Navigating the Algorithmic Landscape: Insights from a Mathematician

welcomes listeners to an engaging exploration of the world of algorithms and their profound impact on modern economies in this insightful episode of the Blind Knowledge podcast.

He introduces his guest, Noah Healy, a recreational mathematician and professional algorithm developer, who brings a wealth of knowledge to the table.

The conversation delves into the intricate web of market mechanisms, particularly focusing on Healey's innovative concept of Coordinated Discovery Markets.

This new approach aims to refine price discovery in commodity exchanges, promoting efficiency and fairness in trading, while also addressing the complexities of supply and demand dynamics.

Transcript

Copy Transcript Copied

SPEAKER_00 0:00

My name is Joey B. This is the Blind Knowledge Podcast. If you know it, you know it. And if you don't know, you do. Today is a good day. Today is a scientific day. We're gonna get mathematical. We're gonna talk about algorithms, my friends. Algorithms. Do you know them? Do you know what they are? Google it real quick. And if you don't have time, hang out with me and my guest today. My guest is Noah Healy. I'm gonna bring him on in just a moment, but just a sec, just for you. If you don't know about it, I gotta tell you, blindknowledge.com is the spot that's hot. Check it out, man. It's new. We're trying to be informative and entertaining. So we got digital content, digital content creation. And hopefully, if you're a digital artist or digital media content creator and you want to get on uh blind knowledge, let us know. Let us know. We're over on Twitter at blind underscore knowledge. There he is. Hey Noah, can you hear me?

SPEAKER_01 0:54

Yes, I can.

SPEAKER_00 0:56

Cool, man. Hey, um, welcome to the show. I've never had an actual mathematician or an algorithm developer on here. So congratulations, you're the first one.

SPEAKER_01 1:08

Thanks. Thanks for having me here.

SPEAKER_00 1:10

No, it's a it's a pleasure. It's a pleasure. Um uh you you said a lot of really interesting things in the prep that I don't want to get into too quick because there's just so much information we have to get out for everybody. Can you tell us just to start? Who are you and and what do you do?

SPEAKER_01 1:27

Uh well, I'm a recreational mathematician, and uh so that means that I I do math for fun. My interest is computational mathematics, uh, which is the kind of math that describes what computers do and can't do. And uh a little while back, uh better part of a decade ago, I was toying around with a problem in information theory and found a new approach to economic markets. And so since then I've been working on developing, patenting, and and promoting uh a better form of economy. And wow.

SPEAKER_00 2:15

Wow, so a better form of economy. So what what does that mean?

SPEAKER_01 2:21

So the the primary problem of economics is organization. Um we have a lot of people, we have a lot of capacity, uh, there's a lot of wealth, there's a big world out there. So the problem is how do you get people to do things that are actually valuable to themselves and other people because it's very confusing. And you could get real good at making, you know, pins, for example. But if the world's got all the pins it needs, then you know, you should stop and you know, stop and smell the roses.

SPEAKER_00 3:01

Yeah, maybe stop making so many pins.

SPEAKER_01 3:03

Exactly. Okay. So we need some kind of system to tell us what's important, what's unimportant, that kind of stuff. And uh the best system that's ever been developed are open markets, where people can come in, buyers and sellers can come together, negotiate prices with one another, and find the spot that causes supply and demand to balance off against each other. I've found a way to change those two-sided markets into three-sided marketplaces, um, creating separate markets for supply, demand, and information and negotiation. And that allows the entire system to be more efficient and consequently less expensive. And so the less money that we spend in the financial sector, the more money we have for increasing production or increasing vacation time, depending on what people want to do with the extra money once we've got it.

SPEAKER_00 4:09

Okay. Okay. So work smarter, not harder, save a couple bucks, and we're talking about exchanges. We're talking about a is it commodity exchanges specifically?

SPEAKER_01 4:21

Commodity exchanges are sort of the foundation of the financial system. Um, because at the base of things, people gotta eat. We gotta keep you know, oil flowing around, electricity, metals. These these basic products are the things that we make our world out of, that we eat, that we drink, and so on. Um on top of that are then other kinds of financial exchanges where you trade companies and debts of people that are engaging in these activities of producing and using these raw materials. So uh I'm kind of starting at the base of the chain, and then we can sort of grow outwards from there.

SPEAKER_00 5:05

Okay. Okay. Um it's it's a lot of information um right off the bat for sure. So I want to make sure we're breaking it down in a way that everyone can kind of digest and understand. Um so what led you into this into this development? Because you're an algorithm developer, correct?

SPEAKER_01 5:28

Yeah, yeah.

SPEAKER_00 5:29

So how did you how did you end up where did that start for you? When did you start doing algorithms? How did that end up into CoreDesk?

SPEAKER_01 5:37

So I started getting into algorithms uh when I got out of college, needed a job, and started working for a local company called Boxer Jam, which was a pioneer in social gaming. And uh they put me on task of doing things like analyzing the web logs to work out player behavior and so on. They had they had a weblog parser, but it had been built in the early days of their website, so it only tracked a few different pages. And uh they also were starting to add extra servers to keep up with the amount of content they were putting out, and it didn't play well with that situation. And the session tracking algorithm that it was using was so memory intense that it was actually choking and it it couldn't actually finish anymore. And so I pretty much learned how to program the language that they used while doing that project. And it took a while, but ultimately uh with a lot of help, I was able to increase the speed of the algorithm by uh or the speed times the resource usage by a factor of a million.

SPEAKER_00 6:56

Um okay, so you gave this bad boy the Heimlich, basically, and you got this thing unchoked and running, correct?

SPEAKER_01 7:04

Yeah, yeah. And that's that's a that's a pretty exciting feeling. Uh and at the same time, I was also uh doing deep dives into the the underlying mathematics of this stuff. And that's just it's really fascinating. And thanks to the fact that the internet exists and the people who came up with most of this mathematics invented most of the stuff that makes the internet work, it's all very easy to find on the internet. And so that's that's what I was doing. I was learning about this new form of math, applying it to my job, making things a lot better than they used to be. And so that became a real big piece of my life. Uh and so then um fast forward, you know, a decade and a half from from there, uh, I've learned a lot. I had finished a job and sort of decided to take some time and space to just think about these math problems. I had money in the bank, so I was okay that way. And I just wanted to see if I could think up anything interesting. So I was I was doing studying and and trying to find some new approaches to some sort of new problems. And I was working on the problem of communicating consensus. And I found an approach using game theory, and I was talking to a friend about it, and he asked about using it to predict what markets would do. And I realized that there was an intriguing possibility that you could build a market with the technology. Um and then and this came out, and so I started analyzing this compared to the algorithms of the existing marketplace, and I got that I got that old, you know, million X tickle. This is actually only about 300,000 X. Um but uh but that's that's highly significant in in real terms. And so uh once you understand that you've got an algorithm that works that much better, and you know about the costs that are imposed by the inefficiencies of the existing markets, um there's there's nothing else I could be working on that's that's more economically valuable than this.

SPEAKER_00 9:44

So are you very educated, very well educated, or are you education? Because it sounds like you have to be wicked smart to do this stuff. Uh what's your what's your educational background like?

SPEAKER_01 9:57

Uh so I uh I grew up in a university town, and so I started going to the University of Virginia while I was still in high school, um, because that's what they do if you finish out a course of high school education. And I was part of several different pilot programs, some of which took and some of which didn't, to accelerate mathematical education. So that meant that I'd finished off the math classes in high school by the time I was a sophomore. Then I went to UVA, uh, just basically wandered around the engineering school taking interesting-looking classes, wound up doing a year of grad school in their nuclear engineering department while they were shutting the place down. Uh, and then got out, like I said, jumped into uh internet startups for the most part since 2000 and uh been using the internet as a as a research tool uh since then. The thing about computational math is that the the entry point is is kind of shallow. Um uh you know, the games like Minecraft provide uh environments that would allow children to do some rudimentary programming. Um but the the depth is so great that it almost doesn't matter how smart you are because you can just keep going until you're you're exhausted and you can't go any further, and there's just an infinite amount more that's that's out there for you to go explore later.

SPEAKER_00 11:38

So um so when you say like go and explore, I'm gonna break this down to like um just regular like grade school education. What are we what are we exploring? How does it go? Like what is it? You know what I'm saying?

SPEAKER_01 11:56

So thinking about grade school education, some of the earliest algorithms you learn are arithmetic. So you learn how to add. You learn that two and three are five, but then you start doing multiplace adding, where you know, eight and seven are also five, but you gotta remember that one that goes over to the next place and turns it into a 15. And so that carrying the one is the algorithm for adding. And then once kids get good at that, you learn about multiplication and the whole wedding cake thing where you do the offset and then you get all the numbers that you can add up and stuff like that. Well, it turns out that that's not the best way to multiply. Computers use something called the fast Fourier transform to multiply in a way that's a lot faster than the way that we teach people in school. It's also a lot more complicated. Um but that's not the only way to multiply. In geometry, you can multiply by making triangles and making similar triangles. And so in every discipline of mathematics, there are techniques that basically let you multiply things, let you add things, and each one of them has an algorithm, and there can be multiple algorithms that do each of those things. And so that's that's the exploration. You find out all the tools that exist, you find out problems, you think about new problems or applying new tools to old problems, and you see how that works. Is it faster? Is it slower? Um of the most important things in computation is sorting, putting things in order. There's a there's a sort of a joke competition around something called BOGO sort, and BOGO is like the worst thing ever. And it's it's sorting algorithms that theoretically will finish, but might take forever. Um so uh one example of a BOGO sort is that you shuffle everything and then you check and see if they're in the right order. So if you had a deck of cards, you could put them back in the in the right order by say dealing them out face up into piles, um and then sorting the piles into the right order and putting the deck back together. Okay, what you would do is you'd shuffle the deck a few times and then fan it out and see if it was in the right order. And if it was, you're done. But if it isn't, shuffle the deck again. And again and again and again until you again and again until well until the universe runs out, actually, because that would that would happen long before you'd hit it.

SPEAKER_00 14:52

Fair. Okay, so you're you're basically you're taking every kind of um which way, um, whether it's randoms or whether it's actually um, you know, planned.

SPEAKER_01 15:03

So it's randomness can be a valuable uh tool in many algorithms. Um there's something called the Monte Carlo method, uh, where instead of trying to figure out a really good theoretical model of how something might happen, um you actually build a model of sort of what does happen and just throw a bunch of of random instances at it. Uh so like if you can see yeah, kind of see what happens. Um so if you if you can run a computer trial of something, uh you can run thousands or millions of computer trials. And uh and then and then you can use those outcomes to give you a good model presentation of how the system actually behaves. And so that's that's actually when I was doing the analysis at back at Boxer Jam, I was using our customer sessions as a Monte Carlo model. Uh there a kind of a special kind of stochastic system called a Markov chain, where I was treating the customer behaviors as just random bounces through our website. And then I was adding all of them onto each other and looking at what the probability of moving from one part of our website to another one was. And that's that allows you to figure out what's popular, what's unpopular, what's making you money, what's losing you money.

SPEAKER_00 16:41

The algorithm, just figuring out like all of these different instances and which way and who's it's and where someone's gonna go, basically. That's I mean, that's that's some serious stuff. It's almost like we should have known about this kind of technology before Twitter, before Facebook, before we actually started hearing about the word algorithm. Have algorithms been around uh for longer than we know about them or have known about them?

SPEAKER_01 17:09

Absolutely, yes. Though the word algorithm uh belongs to the same uh root as the word algebra, actually. And they're both named after a guy. He's uh he's a mathematician from the Arab world from close to a thousand years ago at this point. Uh but a great deal of mathematics actually has concerned algorithms. Uh, one of the most ancient pieces of mathematics we have is something called the sieve of Aratsenes. And the way that works is you just write out a bunch of no all the numbers, you know, two, three, four, five, just keep going. And you circle the one at the front, the two, and then every two numbers you strike it out. And so then the first unstruck out number is three, circle that, and then every three numbers strike out and go like that. Well, what happens is every number you circle is prime. Okay, Steve Verosthenes uh up until about 30 years ago was the most efficient way to list prime numbers that had ever been discovered by human beings.

SPEAKER_00 18:27

I feel like we went over this in the fourth grade somewhere. I don't know.

SPEAKER_01 18:31

Probably did, actually. Um, when when they introduced prime numbers, that's a pretty classic thing to show kids that they kind of put the first hundred numbers in a little box and like you know, trust out the fives and cross out the twos, yeah, go down with the threes. That that's an algorithm that is a couple thousand years old.

SPEAKER_00 18:53

Wow. Wow. So, okay, so algorithm, big scary, funky word that we all know, but we don't really know how it works. But we're learning, we're all learning because we're here with Noah Healy. Noah, thank you again for coming on to the show. Um, we because we we need the knowledge, we need you to drop some knowledge on us about algorithms, especially before we get back into more of what you do specifically in this huge major project we got to get into. Um, you know, algorithms are fascinating. You know, I don't know about you, and I and I assume you you find them fascinating because you work in the field, you are an algorithm developer, but I I find them fascinating because they can be tweaked and they can be um just kind of the like the Google algorithm. It seems like Google itself has actually changed its search algorithm over the years. And it seems like recently, I don't know about you, but I see a lot more ads come up on the first two pages than I ever did before. It's almost like you have to search harder to get to the answer on Google, which wasn't the case like five to seven years ago. Why do they do that? Why why do they change it? And how do they change it? Like, is there like an algorithm box that they have to open up with a key and say, all right, we're gonna change this five to it too?

SPEAKER_01 20:14

Um, basically, yes. Uh that that's that's kind of how it works out is computers don't just do stuff, they do what we tell them to do. And there's a lot of different ways for us to tell them to do things, but no matter what technique we're using to tell to do things, and no matter what they're doing, at the heart of what they're doing is the algorithm of what they're doing. And so because that that computer thing is just pretty much, you know, text, numbers, stuff like that. Um if if you think about like uh if if a mechanic and metal worker owned a car, they could go in and and move stuff around if they wanted to. They could they could raise the engine by a quarter of an inch or or move the struts of the wheels around if they felt like it. And that might make the car work better or it might make the car work worse. But if they wanted to, they could change those things. Well, computers the the thing that puts them together is just a file. It's it's just it's just like anything you might have written in an email. And so those things are very easy to change. And it's not like Google is a car that we have several million of and everybody's driving it around. There's one Google, and they're constantly trying to figure out how to get what they want. And so they just go in there and you know, get the crescent wrenches out and move stuff around and see if that works better or worse. Uh, and that's that's a really severe problem that we have with these with our current leading companies, is that whether you're talking about Google or Amazon or Facebook, Netflix, Twitter, uh, what makes these companies function is the algorithms that allow them to have the presence, show you what they're showing you, do what they do, and make whatever money they make. And they need to keep those algorithms secret, which is why we've got kind of this heebie jeebies type of feeling about them these days. Uh, and that gets into the game theory. Um Google is essentially deciding what's important and what's unimportant on the web. That's a lot of power.

SPEAKER_00 22:55

Right. Yeah, true.

SPEAKER_01 22:57

So if you knew how they were making that decision, you could do things that they