Table of contents
About the talk
Speaker: Garud Iyengar
Professor Garud Iyengar joined Columbia University’s Industrial Engineering and Operations Research Department in 1998 and teaches courses in asset allocation, asset pricing, simulation and optimization. His research interests include convex optimization, robust optimization, queuing networks, combinatorial optimization, mathematical and computational finance, communication and information theory. Professor Iyengar received a Ph.D. in Electrical Engineering from Stanford University. He also holds a Master of Science in Electrical Engineering from Stanford University and a Bachelor of Technology from the Indian Institute of Technology.View the profile
Hi everyone. My name is go to die. And are I am a faculty member in the operations research department at Columbia University. And before that, I used to be the associate director for the data scientists found a way, blockchains are revolutionizing, all things data had been a central part of the data Science Institute was doing and my particular interest in this aspect, it's not so much from the computer science aspect, but also looking at it from the economic aspect of why you blocking solution, sometimes do not get adopted, and what can we do to make them
adopted at scale? Is a simple model to understand why permission blockchains not make a mess but permission, blockchains are not getting about it. And identify the reasons why they're not getting adopted. I tried to see if there are ways that we can mitigate. Did the people that I'm going to be presenting today, is called economics of the mission. Blockchain adoption is a joint work with my colleague at Columbia, or joint, and a collaborator from Wake Forest
University for hard, sole Jamal chains on an Ideal data structure for information sharing they provide immutable and prints which means there's an audit Trail for everything that has been done and in particular are better because the entities that are providing the consensus are, in fact connected to physical entities, which means that you can actually put a name on a privacy preserving operations that you can do in the data with such that only certain part of the data is visible. And often only certain functions of the day. In setting up,
all kinds of depredation metrics trying to make sure that various participants in the blockchain are only able to see the data. That is relevant. It's also distributed, which means that from a, from a security perspective, it cannot be attacked and brought down in Corpus form. Collaborate from a computer science perspective, from a perspective of the underlined consensus mechanism. All things are great then, not being adopted in many Industries. And one of the things that we wanted to investigate was what are the industrial organizational aspects that are preventing. Blocked him from getting
in getting adopted in some Industries. And what we doing this paper is a very minimal Network infrastructure, that is able to adopt identify such an adoption failure at able to figure out why these things do not get adopted. And then Words understanding what happens if that works here? We're trying to bring the problem down to its simplest essence. Show me when I get down to the actual problem, I want you to think about a very for example, the game consists of three agents but only two players.
Meaning that are only two players that make decisions agents that get welfare out. So that's the one that plays on Empire to adopt. They both end up being a cost of one unit. That's the one unit there but player agent 3 sorry I correct myself is not clear because they don't take this agency ends up getting a pay off of 4. So the next recycle day off is of two units. Similarly, in this particular style of the players actually end up paying any cost, but agent 3
does not get any benefits. And if you just stayed at this Matrix, for a moment, you can convince yourself of the unique, it really bring them here. Is going to be, don't adopt a dog. Both players do not adopt or take the opposition and Society ends up. Losing two units are. So she had to go to Boston from here, which is a social optimal at compare, and then move to this. Where are discussions began about looking at blockchains because what was happening was the cost of the blockchain was coming from. One set
of entities are the benefits were being externalizing subsets. And unless you're able to, in endogenous internalized, some of those Network benefits of becoming the chances, are that people that go to the This particular example, we can solve the problem by transferring welfare, from the third agent to agent that basically gets all the way out there to the first three units. Off welfare. This cell was four units do 1.52 Player 1 and 1.5 two-player to. So therefore now their wind
5.5 + 3 and similar things have been done in the other spouse transfer a budget balance. That means you do not have to inject any more cash. Does is moved equilibrium from 000, which was not socially optimal, optimal players. Adopting the The solution to become socially optimal. An impacted comes to Unique it. So what I want you to take away from this basic example is there are often situations when the social welfare is coming from. A third agent, was not the decision-maker cannot influence the decision makers of player. One and two are wanting to end up
financing the social welfare. And so there is a possibility that the equilibrium equilibrium did not adopt the solution that could be socially optimal. And one way to solve this problem is to do transfers and then we're going to do is take the story over to A more realistic supply-chain Network and see how this Essence can be. Almost represented in that Network. All manufacturers in this particular example. All of these manufacturers are competing for a
bunch of consumers and there is what we're showing is that when there is information sharing across manufacturers, that is a lyric, which information-sharing happens is a single vendor. And so there isn't really much information sharing that's happening between the manufacturer and the weather. We have these are just called Product quality. Displayed over here, and implicitly would be resuming, is that the consumer makes decisions of which manufacturer to buy from based on the quality of this manufacturer. Get down to the details of this animal.
Windows Windows, Just Produce the goods and ship them to different manufacturers for every manufacturer activities. Is that the Wii manufacturers differentiate themselves is on the basis of quality and will get to Quality in a couple of slides. Right now, if you want to look at the situation, is why in such a setting, blockchains might be useful. And one example that I wanted to keep in mind while you're going through this like the vendors here, they potentially Supply food or food items
that you want to keep some item in line think of spinach. So they Supply spinach to all kinds of grocery stores. More often than not, things are fine. But food can get contaminated and these can, cause serious illnesses. And even if the cause is often known trying to do, Recalls can become expensive because you don't often keep track of exactly which vendor this, but the Spanish came from what other grocery stores was the same Spanish sentence about Jaden able. Sufficient reason of these defects and thereby saving cost. and this is this is the kind of motivation that I wanted to keep in
mind that vendors sometimes produce defective Goods, but the defective goods are not recalled in time you end up getting You end up getting serious application. No cost and manufacturers. Want to avoid and one way to do that is to get information about the defective Goods. Early. So before we talk about manufacturers, we're going to jump quality manufacturers here. Does not mean the fact it does not mean because I have not sold at a decent time so some manufacturers might be better at keeping pressure. True identity of
the source that can determine whether they are able to provide others are able to provide you information about its quality. And like I said, He does not help determine if that's because if that was the case, manufacturers would have not sold at item at all. One thing the BBC is that even though information about qualities welfare, enhancing for the manufacturers of this sharing this information and what BCE to our paper is that the ship could be beneficial for all parties.
Off the consumer Center is enhance and it has to be somehow sent back to the manufacturer. Manufacturers and consumers. Again, just to conceptualize the food, change the food, supply change during closer so stupid. Grocers favorite station, possibly sell contaminated food. If we know that a particular grocery has sold contaminated food, Identifying contaminated shipments expediently, reduces cost cost reputation has information quickly and then handed it over to potentially.
Any beautiful Auditorium from the farmer to Consumer can have also there is a negotiation that happens around food recall as to who was at fault, it happened at the time of transfer from the vendor to the grocer and so on. And so have the supply chain is going to be the chief, would be very useful in the cars can be identified. The cost to be identified early is that if all the groceries in the market around the block than anyone discovering that particular, get
communicated to all the brochures. So instead of each grocery relying on their particular time to Discovery, they all get to see it at the minimum time to discovery of all the all the groceries that are on the supply chain. For the difference is my time or everybody else this second piece which is a minimum of everybody's time and something much much smaller than my own. The food supply chain was a motivation, and it's a very strong motivation for the
three components by signaling quality, I'm going to put all my transactions to verify it did not going to do it, but maybe the third party. Play recommend and set up some sort of things of that, sort levitation, early, things are unequivocally good for me. There is an issue with the attracting customers part that I'll come in the moment, but on the face of it, the first look like good for a manufacturer of Ender, don't block the third one being the bad asthma. Paper progress, be found that even the revenue bar
ends up being a reason for not adopting Roblox. So this is a repeat of what we're saying that he thought the single vendor into simplify the story. So when everybody adopt the blockchain, you see the current time in Incheon is identified without as you can also. See this is not clear from a simple analysis but a little bit more detail. As you can see that the benefits The blockchain also helps higher-quality manufacturers because it reduces the error in the quality signal,
past interaction data, on this blockchain, consumers are able to view it as a result. Top has bad indications for detecting, because now, everything is being done by very few manufacturers and potentially has diversification affected. You were getting from block James, may not happen. And that's another reason why blocking various results that we been able to update. Just to fix ideas. Let me just go and remind you of the network. Is the network single vendor with
trying to understand when, and why manufacturers adult or Downstream, the consumer gets better quality information if the manufacturer's happened to be. All right, so going back to the results section now. Find in this simple, model is the blockchain, always enhances consumer, welfare of useful information on average because the blockchain improves what happened to their quality. And by the way, back in the beginning, is a paper that goes with it. It's the title of the paper is economics of information available.
Online if you just search for it and that's where all the mathematical model has been here. I'm trying to give you the mean messages without getting too much into the details of the mathematics. Enhancement enhancement result is true. Even if the blockchain do not fully reveal benefactor equality, even if it just improve manufacturing quality about Ted goes off. There are no ifs and buts and she must always benefit. Now, we going to understand what happened to the manufacturer that have to decide
whether to adopt consumers, cannot do anything about that, manufacturers going to a blockchain but they can't make that decision for them. That would be fine. Is that blotchiness Belle Tire in Lansing for the manufacturer's only for intermedial? Defect defect detection rates, should be a uniform result and it ends up happening. Investing in the blockchain is no longer profitable because the rates are so low that you're not able to manufacture is able to detect soon enough to
avoid significant definition. If the defect detection rate is high because where it's not so that I can detect it on my own or so low that he doesn't help me. That's the sweet spot at which manufacturers are going to be interested in application. Okay. Intermediate where manufacturers are indeed interested in. What happens there? It turns out that blocks, it may reduce welfare, even when there are no dogs, we are in Jackson braids are decent so that potential that is interested in being on the blockchain. And
even then and then remove the component of the car. So we can say that they are not adopted because it's the cost is too high. Even then it turns out that adopted lock basement. I may not be very good for the manufacturer. No change improve the quality signal because of this. manufacturers that RX and the sort of more compatible end up becoming more separated in the Bible as a result of the higher-quality manufacturers, concentrating demand more and as a result of this concentration of the man, Detection, the
early detection part gets pamper you have fewer. So the way to think about this is that if if earlier that the man was spread out and they were locked and different manufacturers, that work independently looking for for the detection times. And as a result Navigation course, at large enough, you end up actually see a Worst Behavior. So it has to be against. These are all when you look at the Dynamics of these, this Market, it seems obvious that locking should be well fed and handsome, but the entire Society things can
go back. And even if it's recyclable, things are beneficial, and being seen a moment under what conditions, even then it will happen to the manufacturer. Think back to the story example that we had in the beginning with two players lose for the third agent benefits dramatically. And we need to figure out how to do that. How to get that solution back into Here's an interesting and tricky part to the flooded. Blockchain adoption depends on beliefs and related to fundamentals. I'll
unpack the statement in a moment. Where's the phone in equilibrium, if you want, brought change to be adopted than all? Manufacturers, must recorded options when others are. That's just a basic statement. When when a manufacturer is trying to compare two decisions, whether to a dog or not, where everybody else and there's a manufacturer that's trying to understand whether they should adopt or not at all. Not adopting is an all-stock action. Meaning of this is not part of the equilibrium. So one has to understand what
happened to this off. I need to understand the impact of this outfit action. You need to I think, in terms of apocalypse, Indeed, adoption decision. Actually does depend on this offer and we will see that these off that believes end up becoming important and unlike any other belief structures. Adoption equilibrium is like conditions necessary for the manufacturer to potentially education as it has to be done with the customers so that they can be sustained. Blocking, as he said, the doctor, the doctor may not arise. Even
went over. Well, it's been total, welfare is enhanced and this happens because consumers manufacturer of beer cost but do not gain from an auction and that's what we call a market failure. And wheat beer designed in the paper. A system of transfers that can be charged to Consumers, condition lead to block cheddar dachshund as a society. We can design a transfer that can make this happen even if the individual looking inside, from this particular supply chain example, was that matches can provide valuable information to Consumers, but determined Block
in adoption, Washington, ambiguous value to manufacturers, I want may have to, but not many, but in most circumstances will have to Define transfer mechanism that I can take the benefits and transfer it to the manufacturer. Just a few conclusions. Chainsaws value in business settings, almost always societal benefit goes up, this adoption centers and one has to look at the economic structure of incentive carefully design, right? Mechanisms of intervention. Sometimes it has to be in terms of transfers and
sometimes it may also be in terms of changing of Barclays, changing beliefs about the bass player, and one of them is necessary in order for us to get blocked in adopted in various markets. Thank you very much.
Buy this talk
Interested in topic “Blockchain”?
You might be interested in videos from this event
Buy this video
Our other topics
With ConferenceCast.tv, you get access to our library of the world's best conference talks.