wanted to have a one in a thousand shot? Which it's still not terribly good, but it means that if I get to try 20 times, then I have a one in 50 chance of actually getting it sometime in the next 20 years, which is sort of as far out as I can look. And it turns out that in order to have a one in a thousand shot at winning Warren Buffett's money,
5:06 Speaker 01 you have to be able to predict with 90% accuracy, who's going to win any particular game. And some of the games is pretty easy, right? When a one seed plays a 16 seed and you bet on the one seed, it's probably about 90%. So that's good. The problem is when the seven seed plays the eight seed.
5:26 Neil deGrasse Tyson? Yeah.
5:26 Speaker 01 That game is really a toss up. And getting to the point that you can predict that game with 90% accuracy is really hard. There are computer methods that will help. Those computer methods, one of the nice things about them is they'll actually tell you how, how sure they are. So you can put in all this information about two teams and say they're playing.
5:47 Neil deGrasse Tyson? Well, could you get to put in, how are the players feeling today? I mean, isn't that a factor that you're not considering here?
5:54 Speaker 01 It is. And you can put in, obviously, I, I mean, you could call them and ask. I don't know that that would actually help.
6:05 Neil deGrasse Tyson? Um. It would help them, you know?
6:08 Speaker 01 You can, you can put in everything.
6:10 Neil deGrasse Tyson? Matt, just want to know how you're doing today.
6:12 Speaker 01 There you go. And um, if they took the call.
6:16 Neil deGrasse Tyson? Did you have your coffee? Did you get your eight hours, right, asleep?
6:19 Speaker 01 Um, I think that, that a lot of the information that you need is actually buried there. So for example, one of the things that I've, you can put in whatever you want. So you can put in, for example, how many minutes of playing time did this guy have over the last three weeks?
6:36 Neil deGrasse Tyson? That's very important.
6:36 Speaker 01 And that's sort of a proxy for how tired is he?
6:39 Neil deGrasse Tyson? Yeah.
6:40 Speaker 01 And you can put in, how many minutes do you have over the last three weeks or the last three days? When was he last injured? How has he been playing recently? So all of this information is public and you can at least in theory, push it all into a machine learning system. And out will come, I think this team is going to win and the probability is whatever.
7:05 Neil deGrasse Tyson? Wow.
7:06 Speaker 01 Now, how?
7:10 Speaker 01 They typically, for a problem like this, they typically use something called gradient boosting, and that doesn't really help you much.
7:19 Speaker 01 What they do is they-
7:21 Neil deGrasse Tyson? Hearing you say that gradient boosting doesn't help us much.
7:24 Chuck Nice? Does gradient boosting not help much?
7:27 Neil deGrasse Tyson? I know, which is it that doesn't help?
7:30 Speaker 01 Gradient boosting is very slick, and-
7:33 Neil deGrasse Tyson? But we need you to define it, right?
7:35 Chuck Nice? Yeah.
7:35 Speaker 01 So what you do, the gradient boosting part doesn't matter. What it does is it says, "Okay, give me tons and tons of data about March Madness, historically, going back as far as you have, just buckets of data." And then it says, "Okay, I have data on 400,000 games. I'm gonna take a thousand of those games out and not look at them. And I'm gonna set them off to the side." And then it tries to figure out what tendencies there were in the 399,000 games it has left. And then it says, "Okay, this is what I think is going on." And then you go and you get the thousand games and you bring them back in and you check, and you see, "How well did I predict?"
8:21 Chuck Nice? Right.
8:21 Speaker 01 Now, in order to get Warren Buffett's money-
8:25 Neil deGrasse Tyson? Because that's the objective of this whole program.
8:27 Speaker 01 No, it's not gonna be in a second. In order to get Warren Buffett's money, you need to predict these seven versus eight games with 90% accuracy. In order to get Vegas's money, you probably need to predict with 60% accuracy. So if you're that good at predicting who's gonna win these games, you should just go to Vegas. You shouldn't, you should not try and win Warren Buffett's money. It's a much, it's a much further reach. And it's not that much, I mean, it's a million dollars, right?
8:55 Chuck Nice? Right.
8:55 Gary O'Reilly? Yeah, but a year for the rest of your life.
8:59 Speaker 01 If you can predict, you can get that money from Vegas, and it's easier.
9:03 Chuck Nice? Right.
9:03 Speaker 01 If you can predict 60% of the time-
9:06 Chuck Nice? Because you just come back every year, and you get the money, you get the Vegas money every year.
9:10 Gary O'Reilly? But Matt, if we use that model you've just described, it talks to me about history. Now, from a player's point of view, that's great. It means nothing to me because we're not playing those teams, those players are no longer playing, we're different players. This, in my mind, that's history, it has no relevance to what is about to happen.
9:30 Speaker 01 If you look back historically, you will see that five years ago LeBron James scored way more points than average. And yes, that's history, but it certainly bears on today. I mean, he was just a better player. And you can also look and you can gather information about
9:48 Neil deGrasse Tyson? Because right now today, he sucks, right?
9:54 Speaker 01 How old is he? How do people age? How do people like LeBron age? And you can just, there is so much data. Now, what makes it hard, the reason people don't just go beat up on Vegas with this stuff is because there's so much data. In something that happens with machine learning, when you have too much data, programs start what's called overfitting, which means they look and they say, wow, whether or not the Lakers won last year was correlated with the phase of the moon.
10:27 Chuck Nice? Oh dear.
10:28 Speaker 01 That is history, and that's just luck. But if you have so much data, some of it's going to look relevant when it really isn't. And the reason machine learning is hard is because you have to somehow filter that out. You have to somehow say, which of this actually matters? And one of the reasons people have this thousand games off to the side
10:48 Chuck Nice? Right.
10:48 Speaker 01 is because then they say, wow, the Lakers in the phase of the moon, and look at the thousand games, and it's like, that didn't work.
10:55 Chuck Nice? Right.
10:55 Speaker 01 And then they go and they try again.
10:56 Chuck Nice? So you're using the data itself as part of the predictive model, kind of like what they do with climate. You put in the history of climate to show whether or not the predictive models that you're using right now are effective.
11:13 Neil deGrasse Tyson? Yeah, that's right. That's right. But so just, I want to emphasize something that you said, Matt. And you said it brilliantly and beautifully. I didn't want it to go by without sort of adding further punctuation. There's a point beyond which you have so much data. That you can find correlations that have nothing to do with cause and effect. And, and so you need enough ways to check against that to remove those correlations from the analysis. Is that a fair way to, to summarize some of what you did?
11:47 Speaker 01 It is, and, and the effect is very real. You can, you can go online and you can, you can look for weird correlations and it, you know, it's things like the number of loaves of bread eaten in each year in Denmark is incredibly well correlated with whether the US stock market goes up or down. Something, you know, they're, they're just craziness. And, and as Chuck said, it really is a matter of you have, there are words for it. So you have one set of data, it's called the training data. And that's the data you use to build your model. And then you have this other set of data, which is called the validation data, which is the data that you use to test your model and see if it actually turned out to work. It's what you can't do, what it seems like you want to do is when you train your data and you look at the validation
12:32 Speaker 01 data and it doesn't work because you were looking at the number of loaves of bread eaten in Denmark. So you make a new model that doesn't consider Denmark. And you try that on your validation data and it doesn't work. And then you do it again and again and again, until you find something that works on your validation data. Well, you've cheated now. Your validation data is sort of dirty because you've been using it over and over and over again. And your basic models think they're working with just the training data, but you're working with the validation data. So you're stuck. So you need, you need clean data and clean data is incredibly valuable. Because what happens, you have these 400,000 games, feels like a lot. And then you take 10,000 aside and say, oh, I'm going to hold them for validation. And you notice, oh, maybe that's not enough.
13:17 Speaker 01 I need to, I need another validation set. And you take 10,000 more aside and then you take 10,000 more aside. And, and once you touch data, it's dirty forever. So how you split data up between validation and training data turns out to be important and hard. And it really matters that you not look at your validation data very often.
13:40 Gary O'Reilly? So Matt, how many relevant data points would you expect someone to want to use for a decent outcome?
13:49 Speaker 01 So it depends on what you're trying to do.
13:52 Gary O'Reilly? All right.
13:52 Speaker 01 Some things you need huge data sets. So if you're trying to predict it with 90% accuracy, seven versus eight, you need at least as much data, you need everything. If you're trying to predict if a number one seed is going to beat a 16 seed, you probably get by on much less. So different problems require different amounts of data. The one rule of thumb there is-
14:12 Neil deGrasse Tyson? By the way, somebody, somebody or some committee seeded the number one team as number one and the number 16 as number 16. So they're using data to do that.
14:23 Speaker 01 Yes. And that data-
14:25 Neil deGrasse Tyson? So, so you're not, you're not approaching those two teams from a pure, you, you are already biased by the setup for that. And I remember as a kid and I would, because I didn't, I was kind of very literal and they would say, oh, uh, Oklahoma upset Texas today by beating them in a basketball game. And I say, of course they upset them because they lost.
14:46 Speaker 01 Yeah.
14:46 Neil deGrasse Tyson? And, and so I didn't, the concept of upset was not, didn't, I didn't understand that is, it's because you won beyond the expectations of some group of people who decided that you should have lost. So does that add another sort of variable of unpredictability to what you're doing?
15:05 Speaker 01 Sort of. I mean, the bottom line is that the, the seeding committee of the NC2A bracket is looking at very limited amounts of data. They're typically looking at who beat whom and perhaps by how much. And they look at the, you know, how well did those teams do? And the records of your opponents, but it's a, it's a tiny amount of data relative to everything you actually could use. Like how long had people rested? Who's injured? Were they playing? And by, on and on and on and on. And I would assume-
15:32 Neil deGrasse Tyson? Because if they, because if they were accurate, no one would ever be upset in a tournament.
15:38 Speaker 01 And the only person who would be upset would be Warren Buffett because he would keep losing. But, um-
15:47 Speaker 01 Probably the data you would use if you were trying to do this seriously is a vast superset of the data that the March Madness guys, you know, the NC2A guys actually use. There's something else that I think it's important to realize here, and this has to do with how people solve problems and how machines solve problems. So I often draw a distinction between what I like to call a 99% problem and a 49% problem. So in a 49% problem, you're trying to distinguish between a 49% probability and a 51% probability. So the stock market, if you can only invest in stocks that are 51% to go up, you will make a killing.
16:31 Speaker 01 The stock market's 50/50 and just that little edge, you'll do phenomenally. 99% means you have to just get it right. So this, my sort of standard example of a 99% problem is stoplights. If you identify 99% of stoplights correctly and just drive through the rest, because you don't realize they're there, you're gonna die. You're gonna have an accident, you're gonna die. So machine learning, the kinds of things I was talking about, turns out to be really good on the 49/51 problems and really bad on the 99/100 problems. Basketball is interesting, right? Because it's sort of in between. Stock market is on one side, traffic lights on the other side.
17:13 Speaker 01 People turn out to be pretty good at the 99/100 problems and we're not so good at the 49/51 problems.
17:20 Chuck Nice? So the answer, the answer is we're not getting that money. Chuck has been single-minded this whole.
17:27 Speaker 01 The answer is, um, you're probably not, you might. Go, go for it, right? You have a, you have a one in a zillion chance.
17:38 Chuck Nice? I don't want to dash your hopes Chuck.
17:41 Speaker 01 That would be sad.
17:42 Chuck Nice? Right, exactly. Yeah. Go ahead Chuck. Chuck, go ahead. Bet your kids 529 on it.
17:51 Neil deGrasse Tyson? Guys, we've got to take a quick break, but when we come back, this is really a Patreon Cosmic Queries for March Madness, and I want to get a few of those questions in related to that. But also, maybe we can think about sort of sports on other worlds and what that might be like when it started.
18:37 Neil deGrasse Tyson? We're back, StarTalk Sports Edition. March Madness, that's what we're talking about. I've got Chuck and Gary here, guys.
18:45 Chuck Nice? Yes.
18:45 Neil deGrasse Tyson? Yeah, we can't do this alone. We needed someone who could bring the analytics into the house. And we've got our favorite analytic guy, Matt Ginsberg. Matt, always good to have you here.
18:56 Chuck Nice? Yeah.
18:56 Neil deGrasse Tyson? On StarTalk. With his math PhD, just spilling it out whenever he's got to do it.
19:03 Chuck Nice? Hey, before we get into it, let me just ask Matt something real quick.
19:06 Neil deGrasse Tyson? What's that?
19:07 Chuck Nice? With all of your cred...
19:09 Neil deGrasse Tyson? Plus, I want to take sports off-world and see where that can, where that goes to.
19:13 Chuck Nice? Oh, cool.
19:14 Neil deGrasse Tyson? With questions that Patreon members have.
19:16 Chuck Nice? Oh, we're gonna... So, all this cred you have mathematically, do you ever, do you ever game the system? You know, with any of your programming, any of your machine learning, any of your mathematical prowess? Do you ever put that to work and just say, you know what, I'm gonna go ahead and make a little...
19:37 Neil deGrasse Tyson? Wait, Chuck, I have the answer to it. I can answer for you, Matt, here. So, either he's a miser, or the answer to that question is no. Because look at the room he's in. There's no butler, there's no staircase going up to three level mansion. The dude is just in a regular room.
20:00 Chuck Nice? I don't know.
20:01 Speaker 01 So the answer is the butler's in quarantine.
20:10 Speaker 01 It's not, it's not fun, right? I had a friend who went and played blackjack in Vegas and counting cards and was very mathematic. But at the end of it all, he said he figured out that he had made 15 cents an hour. And it's such a grind and I, I like, I'll tell you the, the best.
20:28 Neil deGrasse Tyson? It's a grind making money. It's such a grind. Oh.
20:31 Speaker 01 The best thing that ever happened to me in terms of mathematical abilities is my daughter uses me as a calculator. So I'll be walking around the house and she'll, Dad, what's 17 times 36? And occasionally we have house guests who look at her and they say, Scott, did you just use your father as a calculator? And she's like, yeah, that's what he's for. So that's great.
20:54 Chuck Nice? Well, guess what? How, how old is your daughter?
20:57 Speaker 01 Well, now she's 22.
20:58 Chuck Nice? Okay, I was going to say, cause at this point she's using you as a cash machine.
21:06 Neil deGrasse Tyson? Dad, can I have 17 times 36 worth of dollars?
21:11 Chuck Nice? That's cool. All right, well, that's a good answer.
21:14 Neil deGrasse Tyson? All right. So, so give me that Chuck, who's got the first Patreon question?
21:18 Gary O'Reilly? Um, I don't care, Gary, you want to go first? Yeah, I'll jump in. So these are Patreon patrons, our exclusive Patreon patrons. Nancy Diaz, uh, she says, here's a shot at March Madness question. How can AI take into account things like motivation, styles of play and performance under pressure or other factors challenging to quantify in determining the outcome? The example she cites is Virginia's national championship win in 2020, given their embarrassing loss in 2019. So Matt, what do we think? Is this sort of enrolled in your, in your data points that we told you?
21:53 Neil deGrasse Tyson? Yeah, can you quantify motivation?
21:55 Speaker 01 Probably. I mean, in theory you can sort of quantify anything. Motivation probably shows up. Let me, let me look at this exact example. And, you know, in the data is going to be, well, how well do teams who got to the finals do if they lost in the finals the previous year? You've got a reasonable sampling. And that's what that can, you know, and then these machine learning algorithms that just you just drop all the data in and they just churn churn it up and try and look for patterns. And if that's a pattern, it'll find it. I've looked for patterns like that and I've never found them. College athletes seem to sort of always be about as motivated
22:38 Speaker 01 as they can be because they love the game. They're trying to get into the pros. They're really working. But if there is a pattern, it should be somewhere buried in the data and you should be able to pull it out.
22:50 Chuck Nice? That's pretty cool, actually. And so what hap- what what happens when you have a team of four starting freshmen? Because there won't be any data on them.
23:02 Speaker 01 So there's there's data on a variety of things. There's there's data on four starting freshmen.
23:08 Neil deGrasse Tyson? Wait, wait, wait. You have the whole season behind them. What are you talking about?
23:11 Speaker 01 You have the season, I was about to say, you have the whole season behind you by the time you get to March Madness.
23:15 Chuck Nice? Right.
23:16 Speaker 01 Um, you have data on how does a freshman's performance in March Madness compared to a freshman's performance over the course of the season? Do freshmen choke?
23:26 Chuck Nice? Right.
23:26 Speaker 01 That's that's.
23:27 Gary O'Reilly? Exactly. That's exactly a point that you need to recognize, whether or not the this the occasion crushes the player.
23:35 Speaker 01 And one of the things that's that's cool about these machine learning algorithms like gradient boosting is you don't have to figure out what you're looking for in the data. You just pour all the data in and the algorithms figure out, oh, this is a pattern. Oh, that's a pattern. And they do the mental heavy lifting for you. Now, they're stupid, so they might find some patterns that aren't really there. And that's an issue. But that's what this validation data is for. So you, in theory, they are these very general purpose algorithms that are capable of finding the signal in the noise.
24:14 Neil deGrasse Tyson? Got it. So so it's not as though you are quantifying. The thing itself is you're looking at the statistics in the larger data set of the manifestation of that of that motivation, right? So, so in other words, you can't go up, well, maybe you can go up to the one person, see they're really jacked, they gave a really good pep speech, pep talk, the coach gave a pep talk, and now they're just gonna win. You can't put that in after the fact, right? You don't know that. You're not adjusting these, these
24:48 Speaker 01 Correct.
24:48 Neil deGrasse Tyson? The statistics in real time. You have to go in with the bet already placed.
24:55 Speaker 01 I have to decide what data I want to put in and train the thing. And then what comes out is sort of how I'm gonna make my bets.
25:04 Neil deGrasse Tyson? Right.
25:05 Speaker 01 And I'm not gonna go
25:06 Neil deGrasse Tyson? Because motivational coaches can have have their influence as well.
25:10 Speaker 01 But that will be there, right? Because you'll see, oh, this coach, players playing for this coach do a little better when the same players went and played for a different coach. So this coach must be good. I don't know if it's motivation, right? So I could imagine a world where for some strange reason, on your 18th birthday, you can't play sports. You just you become uncoordinated on your 18th birthday exactly, and then it gets better. In the data, that might look like freshmen occasionally have a bad day because they turn 18 on their somewhere. And I wouldn't know if freshmen tend to choke because the pressure gets to them and they behave a little bit less evenly, or freshmen,
25:53 Speaker 01 or there's this miraculous 18th birthday thing. Now, if their birthdays were in there, well, then I would see, oh, look, it always happens to be on the day they turn 18. And then
26:03 Neil deGrasse Tyson? It's because they got shitfaced that night of their birthday.
26:05 Speaker 01 Well, that might be the reason. But it's the kind of thing that that's everything.
26:13 Speaker 01 Everything is lurking in the data. It's very difficult to imagine a phenomenon that that both matters and is not somehow present in the data for you to tease out if only you knew how to look.
26:28 Neil deGrasse Tyson? So that's the lesson here, really.
26:30 Gary O'Reilly? All right, let's let's.
26:31 Speaker 01 That is the lesson.
26:32 Gary O'Reilly? All right, let's let's fly this thing out of this atmosphere, shall we? Uh, Abby Chris, hey, oh, experts, he says, I've been watching the expanse, which made me think how we could we conduct basketball tournament with people who are from various planets and asteroids that have been settled by humans like, for example, champions of Mars versus champions of Jupiter's moons. Right now, this for me is interesting because you've got a whole new set of metrics to factor into your machine learning here. What are some of the things that we would need to think about to keep the field equal and fair for all teams? Love the show and everything, keep informing fellow, so you're welcome.
27:10 Neil deGrasse Tyson? Wow, so let me just, as a way of lead into that. Um, so so Matt, when people started training for the Olympics at high altitude, no, it's not another planet, it's this planet, but it's a different environmental conditions under which their body is getting trained. And now they all go into the same stadium and some people outperform the others. So that's just an interesting realization there that maybe the environment in which you train, the gravity, the air quality, the air density, the uh that can definitely show up in your performance.
27:51 Speaker 01 I think that's right, and I don't think we don't historically, I mean, I have no idea what's going to happen when we have people from Mars, but historically, we don't try to compensate for that. So, you know, Kenyans win marathons. It's just how it is. And and there is, but there is not anything in the rules saying that any Kenyan entering a marathon has had ankle weights.
28:15 Neil deGrasse Tyson? As a handicap.
28:18 Speaker 01 As a handicap. There's nothing in the NBA to try and make it easier for short people to play.
28:24 Neil deGrasse Tyson? Give him a step stool.
28:27 Neil deGrasse Tyson? You put a little trampoline for their dogs, right?
28:30 Speaker 01 We don't do that. So I think that if, um, if Martians, you know, people from Mars have some physical difference that makes them better at a game, hopefully we will just celebrate with them. Probably it sucks to live on Mars. So we'll celebrate with them that they're better. I mean, you love watching marathons because the people who are so good are so good. And I don't know of anybody, I mean, I never bemoaned the fact that I'm, I'm not going to ever be a professional wrestler. It's just not going to happen.
29:03 Chuck Nice? There's still time. There's still time, man.
29:08 Gary O'Reilly? You're being vicious there Chuck. I mean, I mean, if we bring it back to this planet just temporarily, if I have to cross the west coast to the east coast, and if I'm altitude and we come and play at sea level, there's recovery, then there's oxygenation and ability to, these things, are these not factors that are relevant?
29:31 Speaker 01 Of course they're relevant. Just like home field advantage is relevant, right?
29:35 Gary O'Reilly? Not anymore because there's no fans.
29:38 Speaker 01 No, but playing at Fenway is different than playing at Wrigley Field. The green monster matters. And Boston players, Boston selects their players in part because they're looking for people who have the natural skills that will exploit the peculiarities of their park.
29:58 Neil deGrasse Tyson? As did the Yankees for so many years because the, the right field line was one of the shortest of all ballfields. It was something like 296 feet or something. It was very short. And so the Yankees had a lot of lefty sluggers that racked in the home runs, simply because that's short porch out on right field. And in fact, if you, you're right, Matt, we don't go back to the record books and say, you know, half your home runs were 310 feet and they would not have been a home run in any other stadium, but you happen to play for the Yankees. So we're going to subtract those. We don't do that. We just allow the circumstances to be the expression of that, of that ability. So that's an interesting take on this.
30:40 Neil deGrasse Tyson? Do you know the tug of war used to be in the Olympics as an event? And the rule was everyone in the tug of war had to be the same profession. They had to be a group that made sense that they competed together. They all had to be like medical doctors or they all had to be, you know, soccer players. They all had to be policemen or, and so it turned out that the mounted police always won. Because they, you also had to wear your native uniform and they all had these steel reinforced boots.
31:16 Neil deGrasse Tyson? And then they'd realize this is a stupid event and let's just get rid of it.
31:19 Chuck Nice? They're also all used to pulling on the reins of an obstinate horse. So like, let's go.
31:26 Speaker 01 So, I mean, the closest I've seen to this, my wife was a hydroplane racer.
31:31 Chuck Nice? Wow.
31:32 Neil deGrasse Tyson? Wait, wait, she was a what?
31:33 Chuck Nice? A hydroplane racer.
31:34 Speaker 01 A hydroplane racer. She was the national champion.
31:36 Chuck Nice? That's a thing. Holy crap!
31:38 Speaker 01 Yeah, it's these little boats that fly across the water.
31:41 Chuck Nice? And it's, it's one of the coolest sports ever, by the way.
31:45 Speaker 01 So her class had a weight limit and she was like the only woman. She weighed way less than everybody else and they actually made her put a plate of lead in the bottom of her boat.
31:58 Chuck Nice? Right.
31:58 Speaker 01 So that she could meet the weight limit and she hated it because everybody else, you know, you're there and you're driving and you lean. And you move your whole weight and she could not move the lead plate. It was just stuck on the bottom of the boat and she could lean, but she was leaning with much less mass than all the guys she was competing against.
32:19 Neil deGrasse Tyson? Load pockets in a vest on her so that she could lean with the weight and match other people's capacity.
32:25 Speaker 01 She decided the lead weight was better. I mean, she was ridiculously, and she, she's beat them all anyway. So she was fine, but, um.
32:33 Chuck Nice? Wow, she's that, she's just that competitive. That's us. She's just like, I could have, I could have destroyed you instead of just beat you.
32:45 Chuck Nice? Wow. Oh, man.
32:47 Neil deGrasse Tyson? Also, speaking of home field advantage in an interplanetary contest. So it would matter if you played your sport in, on your planet.
32:58 Neil deGrasse Tyson? Right? I mean, that would matter.
33:00 Chuck Nice? Presumably.
33:00 Gary O'Reilly? Yeah.
33:01 Neil deGrasse Tyson? That would be the ultimate home field advantage because you know your gravity and your air quality and your, your, um, all the peculiarities of your environment.
33:10 Gary O'Reilly? Plus Neil, if I have to travel from planet A to planet B and it's X amount of light years, I have to get there and acclimatize. Therefore, it's recovery. I'd have to turn up however many years in advance to.
33:22 Neil deGrasse Tyson? Years ahead of time.
33:24 Gary O'Reilly? Plus my journey. Yeah.
33:28 Neil deGrasse Tyson? No, you know, it'll be, it'll be like the, the, the, the, uh, what do you call it in tennis where there's the four events, but they're not all on the same.
33:36 Chuck Nice? The Grand Slam.
33:37 Neil deGrasse Tyson? The Grand Slam. So the, so Wimbledon is on grass.
33:40 Gary O'Reilly? Yeah.
33:41 Neil deGrasse Tyson? And who is it on clay? Uh, is it?
33:43 Gary O'Reilly? It's a, Ronald Garros, Ronald Garros, yeah.
33:46 Neil deGrasse Tyson? The French opens on clay and then you have the concrete, you know, at Forest Hills. So that's interesting. So if you, if you do the whole circuit, then you need a combination of abilities. So, which is what makes winning the full Grand Slam so that so much more impressive than anyone who's only has the talent for just one. So, you know, that's cool.
34:07 Gary O'Reilly? All right. Let's get, let's get another one of our Patreon patrons.
34:10 Neil deGrasse Tyson? Okay, go ahead, go ahead.
34:11 Gary O'Reilly? He is the delightful name of Craig Woolhouse and he is from New Zealand where he's proudly flagging up the fact that, that they stopped COVID. Congratulations.
34:21 Neil deGrasse Tyson? Yes.
34:22 Gary O'Reilly? Um, if a game of basketball.
34:24 Neil deGrasse Tyson? It helps when you're a tiny island and you don't let anybody in. Okay, go on.
34:27 Gary O'Reilly? Don't tell them that. They won't listen. They're very, they're a very proud nation and they're great at rugby. Uh, if.
34:37 Chuck Nice? Plus I don't want, I don't want anybody showing up here doing the huff.
34:40 Gary O'Reilly? That's, that's a hundred percent for certain. Right. He says, if a game of basketball was held on Mars indoors with Earth's atmospheric pressure, would we finally be able to dunk it from the three point line and would it count as a three? Or do we need another planet?
34:55 Neil deGrasse Tyson? And he says. Oh, I like, I like that. And we will get to that answer when we come back on Star Talks Sports Edition.
35:14 Chuck Nice? Time now for another Patreon shout out to the following Patreon patrons Erdem, Memes Yazici, Priscilla and Kyle, and Stephen Severin. Guys, you're the greatest. Thanks for helping us make this show the best that it can be. And for anybody listening who would like their very own Patreon shout out, please go to patreon.com/stargtorkradio and support us.
35:47 Neil deGrasse Tyson? We're back. StarTalk Sports Edition. Cosmic Queries. We started out with March Madness, and now we're thinking about sports on other planets, what role AI could play in predicting winners. And we've got Matt Ginsberg with us, becoming a friend of StarTalk. So it's not your first rodeo with us. Thanks for coming back, Matt. So we're picking up on a question. I love this question. If you could restate that.
36:11 Gary O'Reilly? Of course. It's from Craig Wallhouse. He's one of our Patreon patrons, an exclusive member. And he says, if a game of basketball was held on Mars indoors with Earth's atmospheric pressure, would we finally be able to dunk it from the three-point line? And would it count as a three? Or D, we need another planet? And then he cites the one that MJ, that's Michael Jordan, came from. So, uh, hmm. Your answers, please.
36:37 Neil deGrasse Tyson? So Matt, if you could, if you have anything to add to what I say, I'd be delighted for you to sort of jump in. But on Mars, there's about 40% of Earth's gravity there. So if you weigh, you know, 100 pounds on Earth, you weigh 40 pounds on Mars. And all of your musculature is accommodating the 100 pounds that you weigh. So now you only have to sort of move 40 pounds, sort of up against gravity. So you can jump higher. And you fall more slowly. Okay, that's the important part. So the hang time is there. Okay. So if you can jump higher, and you have good hang time, and if you get a good running start.
37:20 Neil deGrasse Tyson? I'm think I didn't run all the equations on this, but I'm thinking you could dunk from the three-point line and count it as a three-pointer because you would not have touched the ground and the ball wouldn't have hit the ground in between the three-point shot line and and the main basket. Matt, what do you think of that?
37:38 Speaker 01 I think that's right. I mean, certainly from a rules perspective, that's the easy part of the question. Um, you know, you take off from behind the three-point line, it's a three-pointer. I haven't done the calculation either. I suspect that a second is still a second and a meter is actually what's reduced by 40%. So you probably can jump like two and a half times as far.
37:57 Neil deGrasse Tyson? Mm hmm.
37:58 Speaker 01 And given that Michael Jordan can dunk from behind the three-point line anyway.
38:03 Neil deGrasse Tyson? No, no, he can dunk from Foul line. The the foul line, not the three-point line.
38:08 Speaker 01 Well, didn't anybody ask him to try?
38:11 Neil deGrasse Tyson? So just just something to be clear about what he's actually doing, which is not obvious unless you analyze it. So generally, if you're trying to dunk, the point where you're dunking is the highest part of your arc because the rim is 10 feet up. Whereas Michael Jordan from the free throw line, he is not still ascending at the point he's dunking. He has already peaked in his parabolic arc and he's on his way down. So he had to jump that high in order to make all of that happen. So if you watch his arc, he's on his way down. So it doesn't have to be sort of the limit of where you're jumping, provided you got up high enough, you could just descend into the dunk.
38:56 Gary O'Reilly? Okay, I want to sound like the smart kids in this conversation, so I didn't do the calculations either. However, I know if we're doing this with this sort of ability, I'm making the court bigger.
39:12 Neil deGrasse Tyson? And the rim taller, the rim higher?
39:14 Gary O'Reilly? Yep, I'm raising the bar, I'm stretching the court and yeah, because if the hang time is seconds, plural, then it's a different game if we play it on this particular, people will be flying out the arena.
39:27 Chuck Nice? It's a better game.
39:28 Gary O'Reilly? Yeah.
39:29 Neil deGrasse Tyson? No, no, wait, wait, wait, but if you do that, then you're neutralizing all these interesting features of the Martian basketball court.
39:36 Chuck Nice? Right.
39:36 Gary O'Reilly? But then again, it's more like basketball because we've kind of made it equivalent in the size of the court as opposed to an Earth court.
39:45 Speaker 01 No, I think it's way more complicated than that, right? Are you gonna make the hoop bigger? If you're shooting from so much further away, you have to make the hoop bigger, but then the inside game becomes tremendously different. So I think it's, I don't think you can, you can't rebalance it. It's gonna be a different game.
40:01 Gary O'Reilly? All right.
40:02 Neil deGrasse Tyson? So here, very good point. Right, just to be clear, just again, to add emphasis to Matt's point, you shoot the ball at the basket and there's a certain margin of error in angle, outside of which you're not gonna make the basket, right? And that is true in any gravity, right? So, so you're not helped at a lower gravity by this sort of margin of error angle. So if you're gonna shoot from twice as far away, then you're gonna make half as many baskets because that, the angle will no longer accommodate the distance over which the ball is veering off course.
40:42 Chuck Nice? Unless you're Steph Curry. But, but I will, I wanna, I wanna just litigate this.
40:46 Neil deGrasse Tyson? Steph Curry on Mars, that's a new movie, we gotta do that.
40:49 Chuck Nice? I just wanna litigate one quick point for you guys to figure out, okay? If shooting a ball from behind the three-point line is what entitles you to the extra point, why would dunking the ball from behind the three-point line still get you that same extra point as if...
41:09 Speaker 01 It's your feet.
41:10 Chuck Nice? Well, no, the shot actually happens at the rim.
41:14 Neil deGrasse Tyson? Okay, no, no, so here it is.
41:16 Speaker 01 A guy standing at the three-point line, leaning forward, so his hands are inside the three-point arc, he's shooting, still gets a three points.
41:25 Neil deGrasse Tyson? Not only that, you can jump from behind the three-point line, land inside the three-point line, and it counts as a three-point.
41:33 Chuck Nice? Okay. So I
41:35 Speaker 01 It's all about your feet.
41:36 Chuck Nice? But it's all about the feet. It has nothing to do with the actual shot.
41:39 Neil deGrasse Tyson? Yeah.
41:39 Chuck Nice? It's just the position of your feet.
41:41 Neil deGrasse Tyson? Correct.
41:41 Chuck Nice? All right, gotcha.
41:42 Neil deGrasse Tyson? Yeah, so technically it works. It reminds me of the movie, which was it? That stupid movie, but it was entertaining, where they invented the alley-oop in the movie. And in the alley-oop, it's like, is that legal? You know, the first alley-oop, you gotta look at it and say, did anything happen illegal there? I can't think of it. It should be illegal, but apparently it's not, and we kept it. All right, there you go.
42:06 Gary O'Reilly? All right.
42:06 Chuck Nice? Cool.
42:07 Gary O'Reilly? All right. Another question. This is another one of our Patreon patrons, James Senior. He says, a question about AI. When do you think we will actually have AI in a sense of an actual artificial consciousness? Also, how would this be achieved? Would it come from an algorithm or from actually uploading the human consciousness into a computer? Or by other means? So we've dropped the ball, timeout, halftime, whatever you want to call it, and we're now thinking about AI. Matt, over to you.
42:37 Speaker 01 There's some assumptions underlying that question that I don't know if they're right. So I think getting uploaded into a computer, if that happens to me before I, I, uh, before my, I guess my warranty is up, that would be fine with me. But I don't think that counts as an AI. That's just Matt inside a computer somewhere. So the question of what it is to be intelligent actually becomes important and interesting. Historically, the definition of intelligence was able to pass the Turing test, which is something invented by British mathematician Alan Turing. And it basically says, if I'm on one end, you're typing into a computer, and an entity is typing back.
43:21 Speaker 01 And if you can't tell if that entity is a machine or a person, the machine's intelligent.
43:26 Chuck Nice? Okay.
43:28 Speaker 01 If you go back to what I said earlier about the 49% versus the 99% problems, the machines that look sort of, quote, intelligent, unquote, are going to be the ones that solve the 49% problems. They're going to do things we can't do. They're going to solve problems we can't solve. But they're not going to look like us. They're not going to pass the Turing test. They're gonna... I don't know how they're going to deal with traffic lights, but it's scary. They're and but they're going to be great at trading stocks. They're going to be great at predicting who's going to win sporting events. They're going to be great at predicting the weather. They're going to do all sorts of things and help us.
44:13 Speaker 01 And they're going to have something that we, I think, will come to think of as intelligence.
44:19 Speaker 01 That there's not going to be a moment where all of a sudden they go from not intelligent to intelligent. We're already seeing that. As machines do more and more.
44:29 Neil deGrasse Tyson? I think you're also implying that we should not hold consciousness as the metric of whether the thing is intelligent or useful or can get the job done.
44:40 Speaker 01 That's right, because I don't know what consciousness is. I mean, are we waiting for a machine to say, hey, leave me alone. I haven't had my coffee. Because that's not that's never going to happen. Why would we build machines like that? Why would we build machines that are grumpy and machines that need coffee?
44:56 Chuck Nice? Well, the fantasy always is that the machine will come to this state of being on its own through some evolutionary process.
45:09 Neil deGrasse Tyson? Achieve consciousness.
45:11 Chuck Nice? All right, it will. So it wasn't designed that way. It becomes that way through so many experiences that it is able to decipher for itself that it indeed is sentient and conscious.
45:25 Speaker 01 So the the number of times that things happen in the movies is not a tremendous indicator of how frequently they will happen in real life. And I don't I don't see it going that way.
45:38 Neil deGrasse Tyson? That was just a diss, Chuck, in case you know.
45:40 Chuck Nice? I never said it came from a movie. I said it was the fantasy.
45:45 Neil deGrasse Tyson? That was a diss, okay? Polite one at that.
45:49 Speaker 01 I don't think it's going to be like Terminator. I think we're going to find that these machines are our partners. We can do things. We can solve problems. They can't. They can solve problems. We can't. And we will collectively do more than either of us could do individually. And I think. And that's now, that's today and tomorrow. As far as will they eventually so completely surpass us that we become unnecessary in some way?
46:22 Speaker 01 I don't know, maybe in some far off land, but I imagine that far off time, I imagine that the team, the man machine team is going to be so much better and it will grow, we will grow together, we will work together to always do things that we can't do individually. I'm incredibly optimistic about this.
46:41 Neil deGrasse Tyson? Although you did say, you did say, just to let the record show, that the future of machines probably won't be a Terminator. That's, I don't know how encouraging that is, because any Terminator at all would be bad.
46:54 Speaker 01 Yes.
46:55 Neil deGrasse Tyson? I'm just saying.
46:55 Speaker 01 Yes, and I, I think that, you know, could we, we could get there if we worked at it, but that would require an enormous amount of stupidity by a relatively large number of people. And that's why it's just probably.
47:10 Neil deGrasse Tyson? Never underestimate.
47:12 Speaker 01 Because exactly.
47:13 Gary O'Reilly? Yeah, humanity's never proved, yeah, humanity's never proved that they can do that, ever. I'm, I'm, I'm just wondering, Matt, if, if machines are constantly learning, that, that part where they can't solve the...
47:28 Speaker 01 A hundred percent problem.
47:29 Gary O'Reilly? Yeah, very well. There'll be a point where they can. And then, then are we not redundant?
47:36 Speaker 01 I don't know, but there will be a point where they can. They really do, they're different, so they're architecturally different, right? So we have a trillion neurons operating on millisecond time scales.
47:49 Gary O'Reilly? Yeah.
47:50 Speaker 01 Machines have, even these massively powerful machines have thousands of processors operating on nanosecond time scales. We're different architectures. We should be good at solving different kinds of problems. And I, you know, could you eventually simulate a human brain and computer and make it all sort of the same?
48:08 Neil deGrasse Tyson? But then why would you, is the point?
48:10 Speaker 01 Yeah, but...
48:10 Neil deGrasse Tyson? I love it. I love this angle on it.
48:12 Speaker 01 There's so much, there...
48:16 Speaker 01 You know, it's like, it's like something your, your parents for your eighteenth birthday, buy you a Porsche. And you say, I only want to use it to drive up the driveway and get the mail. Nobody would do that. You're gonna use the Porsche as a Porsche. We're gonna use these, these computers that have abilities we lack in the areas where we need help.
48:37 Chuck Nice? Just learned that I've been using my Porsche wrong all these years.
48:40 Neil deGrasse Tyson? But you got your mail fast. Yeah. That was the early internet. Let me get to my mailbox as fast as I can. We got time for one last question if it has a quick answer.
48:53 Gary O'Reilly? All right, I think, I think we've got to go back to The Factor Man and God's Algorithm, Matt Ginsberg's book. Didn't you want to know something?
49:01 Neil deGrasse Tyson? Yeah, I just, I just, tell me what happens. What, that's a novel, right? What is God's equation?
49:07 Speaker 01 Um, it's about a guy who finds what's called God's algorithm. It lets him solve basically any problem. And my view is that he's, anybody who finds this is, it's a race, whether the government kills him or he takes over the world first. And this guy realizes, he realizes he's in this race. He doesn't actually want to take over the world. He, he mostly just wants to go to Disneyland with his kids. And it's about his attempted journey to make the world a better place before this technology is used to, to mess everything up.
49:39 Neil deGrasse Tyson? So this is very-
49:40 Speaker 01 It's a thriller. It's supposed to be fun.
49:42 Neil deGrasse Tyson? Uh, this is, uh, uh, this is a, um, uh, do you have confidence that such an equation exists, such an algorithm exists?
49:50 Speaker 01 So there's, this is, the biggest open question in computer science is whether such an algorithm exists. And I believe it does. Confidence is probably a little bit too strong because I'm in a pretty small minority. Um, people occasionally measure it. And I think, I think something like 10% of the serious computer scientists believe it, something like that.
50:11 Neil deGrasse Tyson? And that would give you access, you'd be able to tap future knowledge of systems that would be without precedent in the history of civilization. That would make you all powerful.
50:24 Neil deGrasse Tyson? Then you can move out of your parents basement where you are right now. You save this as you. Neil. I can go get the butler back out of quarantine. Get the butler. And start living it up like I would like. I love that. Get the butler out of quarantine.
50:40 Neil deGrasse Tyson? All right, we got to call it quits there. Matt Ginsberg, great to have you back. Always. StarTalk Sports Edition. Thank you. And there's more to plumb in your, in your expertise on these topics and we'll surely come back to you on this. Thank you. That'll be great. Excellent. And Gary, always good to have you, man. Pleasure, my friend. All right, Chuck. Chuck, you baby. Love you there. We'll see you. All right. I'm Neil deGrasse Tyson, your personal astrophysicist, bidding you farewell from StarTalk Sports Edition. As always, keep looking up.
Previous Cosmic Queries – Origins of the Universe, with Janna Levin Next Cosmic Queries – Life on Exoplanets with Sara Seager Page updated September 28, 2026 · transcript and summary are machine-generated
