Legendary scientist and cryptography pioneer Ralph Merkle joined us to discuss his recent paper on DAOs. Merkle examined how the voting mechanisms in today’s democracies are flawed and how a decentralized, transparent DAO making decisions using prediction markets could create more efficient democratic systems.
Topics we discussed in this episode
- Merkle proofs, Merkle Roots and his early forays into cryptography
- Blockchains as living organisms
- Why DAOs will be subject to a Darwinian evolutionary process
- Why voting is flawed and we need new governance methods to save democracy
- The concept of a DAO democracy
- How prediction markets and futarchy would help govern a DAO democracy
- DutchX: The open, decentralized trading protocol for ERC20 tokens using the Dutch auction mechanism. More at epicenter.tv/dutchx.
Sebastien Couture: Hey, everyone. Thanks for tuning in. So we are all traveling this week and we couldn’t record an episode. Now I know what you’re thinking, this hasn’t ever happened before in 250 episodes, almost five years. The universe must be collapsing. Well, you’d be right about the first part, but we do have something for you. We’re not going to leave you with nothing. Actually we’re going to be re-releasing an old episode, the episode we did with Ralph Merkle over two years ago. It’s a great episode. If you haven’t heard it, definitely listen to it. If you have, think you’re going to want to listen to it again. Yeah, that’s coming up in a few minutes.
Other thing I want to mention, we are going to be part of the San Francisco Blockchain Week. We’re co-organizing an event called SF Blockchain Epicenter. It’s going to be at the Hilton Union Square on October 8th and 9th. And members of the Epicenter team will be there including Sunny. And developers can get reduced rates on tickets and you can find that at sfblockchainweek.io. Hope you enjoy this re-release of the Ralph Merkel episode and we’ll see you next week with a brand new episode of Epicenter.
Brian Fabian Crain: Hello and welcome to Epicenter Bitcoin, the show which talks about the technologies, projects, and startups around decentralization and global blockchain revolution. My name is Brian Fabian Crain.
Meher Roy: And I am Meher Roy. Today we have a very special episode because we’ll be interviewing Ralph Merkle. Ralph has a list of achievements which is so long that it would take me half an hour to read it, but briefly is the co-inventor of public key cryptography, and was the first to publish a paper that discussed asymmetric encryption. He invented Merkle Trees. It’s from a background of the Bitcoin system, Merkel’s signatures which are signature scheme that is potentially quantum proof. After his achievements in cryptography he shifted to the field of molecular nanotechnology where he worked on self-replicating machines. And is currently the co-founder at something called Nanofactory Collaboration.
In addition he is also a faculty member at Singularity University. We’ll be talking about his latest idea called DAO Democracy and how they can be used to go on countries and other organizations. Before we start perhaps let’s have some words from Ralph. Ralph, tell us what interests you back into cryptography and in the idea of decentralized autonomous organizations.
Ralph Merkle: I got interested in the whole concept of molecular manufacturing and I’ve been working on that now for many years. And essentially the conclusion you reach when you work in this area is that this is a very powerful technology that will have major ramifications for everybody and everything. So one of the concerns that I had was I thought about where this technology was going and how it would be used was the state of governance in the world today. And I invite anyone who was not concerned about the state of governance to consider the presidential elections in the United States, which seem to be a fairly timely topic, and that should sober them up fairly quickly.
The problem essentially is that democracies as they are currently defined and operated really don’t have the ability to respond quickly, reliably, and in a sensible fashion to anything new or different. It takes a long time. And even on things that have been around for a while, democracies often stumble rather badly. So the question in my mind was very simple, how do we improve the governance process so that we’ll be able to handle this new technology in a way which is sensible and leads to good outcomes? And that was the basic genesis of the whole DAO democracy concept. And it’s a good idea. It has roots I think in a work that a lot of other people have done as you mentioned, Robin Hanson has been doing work in prediction markets for many years and has also been looking at how to use prediction markets for better governance so it draws on those ideas.
And it draws on the whole set of ideas that have been developed with decentralized autonomous organizations. And it stirs them together and produces something that hopefully will be useful. At the very least, I’m hoping that it will result in something that other people will wish to work on and improve so that it produces something that actually would be used out in the field somewhere and started in some fashion.
Brian: Before we’re going to get into the details of democracy DAOs and some of the implications there, you worked very early on public key cryptography and on some of these fundamental technologies you developed in this context. Can you tell us a little bit in what kind of context and environment did you work on these things?
Ralph: Oh. Well, I was at Berkeley as an undergraduate and I was taking a security course. And one of the requirements for the security course was a quarter project. That was back in 1974. I thought about what projects I could do. And I was up one night thinking about it and I thought, “Well, gee, if you’ve got a computer terminal which is talking to a timesharing system, the security of the timesharing system might be compromised, how would you reestablish a secure link between the timesharing system and the computer terminal?” If you assume that all of the information has been compromised, how do you reestablish security? And that problem, of course, is an instance of public key distribution.
And when I began to think about it, my first thought was, “This must be impossible.” I thought about it and I tried to do something which was rather straightforward. I tried to prove that it was impossible. And I found that I couldn’t prove it. And I worked on it for, oh, several hours. And then after working on it for a while I said, “Well, let’s flip this around. If I can’t prove that it’s impossible, let’s try to do it and see what happens.” I proceeded to try that. Now, having just spent a few hours working on proving that it was impossible, I knew where I was having problems with the proof. And so I went straight to those spots to try and get something that would work.
And I found that it wasn’t too difficult. Within a matter of, I think I was around midnight or something when I realized, “Oh, I could do this,” and I could do it using what’s now called the puzzles method. And the puzzles method is in some sense alarmingly simple, if two communicants want to establish a key, they simply generate random puzzles. And the definition of a puzzle is a problem which you can solve and when you solve it you get a bit of information out of the center of it. But it takes some work to solve it, and you can calibrate the amount of work. The conceptual idea is very straightforward. The two communicants generate random puzzles from a confined space of possible puzzles and they exchange the puzzles.
And according to the birthday paradox, the probability that they will have a common puzzle is amazingly high, so they simply keep generating puzzles at random until they generate one in common and use that for the cryptographic key. There various improvements in the final version, but that’s the essence of the concept. And it shows that you can get an N squared amount of work on the part of any attacker compared with N amount of work put in by the two communicants to establish a key. That was the first public key distribution system and in fact, if you fiddle with it you can turn it into a public key system. And I then spent the next several years trying to get anyone to pay attention and it was rejected.
And there was a reviewer who said it was not in keeping with current cryptographic thinking, which was certainly true, and he therefore recommended that it be rejected. And it was all very amusing, at least, in retrospect it’s very amusing. And eventually after Whit and Marty had completed their work while the reviewers said, “Oh, we know what this is.” Now, that we’ve seen someone else described it who’s not an undergraduate, it’s a very straightforward idea which should be published. And at that point I was accepted for publication and eventually it came out in the communications of the ACM, but only after it had been delayed for three years which was amusing.
That’s the basic story of my original tangling with both the concept of a public key cryptosystem and with the publication process which really is quite badly damaged publication. And the whole review process is almost guaranteed to produce bad results when you look at it carefully. And indeed you find that in a lot of institutions, a lot of institutions that review concepts and ideas, they are very poor at doing that. And that, of course, got me interested in prediction markets, because prediction markets look like they do a better job of reviewing concepts and ideas than the traditional committee that reviews an idea and says, “Oh, we don’t understand this, therefore, it’s bad. Therefore, we rejected it.”
Meher: Did you anticipate that when in the 70s you were working on public key cryptography and Merkle trees that these inventions would become so big as . . .
Ralph: Well, we anticipated at the time, I mean, it was pretty obvious that there would be a lot of applications of public key cryptography and that they would be used fairly widely. I think one of the things was at the time we were looking at a world where commercial security was simply not viewed as important. It was multiple years before the widespread use of the internet for commercial activities or, well, transferring money, financial activities. And once that happened then there was a great deal of interest in improving the security, and of course, at that point the public key systems were available and people began to start using them.
Meher: I’m not sure if you’re aware or not, but there’s a project in this space currently called IPFS, which aims to build something which is called the Merkle Web. The basic idea is that you would take all kinds of data you want to put on the internet as data BLOBs, have unique hashes referencing each data block and link, all of these BLOBs of data in a directed acyclic graph. And this directed the acyclic graph of all of the data in the world if this project succeeds would be called the Merkle Web. Now, I mean, what I wondered is like when you were working on this idea of the Merkle Tree, what problem were you trying to solve? Yeah, that would be the question.
Ralph: Well, the original problem was fairly focused. I had been working on the knapsack, the late lamented knapsack. And it was difficult to extract digital signatures from the knapsack problem in a way that it seemed reasonable and secure and all of those things. I was focusing on a way to get digital signatures that had minimal assumptions about mathematics and anything else. And the use of hash functions struck me as a wonderful idea. And after thinking about it for a while I realized that I needed to have Merkel trees, as we would call it today, in order to reduce the amount of authenticated information I had to pass around.
But essentially it was the signature application that was motivating me at that time. Once I developed the concept I realized it had broad applicability to pretty much everything. And so I went over to the Stanford patenting folks and said, “Hey, here’s a concept. I will write up the basic concept, let’s patent it.” And in fact, there was, how do I phrase it? There wasn’t a lot of interest. I wound up writing the claims myself. Anyone who looks the patent will say, “Oh, it was so sparing in the claims.” Well, the reason it was sparing in the claims because I wrote them. And I simply sat down and said, “Okay. What’s important?” And I didn’t have a full appreciation that in patenting things you generally cover the landscape in all possible ways you can think of, so I simply wrote a couple of claims and said, “Oh, that seems to capture.”
And then sent it off, and the guys at the technology licensing office said, “Great.” And they turned it into a patent and filed it. But that was the core application. And after I did, as I said, after I developed the core application I realized that you could do a lot of things with it. And in particular, the one of the properties that was fascinating was you could have the mafia compute the tree. And if you were using the tree, for example, to have an authenticated list of public keys and the names of the people who own those public keys, then literally you could have the mafia generate it and you could still trust it, which struck me as a fascinating property.
Unfortunately, I wasn’t able to figure out the next step. I mean, it was pretty obvious that this was useful but it was fixed. It was rigid. You couldn’t update the tree. Updating the route required re-computing everything and sending things around again. That looked like the difficult process. And the concept of updating the root, the concept of having a root of all the hashes, and having some algorithm that would update it, I just couldn’t figure out how to do that. And of course, that’s sort of the core of the whole blockchain concept is you do have an algorithm which will allow you to update the root in this delightfully distributed secure fashion.
Brian: Yeah, absolutely. In reading your DAO democracy white paper, one of the things that struck me, and maybe that’s a way to go a little bit deeper into the blockchain topic, is how you described Bitcoin as the first example of a new form of life. Can you explain a little bit why you think of Bitcoin and the systems as living organisms and why that is important?
Ralph: Well, I was trying to get across a description that non-technical people could understand. And when I thought about it from that perspective and tried to strip out all the technical aspects of it, the simplest metaphor that would convey what was going on was that one of these distributed autonomous organizations is a form of life and Bitcoin is the first example of this new form of life. It follows rules. It exists as long as it can do something useful. I mean, there have to be people willing to set up servers and actually put in the resources to carry out the algorithm. And as long as it’s performing a useful service, then it can use its own cryptocurrency and it can use its own cryptocurrency to pay people to carry out the useful services it requires.
It seems to satisfy the core requirements of a living organism and it has the fascinating property of having radical transparency, and in addition to this ability, to carry out any set of rules that you want. It seems to be an easy way to convey to non-technical people what it is that is interesting and advantageous about DAOs or DAOs. I don’t know, I like the DAO pronunciation because it has this vaguely oriental wisdom sound to it.
Meher: Do you think DAOs will also have a process of dominion selection? Will there be dominant mechanisms in the DAO one?
Ralph: Well, clearly we’re already seeing that. I mean, obviously there are a whole raft of competitors to Bitcoin that have been growing up and pretty obviously the selection process among them is entirely Darwinian, so we’re already seeing that. And as we go into the future, I mean again, we’re going to see Darwinian selection processes. Those DAOs that are better able to carry out a useful function will survive, and the ones that eventually shrink to the point where nobody’s interested will vanish. The last server will be shut down and that will be it, it will be all over. We’re seeing a Darwinian selection process and one of the fascinating questions is, what are the DAOs that are going to survive from this Darwinian selection process?
One of the obvious ideas would be to use a prediction market and couple a prediction market to a DAO and then have the DAO use the prediction market to, for example, maximize its own value. That would be an economically-driven DAO. And there are a lot of people who are sort of looking into this whole concept of tying prediction markets into DAOs and making them work. I got interested because I thought you could tie a prediction market into a DAO and use that for governance. And in particular, Robin Hanson’s ideas on governance uses the term Futarchy, which I thought was a really terrible selection of terms. I mean, as a word I think it has all of the wrong association and flavors with it.
It’s difficult to see how that particular term will get widespread usage, although, it’s a very good concept that he’s described. I grabbed his concept, grabbed the DAO concept, gave them a nice name, and worked out a few of the other issues, and glued them all together, changed things a bit here and there to my own liking, and then wrote the paper. And as I said, the primary purpose is to focus on how do you have a governance process which can deal with major societal changes and produce sensible decisions or hopefully sensible decisions. And that looks like it’s at least possible. At least, it gives me hope that that’s possible. I mean, people who look around the world and ask, are people engaging in sensible decision-making in sort of governance often come back with the answer, “No, the governments of the world are not behaving very sensibly.”
The hope is that using this mechanism we would see an improved stability in the decision-making process and that would allow us to have a better ability to cope with very advanced technologies which will turn our world upside-down.
Brian: Yeah. About one description you made of essentially the problem that we have with voting and what’s interesting about DAOs and using prediction markets was that you have first of all, this economic thing factor that my vote doesn’t really matter, right? Then that it just statistically right. The chance of you changing the outcome of election is zero essentially, from a rational choice theory perspective like nobody should have a vote. And then you also have to factor right that people don’t really know much about the issues and have an incentive to make up their mind. And then even when they make decisions how those were implemented is very unclear.
I think you did a very nice job at explaining those issues with democracy. And in today’s world I think it has become abundantly clear to most people just how big those problems are if you look at what’s happening in the U.S. with some rather questionable people having a good shot at becoming a president. Or even if we have things like in the U.K., we had the Brexit decision which now whether one agrees with that or not is a different question. But what’s pretty clear here is that a decision was made with massive implications without anybody really taking those into account and like doing the kind of balanced assessment of, is this going to lead to a better world or not?
But just moving ahead based on instinct or aversion. I think it’s become very clear just how big those problems are. Can you briefly explain why do you think that future key or DAO democracy could solve those problems?
Ralph: Well, the problem I was facing was fairly straightforward. Democracies, as you say, voting has various terrible properties and democracies make decisions where, well, the old joke, I mean, democracies are run by the majority and half the population is below average. There’s that. And then there’s the fact that there’s very little incentive to vote and people oftentimes don’t spend a lot of time thinking about what’s going on. And then there are these people running around who have their entire lives, their entire careers focused on, how does one put it? Influencing the voters in ways that do not necessarily involve telling them the complete truth about what’s going on.
You have a whole raft of factors. The question I was asking myself with this, what we want to do is we want to get very smart people, very bright people. And we want them to figure out what it is that would make sense for the vast majority of people. And we want to do it in a way so that the very smart people don’t wind up saying at some point, “Oh, and by the way I can just take over.” Because if you look, what you find is that previous thoughts about, “Well, you should have very intelligent people making the decision.” That’s just an excuse for the very intelligent people to grab power and then create problems for the rest of us. The question really boils down to, how do you harness the intellectual abilities of really bright people so that they will actually think about what the issues are that would make most people happy?
And this whole concept of prediction markets seems to do that trick in the sense that anyone can participate in the prediction market, anyone can join, anyone can buy. Like the stock market, or like any other investment, if you want to buy a stock or you want to put your money on an idea, you can do it. And if you’re wrong, you lose your money. Suddenly there’s a very strong incentive for the individuals who are involved to get it right. And now, by coupling the prediction market with ideas that are important to awesome terms of governance you get the best and the brightest thinking about, or at least, you get those who have spent substantial time thinking about and are willing to put their money where their mouth is, you get those people actually providing strong information about whether or not a particular idea would work.
Whether a particular bill would work out for the benefit of most of the people in five years’ time. Whether a particular action, whether a particular strategy is going to be beneficial. And I hope that’s not too abstract, but the idea behind the prediction market is that it lets someone else worry about whether it’s a good idea and it rewards them not for what they think about the idea, it rewards them for accurately forecasting what the great bulk of people think about it.
Meher: It seems in your paper you say that we rely on specialists like doctors when it comes to matters of healthcare, specialists like pilots when it comes to matters of flying planes. But in the matters of governance, we rely on the population to vote correctly and put the correct people in government, right? And this is so different from other areas where you expect specialists to nominate, “Okay. You want everybody to participate,” right? And this is, in a sense, like a principal and agent problem, right? All of the people in the country say it like, the principal who would like their lives to improve through governance and they want to select set of laws or set of agents that could propel them in that direction.
What you are essentially saying here is that you could have prediction markets to solve the agency problem that the agent becomes the prediction market or the relation prediction markets give a mechanism by which loss could be made, and very highly skilled people could participate in the prediction market and together form this agent. And then you would need a mechanism by which the, let’s say, the welfare of the principal or welfare of the people would be taken into account by the people who are betting on prediction markets. What would that mechanism be?
Ralph: Okay. The question, for example, would be how do you run a presidential election and use this concept of a prediction market to improve the process? If we look at the current presidential election we see a variety of problems on parade. If we use prediction markets, the question is, how do we improve that process? And the first thing we say is, “Well, if we’re using prediction markets, we have to predict something.” What is it we’re going to predict? And the obvious thing to predict is the rating of the president four, five, six years from now. In other words, we want the judgment of history. What you’re trying to do is decide what the rating of the president will be and we have polls on the president’s performance, how is he doing, and so forth and so on.
You might do something that was stronger than that. You might have, after the president has served in office, you might have an election on whether the president should’ve served in office. In other words, the idea is that well after the president has completed his term, you have essentially a really good poll where you ask everyone what they thought of the president and they rank the president. Then what you do is you say, “Well, that information, that ranking of the president is only going to be available to us in the future after we have completed the selection process.” Now what we do is we say, “Let’s use a prediction market to predict what people are going to think about a particular president after he’s been in office for several years.”
Now, I put it to you that it’s a lot easier to decide in hindsight whether you like the president than it is in foresight. I mean, you vote for a president and then you find out four, five, six years later you didn’t want to vote for him. Or you don’t vote for him and you find out four, five, six years later the other guy did a really good job and you’re very happy with him. If you try to divorce the evaluation after the fact from figuring out what your evaluation will be, you wind up with a prediction market which tries to predict what people will think about the president after he’s been in office. Now, you can run an election. And this prediction market election consists of having a prediction market for each of the candidates.
And what the prediction market does is it tries to predict the rating that that candidate will have a year or two after he’s completed his term. Now, if you do that you now have a series of prediction markets, one for each candidate. And those prediction markets predict the ranking that that president will get. Now, it’s a simple matter, you pick the president or you pick as president the person whose prediction market says that people will rank him the highest after he’s completed his term in office. Now, that process of running a bunch of prediction markets and then picking the top-rated candidate in accordance with the prediction market has a number of obvious advantages.
The biggest advantage is that instead of having ordinary folk trying to figure out all the baffle gap that the candidates are producing you have really bright, really sharp people trying to figure out what Joe Sixpack is going to think of the guy after he’s left office. Now, you’ve got people who are able to spend the time to get through the, how does one phrase it delicately? The promises that politicians make. They can analyze the puffery which is involved in the election process and they can cut through all of that and figure out, “Is this guy really going to be able to deliver the goods?” And if he can they’ll say, “Well, after the election is over people will probably like this guy because he’s really competent and has a good track record.”
Or alternatively they’ll say, “People might be wild for this guy at the moment, but after he’s been in office for a couple of years they aren’t going to like him, because he just doesn’t know how to run a country and he doesn’t know how to run an economy and he’s just going to do a terrible job.” That shifts the burden of evaluating all of the complex issues to a core of very bright people and leaves you with the primary result that that core of very bright people only makes money. They only gain advantage by accurately predicting what Joe Sixpack is going to say after the election is over. And that’s the idea.
Brian: And I think that’s very interesting actually, you go from what you just described here to the concept of DAO democracy, because with what you described here, like one of the issues that I would, for example, say is, “Well, people don’t tend to recollect things very well, right?” They will have a strong recency bias of presidents that might have been not very good, but they made some popular measures in the end and then people have a much more positive memory or people will try to optimize specifically the thing in their polling, how well they poll at the end of their time. And so you got those kind of things, right? But I think that the nice thing in this DAO democracy is the idea that people are not going to evaluate the president but they’re going to evaluate their own life and how happy they are with their life every year.
And then you have, in a way you have this measure of how well are people doing, so if you didn’t have a prediction market. About how well are good people going to do if Trump gets elected versus how well are people going to evaluate their own life if Clinton gets elected all of a sudden, you have a very interesting basis on which to make those decisions.
Ralph: Yeah, so the example I gave was specifically oriented towards something that people are familiar with which is the presidential election. And I tried to give an example and said, “Okay. If you took that process and tried to use prediction markets, here’s how it might play out.” As you point out when I was thinking about it and when Robert Hanson was thinking about it in terms of Futarchy, the question was, how do you get a broader metric? Now, Robin wants to have voting, so his slogan is voting for values, but, I forget what it was, prediction markets for, voting for values, betting on an outcome, something like that.
Brian: On beliefs, I think.
Ralph: Beliefs. I think that the problem with that, though, is it still uses this voting mechanism. And looking at the voting mechanism I really can’t back a process that uses voting. I’d rather have a process that had something else. And as you say, the question is, what is it you’re going to predict? I mean like I say, the core question on a prediction market is you need to predict something. What do you predict? Well, the thing to predict is the collective welfare of everybody and that collective welfare of everybody is perhaps best computed by simply asking everybody what their welfare, how do they feel? Are things going well?
And then you simply add that up and people are, as you point out, they’re very well acquainted with their own lives, how things are going for them, whether it’s going well, whether it’s going badly, whether they would wish to see improvements. You don’t ask people a complicated question like evaluating a president. You ask people a question which they are very qualified to answer which is how they themselves are doing. And then you sum it over all of the people in the nation and you’ve got a collective welfare. Now you use a prediction market to predict whether or not a bill or an action will cause the collective welfare to go up or not. And that really begins to give you something which I think is very good.
Now, the way the DAO comes into it is that if you want to use this concept of prediction market there’s a whole lot of mechanics and you want to make sure the mechanics are carried out reliably and you want to make sure the mechanics are carried out without bias and without people mucking it up in some fashion. By using the whole DAO mechanism you can improve the integrity and the reliability of the core governance process and make sure that what you see is what you get so to speak. That’s where the DAO part comes in. And then you combine it with the prediction market of this core value of collective welfare, and that’s pretty much it.
Brian: The thing that I also found fascinating about your paper is that the underlying structure is, of course, in a way, it’s an implementation of a political theory called utilitarianism, right? Where you say, “Okay. It’s good days if the sum of everybody’s well-being is maximized.”
Brian: And what struck me here is I’m not sure if I’ve ever seen somebody describe a process of how you’re going to go from this political theory to then actually having a society taking decisions that then maximizes and functions according to that political theory. Are there other models of how that can be accomplished or is this the first?
Ralph: I’m probably a terrible person to ask, because my expertise is cryptography and molecular manufacturing, and I came late to the party on this whole political process set of issues. I’m aware of the concepts and ideas that I’ve been directly involved in, but I don’t have a lot of breadth in the political concepts and ideas. I really don’t know if someone else’s implemented ideas in a way that would be a direct implementation of one of the classic concepts of societal well-being. But it is interesting that this is, how do I put it? I mean, it really pushes you towards the concept of, as you say, utilitarianism, the idea that you want to maximize the well-being of everyone.
And what’s also fascinating as I was thinking about it, it turns out that if you ask people directly how they value things you get a pretty good answer. And the other thing is when you think about it, should you be waiting some people more heavily than others? The “all animals are equal but some animals are more equal than others” line. And it’s really hard to justify. And when you get into this whole process, as long as you’re talking about, well, the wise should have a greater influence. Okay. You can argue that, but the moment you get into this process where the hard work of an analysis is carried out by the prediction market and the core function of people when they are providing their input is to say, “Yeah, I like my life. It’s a good life.
Whatever the society is doing is good,” or, “My life is really bad. I don’t like it. Something’s wrong. Do something.” It’s really hard to justify different weights for different people when you boil it down to this level. And you have a natural tendency to wind up with some democratic process. I don’t know if I can express it any more formally than that, but it almost seems as though this is some natural structure that seems to emerge from a set of concepts.
Meher: In a DAO democracy, were the opinions of, for example, what children say about their life, right? In this case, you go to every citizen and ask, “Rate your life,” right? Like, “Rate how well you are doing,” where we also ask children to rate . . .
Ralph: Sure. And the whole purpose of not having children vote today as we say, “Oh, children. They’re not competent to vote. They don’t have the intelligence to understand the issues.” Well, once you’ve factored out the analysis process from the emotive “Is my life going well?” process then sure, involve children. All you have to do is be able to say, “Is this person happy or sad? Are they pleased with the way things are going?” And the whole purpose here is to reduce the intellectual burden on the average citizen while still honoring the emotion of the average citizen. And that’s the tricky part because it’s really easy to get a system where the very bright people run away with the system.
And you don’t want that. You want a system where the very bright people are harnessed to the well-being of everyone and getting that structure is the tricky bit.
Brian: One of the key things in a system like that is the discount factor. Can you explain what a discount factor is and why it plays a role in this process?
Ralph: Well, if you’re dealing with the future then you have to start saying things like, “Well, the future is, 10 years in the future is farther away and it’s blurrier. One year from now is a lot closer, it’s a lot more important. Let’s discount 10 years by some amount.” And this is typically expressed as an interest rate. People normally say, “Well, there’s an interest rate, it’s five percent, it’s four percent, it’s three percent.” And if you’re dealing with this this whole decision-making process you want to have some uniform way of evaluating the future and things that are going to happen. And in particular, if you’re annually taking a poll among all of the citizens, so if you’re taking this annual poll and all of the citizens are providing their input, you want to wait the next year, the next two years, the next five years more heavily than twenty years.
Well, actually, don’t, but it would seem that you would want to wait the nearer term more heavily than the longer term. And certainly when we begin this process it might be advisable to do that so we can gain some experience on it on short time horizons. But in some sense, if people are unhappy a hundred years from now, shouldn’t we be concerned about that? If we’re making a decision where we can actually tell that a thousand years from now it will cause great suffering, shouldn’t we not make that decision? There are a couple of issues here. First off, people have a hard time forecasting things a year, two years, three years from now. Forecasting things twenty years from now is almost uniformly terrible.
If you look at 20-year forecasts you find that they’re almost always wrong, misleading, point you in the wrong direction, etcetera. And that’s not so much a function of whether it’s important to know what’s going on 20 years from now. It’s a question of, it’s just hard to figure out anything significant about 20 years from now. This is captured imperfectly by using a discount rate, and you can argue philosophically that you should care about a thousand years from now, but the pragmatic answer is we’re just not bright enough to be able to make statements about a thousand years from now, or even a hundred, or even twenty. I mean, if we could do a good job of forecasting five years, that would be pretty good.
A lot of society would be greatly improved if we did a good job at figuring out what was going to happen five years from now. And that would allow you to deal with a lot of issues and a lot of problems a lot more effectively than we do now. That argues for a discount rate on the order of five, ten percent. A five percent discount rate would give you roughly a 20-year look ahead. And I go into the math of this in the paper, but the basic idea is we use a discount rate as an imperfect way of saying we have a hard time figuring out what’s going to happen in the future.
Brian: This is very interesting that the way you use discount rate in the paper, because generally in economics the assumption is made at a discount rate reflects a time preference that people have. They prefer having something today from next year just because they don’t value the future as much as the present regardless of the factor of uncertainty and difficulty in predicting things. Now, it seems to me that maybe some people have moral reasons why they think a discount rate should be zero and if there’s certainty then in the future we should value that just as much as today. But I would assume that people are going to have very, very different views on this. Isn’t that a massive problem for this system?
Ralph: Well, I think if you start looking at it and you ask what should the discount rate be, I think most people will gravitate towards something rather plausible. But if your question is, hey, if you implement this thing, are there going to be fights over how to implement it? The answer is, I would expect so. And if you’re talking about a mechanism of governance, then there are going to be a lot of things that people are going to argue about, that they’re going to have very, very strong opinions in multiple ways. I’ve tried as much as I can to reduce the range of things about which you can argue, but in the end, you just have to implement it and see, hey, what happens?
And that raises a number of interesting questions which is, how do you go about implementing these things? I think the simplest way is to start small and perhaps you can start using it as an advisory approach. In other words, there is a concept here of having a collective welfare, and measuring the collective welfare, and using a prediction market to predict the collective welfare depending on whether you pass a bill or don’t pass a bill. And evaluating your actions today, evaluating the actions today based on their predicted results, predicted by the prediction market. Well, you can do that in an advisory capacity without changing the underlying governance process.
You could have an organization and could set up a prediction market, and it could run the prediction market, and it could just use that as input and it could ignore the results if it thought that was appropriate, but it could also use the results and say, “Hey, here’s this prediction market thing. It’s telling us that if we increase our dues by 10% we’ll lose members, but if we increase it by 5% we won’t lose as many members and our growth rate will improve,” or something very quantitative like that. You can ask very detailed questions of a prediction market if you’re trying to figure out whether or not some very pragmatic decision you’re making is going to be a good one.
For a small organization you could simply adopt the concepts of prediction markets, see what happens. And then as you get used to it and you decide maybe it’s a good idea you start solidifying the decision making process and you start to commit the organization more extensively to actually following what the prediction market says assuming that it’s producing reasonably good answers. That would seem to be a natural way of moving forward with putting our toe into the waters and trying these things out.
Meher: Is a DAO democracy as you conceptualized it keep it below for recursive self-improvement? Meaning, I’m assuming that all of these parameters like discount factors and whose opinion counts in the collectivist and metric are implemented as parameters in a cone, can these be changed in the future?
Ralph: The answer, by and large, is yes. In other words if you have a DAO democracy and it’s implemented using a code base then one of the allowed functions would be to ask, “If I make this change to the code base, will it improve my parameters? Will it improve the collective welfare?” And if the answer is yes, and you’d go ahead and make the change. Now, that allows a self-improving process and I think it’s essential to have that. On the other hand, it’s also clear that if you allow self-improvement, the DAO could change in unexpected ways, and it could change in ways that perhaps are not ones that you would want. You’d have to ask yourself how much flexibility should we have versus how much do we want to preserve a core set of values.
You really have to start asking yourself what are the core values that we’re trying to preserve. And then anything that was not a core value would be up for grabs. How you propose bills, the discount rate, all of that, you could adjust and optimize. There is an interesting question which is, given that we’re going to have a lot of DAOs running around and given that they’re going to be based on all kinds of assumptions, we’re going to have a Darwinian selection process. A DAO democracy, a governance process that is based on this would also be part of this Darwinian selection process. And the problem is if the DAO governance which is after all supposed to be the mechanism that’s looking after your welfare, and my welfare, and everyone’s welfare if that DAO, if that DAO democracy which is looking after everyone’s welfare loses in the competitive process that could be a problem and it particularly becomes a problem as you transition to a society where the dominant actors are no longer people.
In all of our history we have dealt with a world where people make it run and we’re moving to a point where that won’t be the case, where you’ll have intelligent computers, autonomous computers carrying out the core functions of society. If you have DAOs that don’t care about you and me, don’t care about people, and they out-compete the DAO which does care about you and me, the DAO democracy, then, where does that leave us? And the answer is in a very uncomfortable place I suspect. You want to have as much flexibility in the DAO democracy as you can because you want the DAO democracy to be as competitive as it can possibly be, because there’s an expectation many people are already running around implementing lots of DAOs, so it’s hard to see how we can have a future where we don’t have a lot of other DAOs that don’t really care a lot about people.
They’ll care about economic results. They’ll care about performance. Let’s face it, you and I are not going to be competitive in the future. You and I are biologically-based, we have brains that operate on millisecond synaptic delays, a few tens of meters per second, propagation velocities, and using componentry which is large and unreliable compared with the kind of systems that we expect we’ll be able to build. At some point we will have systems that are a lot brighter, a lot faster, a lot smarter, a lot more competitive than we are. And there are a lot of people who are perfectly happy to set up an economic system where those people who don’t carry their weight, don’t get to be part of the system.
They’re losers. Well, I got news for everyone, we’re all going to be losers, all of us. Unless we set up a system which looks after slow-moving, not very bright people, which is all of us, we’re going to create a system which will eventually simply dispose of us. Which is not something that, as a human being, I’m very happy about. One of the questions in this whole DAO governance process, this whole DAO democracy process is, will this system remain robust and reliable when you’ve got computers that are a lot brighter than we are? And hopefully the answer is yes. And I think at this point, the conversation and the discussion about these is just beginning. There are a lot of people who write papers about super intelligence, and AIs, and the AI revolution, and friendly AIs, and all of that stuff.
But the hope is that a DAO democracy would be of such a nature that it would remain intact, it would survive, and its core function of responding to the well-being of the people who are composing it would continue to be served.
Brian: That’s a extremely interesting point, and if one looks at it strictly, you would think that if you’re going to have a DAO who design objective and the function it’s trying to optimize is just growth, growth and dominance. And that’s the only thing it tries to optimize. Versus another one that tries to optimize human well-being, it seems inevitable that the latter one is going to struggle against the first one.
Ralph: We get to set up the initial conditions because right now we’re running the show. I would suggest that we arrange the initial conditions in such a way that we have long happy lives. That’s my suggestion. And then as you point out, if you have a DAO whose goals are limited to simple growth then we probably want to give an edge to our DAO democracy in some fashion. Like it gets there first, it owns most of the resources to begin with, so these other DAOs are smaller, and littler, and can be dealt with in some fashion. There are a number of very interesting issues and I’m sure that some of your listeners will understand, these issues are going to be very important for people who expect to be alive as this transition to intelligent computers takes place, which I suspect is a fair percentage of the listeners.
Meher: Now, let me tie in your interest with cyogenics and life extension into this question.
Ralph: Sure. Cryonics by the . . . cryonics by the way…
Meher: Cryonics, yeah. Okay. The basic idea behind cryonics is if I am suffering from a disease that’s going to kill me . . .
Meher: Like living, yeah.
Ralph: We’re all terminal, it’s just a question of how long.
Meher: Then I could either freeze my brain or my complete body in order that in the future if this technology, if the technology level is such that the people in the future can actually resuscitate me, then they would go ahead and do it, right? Now, if you had a DAO democracy system and I was in that position that I had a disease, I would want a system where if I froze myself cryonically, then the DAO democracy would ensure that in the future if there is technology to resuscitate me it would go ahead and resuscitate me. Is there a way you could tweak the system to enable that?
Ralph: Yeah. You just keep saying, hey, when your cryopreserved you continue to enter your annual evaluation of the situation. And my evaluation when I’m cryopreserved is I don’t like the situation very much. And if I reach a point where I’m revived, I’ll like the situation a whole lot more. A DAO democracy which is getting as inputs from all of these cryopreserved people that they don’t like the situation very much and it can wake them up at which point they’ll like the situation a whole lot better, the DAO democracy will say, “Huh, if I wake these people up they’ll like the situation a whole lot better. Oh, I should wake them up.” Now your DAO democracy has an incentive to go ahead and wake up all the people who are cryopreserved.
Get that structure in place and, hey, suddenly you’ve got a DAO democracy looking after your back. Just so you’re aware that this is a good, by the way, for those who are listening, there’s a really good blog on cryonics, Tim Urban wrote an article on why cryonics makes sense. And he put it up on his Wait, But Why? Which has gotten a great deal of popularity because he writes very well and does really good in-depth analysis of a variety of subjects, so I can encourage people to go read about cryonics there. And also if you’re assuming that cryonics won’t work, guess again. There’s a lot of evidence which has been coming in the last couple of years that says, “Yeah, this looks like a lot of work.”
If you want to stay alive in the future and you’re a little bit uncertain about how long you might manage to make it, look into cryonics. And by the way, yes, I am signed up.
Brian: Yeah. We’ll definitely post links to I think also Alcor you’re on the board of if I was correct.
Brian: One of their cryonics organizations and to the blocks of people who have interests in that, which is certainly an extremely fascinating topic should definitely look into that.
Ralph: Well, anyone who thinks life is worth living should probably check it out.
Meher: Now, let us also try to tie in one of your other interests, which is molecular manufacturing, into the DAO democracy idea. For our listeners who have never heard of molecular manufacturing, it’s the idea that like when human civilization started we ended up playing with atoms, and right now we can fabricate things using atoms but we always deal with chunks of atoms like that, a micrometers big or some semiconductors nanometers.
Ralph: Yeah, we deal with matter in great thundering statistical herds, so large numbers of atoms at once. And I really have a hard time getting every atom where we want it to go. That’s a little bit tricky.
Meher: Yeah. And the idea is that we’ll have technology that allows us to arrange atoms any way we wanted to, and this would have massive consequences in the way we do pretty much everything, healthcare, computation, etcetera, right?
Ralph: And if I might intervene for a moment, for those of the less technical sort who are listening, imagine Lego blocks. Nature has given us a Lego set and right now we got boxing gloves on our hands and we can just scoop up the Lego blocks into great untidy heaps. If you got molecular manufacturing, you can take off the boxing gloves and take the fundamental building blocks of nature and snap them together in exactly the way you want. And all of you chemists out there, before you kill me go read our technical articles on the subject.
Meher: There’s a whole body of literature that analyzes this molecular manufacturing, starting from if you like to read about it more, Eric Drexler. And then there are a lot of papers from last month of himself and his, I think, collaborator Robert Freitas, about how this could impact medicine and how this could lead to technologies like regenerating people from their frozen bodies, right? There’s a lot of body of work around this idea. Now I have always felt that one of the reasons why this technology has not come to fruition is that there hasn’t been a focused effort to make sure we get to the point where we have molecular manufacturing, because this is something that requires a lot of money and political will.
Do you think something like a DAO democracy could help prioritize this research, because that could tremendously impact human well-being in the future?
Ralph: Well, the DAO democracy obviously would have an incentive to pursue it because as you say, it would greatly enhance human well-being. And I think that as you point out, a focused project to develop molecular manufacturing would be, I think, critical, we didn’t get to the moon because people sat around and said, “It’d be nice to go to the moon.” We got to the moon because we had a project that said, “There’s the moon, we want to go there. We’re going to do it and we’re going to pour in a lot of time, effort, and energy to make it happen.” And by the way, I think it doesn’t require that much money to get the projects rolling and get things started.
If anyone is interested in making this happen and happens to have a few tens of millions of dollars lying around, give me a call, because, how do I put it? I think I know how to do it and we’ve got collaborators who are knowledgeable about all of this. If you think this is a worthwhile endeavor and you’ve got some money you want to put in, why . . . give us call.
Brian: You’ve written this white paper now on DAO democracy. It’s a super exciting concept and there’s certainly a lot of work to do until that actually gets built and gets realized. What are your next steps for you? Are you working on this? Are you going to be looking into developing some of the tools, some of the code around this?
Ralph: Well, I’m going to be supportive of it. I’m going to stay focused on the molecular manufacturing. I’m going to stay focused on that as my primary activity, but I’m certainly supportive. And I think there’s a marvelous opportunity for anyone who wants to leap in and make things happen to do so. The simplest thing as I said is to take an existing organization and start setting up the prediction market apparatus just using manual techniques, and then using that as an advisory input. The other technique, of course, would be to set up the code base for a DAO democracy and find someplace that really wanted to try it out. And there are a couple of reasons you might really want to try it out.
One is you’re a student organization that likes to try things out. Another would be that you’re flat-out desperate. Bankrupt cities, or Somalia, or someplace like that, now, you might want to try those a little bit after you got the code base up and running, and you’ve tried it out with a student-based organization, but you get the idea. Find some group that is interested in trying it out, give it a shot, see what happens. Write up some code, put it into the student organization, get things rolling, and see what happens. And we’ve got a lot of excitement around DAOs in general, and I think this is one of the efforts where there’s going to be a big payback. The core question is, how can we use DAOs to improve governance?
And anyone who wants to go into that area and do it, I mean, this is going to transform society. So it’s got a lot of pluses for it.
Brian: Absolutely. And the great thing, the excellent situation that we currently have is that there is a money opportunity. People have certainly seen that with the DAO, that was a rather poorly designed attempt in many ways, still they raised so much money. I think it’s quite obvious that there’s a big opportunity in getting this right. And, of course, that would mean a ton of people are going to be working on this over the next years. I think we’re going to see extremely rapid progress there. And that’s the nice thing here is that you will have this rapid progress and being done in an open-source way. Even if people do it for profit reasons, one will be able to be used a lot of that work for things such as a DAO democracy.
We’re at the end of our show, Ralph. Thanks so much for coming on, that was extremely interesting. Thanks for your work and we are really, extremely looking forward to what’s going to come out of this, off the DAO democracy concept and your further work in this field.
Ralph: Well, thank you very much for having me and I look forward to the progress in all of these areas as well.
Brian: And hopefully we’ll see a follow-up paper from you at some point or have another reason to have you back on the show.
Ralph: That would be great.