QUOTERSCAST #1 – Interview with Noah Healy of Coordisc.com Podcast link below…
TRANSCRIPTION…
Today on the QuotersCast, we are talking to Noah Healy of coordisc.com.
Noah is the creator of Coordinated Discovery Market, which is a technology product that has the potential to completely re-order the global financial system.
Here’s our conversation now……
Host Renee: I wanted to get your take on this. Number one, I would like you to explain in layman’s terms, exactly what it is that you do and how that may correlate with the insurance industry, because I know it’s a little off-topic, but it’s so fascinating, and I think you’re doing something so unusual, that I want people to appreciate what you’re doing.
Guest Noah Healy: So what I do is algorithm development and basically strategic manipulation or structuring… I have a system for rebuilding the market place pretty much from first principles, using computer technology to allow better price signaling effectively, existing markets are basically about and sell, and while there’s a little bit of information about future desired strike prices and so on, at the core, people who are buying, think prices should be higher or will go higher, and people that are selling things that… Prices will go lower. So in information terms, there’s sort of a maximum of one bit of information in that signal, and since people aren’t perfect and they make mistakes, there’s actually less than one bit in that signal, and so since the actual information about the economy is in fact a very rich signal that has an enormous amount of information in it that’s being constructed out of these sub one-bit pieces, there’s essentially an enormous amount of processing that has to go on with the existing model, and that processing that’s effectively just noise. It would be like if you had an engine that you didn’t oil and it overheated and was just sitting there. Basically, that’s what’s going on right now.
Host Renee: Okay, just to sum that up, so it’s kind of like your algorithm is streamlining all that, is that it?
Guest Noah: That’s a fair assessment. The thing that’s tricky is that within the current confines of the existing market structure, people have an interest in effectively lying about what they really want, because if you can sort of fool your counter party into taking a worse deal, then you’re better off and both you and your counter-party are in that condition. So there’s sort of the worst deal that you will take and then the worst deal the bell take, and all the room between those two deals is where negotiation can happen.
Host Renee: Okay, so does your algorithm work in tandem with artificial intelligence? Am I on the right track there with that?
Guest Noah: Not really… With the existing AI revolution is really just based on one learning algorithm that is the neural network learning algorithm and a handful of techniques for training neural networks. So neural networks are a very interesting kind of programming system that can do something called programming by example. So rather than having to write a program that can say, take in things and sort them into the proper boxes by figuring out what rules exist or should exist to close the sorting to happen, you can instead have an existing system that is sorting things in the proper boxes and let the neural network watch an attempt to emulate, eventually learn how to emulate that system until it does.
Host Renee: Okay, yes, and which is kind of what’s happening now with cell phones and the technology and how they track us….
Guest Noah: That’s an aspect of it. With cell phones, like error corrections on your text is basically… It’s learning. People don’t remember the “I before E except after C” rule, except… It’s not thinking about it like that. It’s just thinking about it. When people are spelling this were neighbor with the in in the wrong order, I’ll just switch ’em. And stuff like that, right?
Host Renee: Your algorithm sounds a lot more sophisticated than that actually…
Guest Noah: Well, so yeah, so what I’ve developed is a learning algorithm that can take on board human inputs as part of the system, so it’s taking advantage of the fact that market places while, say the two of us might have some distance between the worst deals, each one of us, we take and have some room to negotiate in a broader market place, the two sides of the markets will tend to converge and mostly will converge on another because there’s opportunities, marginal businesses can be created within that open space between the buyers and sellers, and so they are… And that increases the total amount of value that is transacted through the marketplaces, if say eggs… If only the person who wanted eggs the most and only the person who could produce eggs the cheapest actually made eggs, then the vast majority of us wouldn’t even know what eggs were, but because there’s room to fill in less efficient egg production and less valuable egg consumption there’s a market for eggs that includes millions of creators and consumers.
Host Renee: And your algorithm finds those… Is that correct?
Guest Noah: My algorithm effectively allows those two sides to directly negotiate with each other over those prices in a way that stabilizes over time.
Host Renee: Okay, so since this is the business that we’re in, insurance, I know there is a tentative connection, can you draw that line for us as well as possible…
Guest Noah: Absolutely, so the thing about insurance is that it’s very much an information-based business and again, there’s this sort of two-sided information, a symmetry. An insurance company needs to have better information than each of their clients, pretty much on average, because otherwise they’ll vanish. So if you don’t know what the actual probability of houses burning down are, and your clients do know what the probability or know better than you with the probably if there is burning down or then they will buy insurance products from you that will pay out more than you can afford… And you’ll go out of business.
Host Renee: I get it, yes.
Guest Noah: On the other hand, because your client base is so large in aggregate, your entire client base actually probably does have better information about what’s going on then you yourself have, and this is why insurance companies actually generally invest relatively heavily in actuaries, other sorts of studies, things like that… And they do some client survey and things like… And such like… But again, because of that motivated variation problem, if the client does know something you don’t know, it’s not in their interest to reveal that information to you, and so it’s not that valuable basically, to get the opinions of clients within the strategic context of the way things presently are… Yes, but imagine a scenario where you had a data source where people could provide information into your model for pricing things, which would give them to the degree that their information turned out to be not valuable and profitable, a share of a share of your actual profit stream. That’s providing a positive incentive for them to fess up or tell you what’s going on.
Host Renee: Okay, so what would be an example of that kind of information that wouldn’t necessarily be that valuable or would be valuable for the insurance company from the clients?
Guest Noah: Well, at the most naive level, you’d be talking about setting up a pricing model, and so with my system that’s affected, which could do… One of my insights is that prices aren’t really valuable in and of themselves because time marches on and as time moves on, everything changes, and so one possibility would be setting up your pricing model as a change over time of what you’re talking about with insurance. There might be other things where you’d be waiting different aspects of the system, so there might not be a single pricing model, but in the health insurance context, there might be different weightings of exercise or weight, or habits, drinking, smoking, those types of things. Having a larger, small family, stuff like that, which might be the kind of thing that either for external general comment or as an internal tool for identifying and rewarding super predictors within your own organization could be used to bring together multiple points of view into a coordinated view of the system, while simultaneously recognizing and rewarding people who are making disproportionately valuable contribution towards towards that model.
Host Renee: Okay, alright, so to that point, I’m not exactly sure how they do it, but I do know most, especially the larger insurance companies do have pretty expansive data collection software and procedures and collecting all that… Our insurance companies or the insurance industry, is that a potential client for you or where you would wanna picture algorithm? How do you see that gelling?
Guest Noah: It’s certainly a possibility. I actually have my sights on slightly higher fair really markets on commodity markets, so that’s tens of trillions of dollars in transacted good flow a year, and something like the CME Group notionally handles quirinus dollars in and trade just because of the way I leverage works in the system, my system would allow a massive collapse of that, and that’s sort of the biggest fish in the sea, so since they’re all pretty much equally difficult to fish for, I might as well go for the wall first and then pick up anything else, but I’m open and hope other people would also be open to new opportunities. And the marketplace system is actually the more complicated version of this, if you’re interested in acquiring knowledge from a disparate network, you can use a simplified version of this technique to either commit a bounded amount of money to the pursuit or a bounded amount of money per information to the pursuit, such that error signals will effectively cancel themselves out of the system and allow you to gather information from a large source group at controllable costs.
Host Renee: So it sounds like you’re describing commoditized intent.
Guest Noah: There’s a certain extent to that, so the way I would actually put it is pricing transaction costs.
Host Renee: Well, pricing transaction costs, it doesn’t sound like what you just described as sort of hard costs, meaning like if you’re gonna make a salad, you need lettuce tomato and say onions, but it sounded more like you were describing the intent of the human beings behind the desire to want those items for the salad, so…
Guest Noah: I’m talking about transaction costs in the Coasian way, so
Host Renee: I’m pretty good with words, but that one is…
Guest Noah: Yeah, I can’t remember his first name. Is it Arnold maybe… So this economist named Coase who actually did some work here at UVA in Charles, Virginia back in the early part of the 19th, 1900s and the problem that they were dealing with was effectively, why do businesses exist? If market places are sort of super efficient and make great decisions, and human beings are sort of super inefficient and make bad decisions, why do we have these businesses that are run by human beings making these bad decisions? And so Coase comes up with this concept called “transaction costs.” Something like the net takes, change is awesome, and he’s telling everybody can write answers to everything all the time, but there’s a cost to using a marketplace, to setting it up, to actually paying attention to it, to making the errors and so on that go into that. And so what my system does is allows you to price, basically decide a fixed value for that set of those hard costs that are associated with transaction, and so the notion of transaction costs is that people actually running businesses and making decisions works out because most of the decisions we make don’t matter very much. If I put cucumbers into the tomatoes in my salad, maybe I won’t like it as much, but the cost of setting up a New York Stock Exchange style marketplace to figure out how to make my lunch is going to be so disproportionate that the amount more than I would like my lunch is just not worth the effort, so I can go ahead and make these relatively small and minor choices like where to explore for oil in West and things like that on my own…. without having to set up an entire complex market system to make those tools for me, however, again, thanks to computer technology and the sort of mathematical formalism decisions that you may right now be making informally or just off the cuff, might be valuable enough and might be cheap enough using systems like this to be able to set up these relatively low-cost markets to allow them to be made in an integrated joint fashion and consequently better.
Host Renee: Okay, so it almost sounds like you’ve created an algorithm that eliminates trial by error…
Guest Noah: Well, I wouldn’t go nearly that far because the information still has to exist within the population, so if you’re out in the… If you’re out in the weeds, people are gonna have to be trialling by error, however, if you have such a system, you might be able to spot a reward structure that could distribute the trial and error.
Host Renee: This is really complex. Alright, so it’s almost like you could… I could buy… Let’s say I wanted to start a company, but I have no idea how to do that, your algorithm could find the information on what it would take and the likelihood and the stats before I even got funded for it. To see what it would take to start it and run it?
Guest Noah: I’ll caution you again, so my algorithm is a way to pay for information from the system, so it’s not free, it’s a way to create the incentives that will cause the desired outcomes you’re looking for, so that could be information, so you might be able to set up such something like this in order to learn more about how to start up a business, you might be able to learn more about how to operate your business or learn more about… Or build essentially a better business brain, it could also be a certain stop-gap against succession issues. So one of the things that happens in companies that get to be more than a few people is you get this concept of “bus” number where certain people are critical and if they got hit by a bus, then the company would be in a lot of trouble… If you have these kind of federated decision models, then even people who are in relatively important positions can become less critical when the company can have some continuity on that. But I think the most important thing is that the better decisions that are made by markets aren’t this mysterious force or a pure freebie… The fact is that the more we can use systems like this, and we’ve been using systems like this for centuries to multiple human brains into super intelligences. So the things that people are talking about with AI now about the prospect for transforming different parts or different industries are available to us because we already have access to super intelligences, we know that we can get multiple human beings together on a problem that’s reasonable, specified to make decisions that would be better than any of the people involved in that system can make on their own. And so if you’re in a business, and finance is very much one of these kinds of businesses where consistently better decision-making is going to be a game changer than building out these sorts of systems becomes a strong potential competitive advantage to hold over other people that don’t build such systems.
Host Renee: Right. So at what point in the market process do you see your algorithm fitting.. is a close to the beginning or in the end?
Guest Noah: So my algorithm is a price discovery mechanism, so it’s sort of right smack-dab in the center. The marketplace is essentially a place where buyers and sellers come together. My observation is that that’s an over-simplification, and that deal not only requires the good or service or whatever, and the cash, but it also requires the information that intent you were talking about, where the two sides each have their information about how this is in their interest, and one of the things about now is that we currently understand information itself to be a physical process that we can measure and transmit and so on, and so what I propose is a three-sided marketplace with a dedicated product space that effectively act as a sort of clearing house and aggregates together many, many sellers, a dedicated consumption space, which again is many, many buyers and a third dedicated forecasting or negotiation space, which is aggregating the intent of buyers, sellers, third parties with inside information and anybody else that happens to feel that their opinions are worth backing up.
Host Renee: Wow. Now, this is getting really fascinating, so it sounds also that there could be like any good thing, there’s also the potential for a downside, and so what do you see that have in the wrong hands? How do you see it going wrong?
Guest Noah: Well, on the one hand, there are limits on any kind of technology like this, we can’t predict the future because sadly, that’s not how philosophy works, you can’t predict the future, so this needs to have recovery mechanisms for when everybody’s wrong at the same time. And this does have recovery mechanisms for that, essentially what happens in that case is that the forecasting market adjust its case to buy in, so when… If this market goes off the rails, it becomes much less expensive to correct it, then it was to push it off the rails in the first place, and the people who come in and sort of white knight and save everything, get a windfall profit as a result of having pulled things back on to course. Okay, however, speaking to the other thing, since what this is doing is creating super intelligences, all of the downsides of super intelligences that show up in debates around AI, which can be quite hyperbolic or in science fiction, or even modern fiction these days is AI and robots, you becoming more and more popular, but even just… We have centuries of experience with the markets themselves, if decisions and important decisions are being made by a system that’s more intelligent than any people, then decisions are made in ways that can’t really deep back checked or understood by the people that are going into them, and so… There’s a bravery requirement, if you will, and a certain amount of faith that you have to trust in the mathematics and play the long game, that getting better decision-making means that you’re not always going to agree with or understand the decisions that are made. And this shows up, even among people, there are numerous games of skill and strategy, which have become popular enough to be… Generally, televote have commentators, and the commentators do what they can, but if they could understand what was going on in the game, they’d be playing it, they wouldn’t be telling us about it. And so the world chess championships, that’s actually been an interesting development over the last decade, where for about a century, it was the case that the commentators on the game couldn’t really understand the games in many cases… Right, but now we have these chess AIs that are much, much stronger than human players, and so the commentators now have access to the chess, telling them what good or bad moves are, so they have access to better information than the people playing the game, but there… So they’re not good enough to actually understand what the AIs are saying, so a move gets made, the AI tells them it’s a good move, they try to figure out why a human being might have missed or seen something there, but their knowledge and the player’s knowledge is mismatched, the players may or may not have seen it, and they may not be thinking about the way that they’re thinking about things, so these sorts of issues are very much a part of the adoption of these kinds of technologies.
Host Renee: Alright, so it sounds like quite brilliant in what you’re doing, but it also sounds like it could be quite dangerous or go off the rails, and I know you mentioned that there are sort of stop-gaps and things in place, but as we’ve seen with most technology, there always seems to be some type of hack or work around, do you see a potential there for somebody to be able to… Or a group of people to get in and really corrupt it..
Guest Noah: That would be extremely difficult because at the core, it’s very simple, this system is actually only about 300 lines of code, the system is actually by construction in the common interest of all parties. And another thing to bear in mind is that we aren’t going from people to super intelligence, we’re going from the super intelligence that we have right now, existing markets which are breaking down because they are being forced to process more information faster than they’re really capable of doing the algorithm is what’s known as a red means race, basically, the amount of work you have to do is slightly more than the amount of work there is to do, and so as we get faster and faster computers, markets are actually functioning worse and worse, and this is something that we see in the news almost every day now. Right. So we have, as a society, we built on market place decision making with markets that are degrading because of the technology that we’ve developed, and a society that depends on good market decision-making so that we can get our regs. So the real issue is that we effectively have to transition if we want to maintain this kind of advanced society, which frankly most of our lives depend on…
Host Renee: Wow. Okay, well, I only have a few minutes left here, but this has been really interesting, a lot to think about, and so just one last question, and I guess it’s about the insurance industry, ’cause I do wanna connect that with all the data that’s collected. A lot of it, biological. Do you see a way for your algorithm to track or can be used for biological or physiological purposes?
Guest Noah: It definitely could be. So information is information, and this is a way to make efficient, the gathering integration of information for multiple source inputs. So that’s a definite possibility. Those systems are really in their infancy, and there are other issues with things like privacy and so on, which right now the younger generation doesn’t seem to care about. But that may or may not turn out to be a good decision. I’ve pointed out on other podcasts that for the majority of human existence, most people lived in environments where they enjoyed since less privacy than we have today, because if you’re living in a tribal structure or a village structure, everybody knows everybody’s business. But what’s very, very different is that reciprocity was a foundational part of that type of the system, so the tribal leader knows everybody’s business, but everybody also knows his business. So for example, there was the case of Elon Musk throwing people on Twitter for outing geolocation and the statement… I have security concerns. Well, everybody has security concerns, and Twitter literally knows where it’s user basis because when you’re tweeting, your phone knows where you are and it’s telling them where it is, so that reciprocity is going bye bye and that may or turned out to be safe.
Host Renee: Yeah, no, I don’t think so. Alright, I would love to be able to follow up with you and follow your work a little bit and see how things develop for you, if you don’t mind, I would love that, and… So I wish you all the best of luck on it, and so I’m gonna have to sign off here ’cause I think we have just a few more minutes left, and so… Is there anything else you want to add at this point?
Guest Noah: No, thanks, thanks for having me on and yeah, I’d love to follow up at some point, I just was great.
Host Renee: Excellent, thank you. And I’ll make sure that all your links and all the information are connected to this podcast.
Guest Noah: So coordinated discovery for disc. coordisc.com
END OF INTERVIEW
