Abstractive Divergence is an Existential Risk

Transcript

hello good afternoon i'm going to risk the ire of the french government and take off my mask please don't report this i see i have intimidated everyone so much with this title that they all are waiting outside the door waiting for me to explain what abstractive divergence is and then deciding whether or not to come in uh unfortunately i'm not going to provide them with an immediate explanation you'll have to wait a few slides but it will come i promise so i'm christopher from inoma and i'm going to be talking about abstractive divergence is an existential risk what is abstractive divergence and why is it an existential risk hopefully in a few slides you will have some idea or you know you might be terrified but at least still have some idea but first i'm going to start with a story a story of economic mismediation so let's say

you're getting up in the morning maybe you live in new york maybe you live in paris let's say let's say paris and you go to a corner store you have a little bit of time before you start work and you want to grab something to eat and a bit of coffee some caffeine to help jump start before you go to your nine-to-five job at the corner store you have some options you know we lived in this kind of capitalist economy you have some nice options to pick from a little bit of illusion of choice um at the corner store say you can choose between three kinds of coffee there's a brazilian dark roast there's a nicaraguan medium roast labeled fair trade and there's the house special of the three the dark roast is a bit more expensive how special and the fair trade seem to be about the same price

and you yeah fair trade seems good i want fair trade i want people to be traded fairly labor conditions important okay so you pick the fair trade coffee get your cup of coffee then you see a few energy bars there's a hazelnut energy bar there's a choco toffee energy bar and there's a blueberry energy bar and the blueberry and taco toffee are a little bit cheaper hazelnut one a little bit more expensive you want to avoid too too much sugar so you take the blueberry one glance the label hope that the sugar is from the blueberries not added afterwards eat your bar uh drink your coffee and go about your day it seems simple right you've got a coffee you've got an energy bar but involved in this interaction which to you is just a matter of choosing between options at

a corner store is a whole network of economic interdependency and it so happens in this case that the reason the fair trade coffee was the same price as the house special was because even though it is it is fair trade let's say that the label is accurate it is fair trade that the uh farmers in question are extracting nutrients from the soil they're mono cropping so because in order to compete while remaining fair trade in order to compete with the house special this coffee producer chose to grow one kind of coffee in the same area of soil for a long time so they'll only be able to grow for about five or ten years and then that soil will be exhausted those nutrients will be exhausted but fair trade doesn't say

anything about that so you don't know and they're able to compete on that basis so in in in the course of purchasing this coffee you're in fact contributing to some kind of nutrient extraction from soil of what they're of which there's a finite amount finite amount of variable land it also turns out the reason the two energy bars the what was it blueberry and toffee energy bars were cheaper than the other one is that the companies involved found a way to manufacture uh wrappers for their energy bars in a way that dumps the pollution into the saturn and the other energy bar was a little bit more expensive because that company chose not to do that but it's not able to do it as cheaply because it can't just dump its solution it has to say

store it safely or tries to come up with a manufacturing process for trappers that doesn't produce so much so you aimed to buy breakfast right you just wanted breakfast and coffee you didn't want to think too much about it but in those interactions you are contributing to the continued existence of those companies and these practices which are not creating a world in which you want to live in uh that's contributing to say the extraction of nutrients from soil which will mean that food gets more expensive in the future or which will mean that the soil can no longer support say trees or a crops which would capture some carbon and you're contributing to pollution in someone else's food or in perhaps even the future water which your children will drink if it's the sun and you live in paris so what's going on here you're using

money and you're paying with money and you're getting something that it seems like immediately is what you want you wanted the energy bar something to eat and you wanted the coffee something to drink a little bit of caffeine but you're also getting these long-term impacts which aren't what you want and if you were to in fact sum up the net impacts of all of these interactions maybe it so happens that uh as a result of buying stuff you're actually making your life worse right that like the net damage maybe not just from your contributions but from everyone's collective contributions and purchases in this fashion is making your life worse so if money is in some sense representing value then something has gone very very wrong because you are paying with money you are using money in order to get something and that which it is what you're getting

not just at the point of purchase but in total is in fact uh perhaps uh a worse quality of life so something's gone wrong there's some uh mistake in this economic system that leads to these kinds of interactions having a result that you really didn't intend but in order to analyze what this is i think we need to dive into what money itself is doing and how it is allowing these complex economic supply chains to exist because as some what we what we have here is a kind of result about the net effect of all of these uh all of these complex interactions but we don't yet know what the contributor contributing components what are the key causes of

the generator functions so in order to investigate this i want to dive a little bit into coordination itself because i'm going to argue that money is just a particular form of coordination and that it's a form which has some really fundamental flaws so what is coordination i take coordination to be very broad say a group of hunters a group of hunters and a tribe coordinates with each other in order to encircle a deer which they want to kill and then eat and in order to encircle that deer they need to communicate with each other these hunters are arranging themselves in some kind of circular position because they want to prevent the deer from escaping and then they'll narrow their circle inwards until they catch it so in order to coordinate they need to use some kind of language they need to

talk with each other a perhaps or they need to use hand signals and it's important in this case that the deer doesn't understand because if the deer understands what they're doing it can probably run away ahead of time so there's one case of coordination another case of coordination might be some some divers say deep sea divers who are trying to investigate shipwreck and those divers need to lift heavy pieces so they're underwater maybe they've got their oxygen tanks they're going down and trying to lift up oh this piece of the titanic that looks like maybe they're even scavengers this piece of the titanic that looks like it has some gold in order to lift that they need to work together because it's too heavy for any one of them so they need to single signal to each other and because they're under water they probably can't use language and

they don't have wonky doggies so they instead use hat signals they're coordinating another case perhaps something which some of you have done could be community members coordinating to build some kind of public good let's say people in a town who want to build a new school for their children and they're coordinating in this case in a written fashion so they're writing down potential lesson plans they're writing down potential architectural diagrams for this new school they need to come to agreement because they need to work together to fund the school or to provide the resources required to construct it in all of these cases this coordination implies a distinction between what i'm going to call an internal and an external state so the divers have some internal state

like they have plans as to where they're going to move or where they're going to put their foot next the hunters have internal state they have plans as to which direction they're going to walk in or where they're going to shoot the arrow and the community members have internal state their preferences as to what kind of school should be built or what should be taught there in order to coordinate then these participants in these mechanisms need to transmit information to each other about their internal states because what they need to agree upon is some like future plan uh that at least uh uh talks about or contains information about what their internal states are going to be so the hunters need to coordinate so that they can plan out their actions in unison so that they can go in uh you know a coherent uh

inwardly circle i have no idea how understeer i'm vegetarian so i'm just making this up so they can go in an inwardly circling fashion and encircle the deer and the community members need to coordinate on the architectural diagram so they you know build a school which fits together instead of having like little bits of wall uh and no protection from the elements and the divers need to coordinate so that they perform the actions at the same time otherwise they would one would start lifting the shipwreck and the other would not be there yet and it would turn over and open up a nest of undersea uh i don't know uh lethal occupied octopi octopi right so there's an internal and there's an external these coordination mechanisms are

embedded in a situational context and in all of these cases there's a particular form a particular medium or a form of language in the case of the divers it's the form of hand signals in the case of hunters it's the form of say spoken language they're yelling to each other and in the case of community members it's written language this form needs to suit the context at hand it wouldn't really make sense the divers obviously couldn't exchange notes to each other underwater and they can't really speak either it won't go far enough and the hunters couldn't really pass notes around while they're trying to encircle the steer while the community members were they to try and communicate or coordinate orally in order to build the school would run into playing a game

of telephone it's quite difficult to agree on precise properties of architectural diagrams when all you can do is state numbers and try to remember what that translates to so the form needs to suit the context but also in each of these contexts there's some sort of shared goal or there's some sort of potential benefit shared by all the people or all the participants who are coordinating so the hunters want to work together because only in virtue of working together will they be able to encircle this deer capture it then eat it and split the you know split the venison uh the divers want to work together because maybe they have some prior agreement to split the proceeds of this of this treasure wreck and the school members sorry a community members want to work together to build the school

uh because they all have children uh they want to be educated there so there's some kind of shared mutual benefit which is which only can be realized through this kind of coordination right so i argue that money is another kind of coordination mechanism but money is a kind of coordination mechanism which has some particular rules and other kinds of coordination mechanisms also have rules so you'd say language has some rules of grammar they're vague sometimes though there are institutions or agents which exert varying degrees of control over the enforcement of the rules of the mechanism itself so and i'm going to butcher the french but uh an example here of the language is the academy francaise which tries to

sometimes come up with new words in french uh to to discourage i suppose young french people from using hashtag uh as far as i know they have not done so successfully but here's an example with language of some some body body politic trying to control the rules of the coordination mechanism itself with money this is a little more obvious there are central banks and governments which have direct direct policy control over the rules of money committed etc when there is a locus of control when there is an entity who is controlling the enforcement of the coordination mechanism if the participants who want to use that mechanism aren't uh don't themselves have control over it or are located elsewhere be that geographically or commercially in some kind of a space of mutual influence there's a

possibility a fundamental possibility of divergence a concrete case of this happening is in say communities in south america using the us dollar because they don't have access to a stable currency but of course the monetary policy of the us dollar is controlled in uh you know by the central bank in the united states which has frankly little knowledge of and certainly doesn't care much about uh whether or not the policies they choose are in the best interests of or even not harmful to the users of the us dollar in south america so when considering coordination mechanisms and in particular when trying to design one as we're going to do throughout this talk we need to think about whether if there is a locus of control of the rules which need to be enforced

is that aligned with the participants of the coordination mechanism itself now let's talk a little bit about money i argue that money is a mechanism for coordination and to demonstrate this i'm first going to give a really simple example in particular money is a mechanism of coordination for the exchange of physical goods so let's say that we have alice bob and charlie alice bob and charlie live in a small community and they engage in various kinds of farming practices which i also don't know anything about i freely admit so uh the examples are a little bit hypothetical let's say that alice uh has some chickens and alice produces sex and bob bob has some cows bob produces milk and charlie uh grows wheat and produces bread and alice

uh would like some milk to go with her eggs bob would like some um wheat or sorry some bread to go with his milk and charlie would like some eggs to go with his bread why they want this for breakfast i don't know but it is uh plausible at least example however since none of them have a sort of bi-directional matching uh desires and offers they can't trade directly uh alice can trade uh with bob bob can trade but charlie charlie can't trade with alice so there's no that there's a potential a potential like share gain right if in fact we could realize a situation where alice gets some whatever it was alice gets some milk

bob get some bread and charlie get some eggs they would all be happier even if they traded a little bit of their own resources for it but without a mechanism that isn't realized so if we introduce money into this scenario then alice uh bob and charlie can buy and sell the goods which they have in which they want so alice can sell sell her eggs bob can sell his milk and charlie can sell his bread while alice can buy milk bob can buy bread and charlie can buy eggs and if they price the goods which they're buying and selling at something like that which they value them at then the market will clear and everyone will get will become slightly happier off in this particular case of physical goods for breakfast so here money is serving as a coordination mechanism it's transmitting information about what people want

and the mechanism and the sort of enforcement of money or the the tie between the exchange of money and the exchange of physical goods allows for the realization of this preferred state for all involved but money is a mechanism has some pretty fundamental limitations so one of the limitations even in these simple scenarios is that money can only express a sort of linear or independently related independent preferences if i prefer uh i can i can only price single goods right without some more complex system of contract i can price the eggs i can price the bread i can price the milk maybe there's cheese but if it's more valuable to me to have a grilled cheese sandwich than the price of bread and cheese those

are not independent variables they're actually dependent on whether i can get both at once but i can't express this just with prices right so there's a pretty fundamental limitation but the other limitation of money comes into play when we're talking about uh more complex supply chains so to go back to my earlier example when you buy coffee or when you buy an energy bar or when you good evans when you buy a cell phone the uh those purchases uh at the end you're sort of a consumer at the end of the supply line and those purchases uh uh include interactions with say a network of hundreds or potentially tens of thousands for a cell phone different suppliers and at each layer of the supply chain there are some suppliers who are say competing with

each other to manufacture memory or to manufacture uh blueberry grow blueberries or to produce some kind of component for the thing which you're buying however even though those components may have a lot of variation certainly the production practices can vary the actual impact which those producers have upon the world with their actions can vary a lot you only see maybe some limited specification or some nutrient facts label on the energy bar into price so that's a reduction in information you care about the total impact uh you care about the world in which you live and you care about the uh uh a net uh uh i guess causal influence which this interaction

has on that world but you only see when you purchase something the price so it's a kind of dimensionality reduction furthermore when you add when you combine that dimensionality reduction with competition at each stage of the supply line you run into what's called a multi-polar trap a multi-polar trap is where there are several suppliers uh say who are competing to produce something which is roughly fungible so they're competing to produce rubber for a tennis shoe or they're competing to grow blueberries for an energy bar they're competing to produce coffee for a cappuccino most of the reason for choice say in the scenario is decided by price

so mostly people are going to buy the cheaper coffee if the quality is roughly the same mostly they're going to buy the cheap pretend issue if the quality of rubber is roughly the same right but then these companies have choices because of this dimensionality reduction the price is not capturing everything that's going on and wherever these companies have a choice between some kind of production practice which externalizes some harm so which dumps pollutions into the river or which chop down chops down trees from the amazon or which exploits local labor or which even just like you know extracts local resources in some kind of unsustainable way if employing that production practice allows them to manufacture product more cheaply they will be able to out-compete all the other companies who don't do

that so if there are five rubber manufacturers even most of them are nice guys even four out of the five uh you know aren't going to cut down the amazon just to make cheaper rubber because they say oh you know that's awful we won't do that if one of them is willing to defect if one of them is willing to start using that production practice they will end up with rubber that's you know even one two three percent cheaper and they will out to beat the other companies and probably drive them out of business so in your engagements with a complex supply chain where there is this kind of dimensionality reduction in competition at each step you are contributing to this kind of uh this is what i call that abstractive divergence this kind of uh chaining process uh by which the

factors of natural selection are selecting for behavior in which companies or producers externalize as much harm as possible because whatever harm they can externalize without it being reflected in price or in factors that affect your choice uh the better they will be able to out compete each other and the longer they will continue to survive in this competitive capitalist system so this is the kind of uh uh then uh explanatory um i guess explanatory mechanism which i think we need to address the scenario we need to take a look at this kind of dimensionality reduction which is happening in each stage

uh since this is causing the divergence and we need to say can we construct a mechanism which allows for the for the realization of this coordination which people actually want so you want to buy energy bars and you want to buy coffee without this kind of divergence at each step from say your long-term preference so there's information here that's getting lost and we need to find some way to factor that information so that as someone who's interacting with this complex supply chain you can start to factor these variables but here we went into a very very tricky dilemma relating to time all dilemmas related to time and that is computational reductionism is useful

if i go to the store in order to buy coffee and i want complete information about everything which is in this coffee uh i'm going to need to read like a 600 page binder that goes over all the production practices and i need to know about soil ecology and 19 to know about the local geopolitical history of the region and i need to understand potential labor tensions and i need to understand even the shipping infrastructure responsible for getting my coffee from nicaragua to france if every time i have to every time i want to buy coffee i have to read this very long 600 page binder uh i'm not going to buy any copy or much more likely i'm going to give up and go to the other store which doesn't require that i read this binder and just pay two dollars for my coffee and go to

work right so the very reductionism which is that this computational productionism the reduction to a price which is so damaging which is an essential component of the subtractive divergence is also critically useful because people have finite amounts of time so we need some way to have our cake and eat it too or we need some way to allow the participants in these mechanisms who are engaging in commercial transactions with complex supply chains to factor all these variables which they care about while not needing to do it by hand every time to me that implies that we need some way to hide this computation or we need some way to

outsource it such that users can articulate they can articulate preferences that factor everything or at least that factor enough that factor these impacts which they care about but mostly in most of their interactions in everyday living uh those uh that preface enforcement kind of happens behind the scenes it's not it's not something that they need to think about or read a long binder every time they want to make a purchase because of this divergence uh then i argue that we kind of have we have three options so the the essential divergence here to attach back to it this abstractive divergence is a result of the combination of the dimensionality reduction of money competition

and really long complex supply chains so we have three options one thing we could do is go back to living in tribes to be honest i don't think it's a bad option i mean it's probably better than nuclear weapons in nation states right like there's a lot of diversity in tribes i mean you could grow a lot of mushrooms you could wander through the forest you know you wouldn't have to deal with uh cellular data connections or taxes so much not a bad option but we'd have to give up friends who live all across the world we'd have to give up some of these cool technologies we'd have to give up i don't know electronic music we'd have to give up spaceships perhaps we want a way to keep these things but still it's an option another option is that we keep going on our current trajectory

and as in each stage of the supply chain where this kind of multipolar trap happens we're continuing to externalize harms the harms will eventually accumulate and the carrying capacity of the earth is finite so that risk is existential if we keep going along this way humanity will in its current form and not necessarily go extinct but it would but the sort of global commercial interplay would end um whether due to catastrophic conflict or simply uh uninhabitable locations or both at once it remains to be seen but that would be option two the third option uh and perhaps the most optimistic option is that we come up with some other mechanism so we find a way to uh get what it is that we want out of money uh namely this kind of coordination

across really wide networks of people who are engaging in various kinds of economic transactions or exchange of physical goods without this dimensionality reduction while somehow solving this computational reductionist dilemma so are actually two questions here and it's important to understand that they're distinct in this design of a potential coordination mechanism the first question is whether this exists at all is there a stable equilibrium where say seven billion people can coordinate where they can engage in complex economic transactions uh with something that uh doesn't have this kind of obstructive divergence doesn't have this

problem that money has even if the answer to that question is yes it's not enough because the second question is whether we can get there from here most coordination mechanisms have some kind of network effect because you're using the coordination mechanism in order to work with other people other participants you know the us dollar or the euro is valuable in some sense insofar as you can use it to get stuff you want in virtue of mostly being used by a lot of other people and if in order to start using some new mechanism you have to immediately completely give up uh the old mechanism it's not going to be very appealing no one will be able to do it and you know whoever could do it uh

would sort of would not be able to compete in some sense with the existing system for participants so we need to answer two questions does a better equilibria exist and can we get there is there some kind of path in the mechanism design trajectory space which is upwards or it's like mostly upwards maybe there's some little windy parts but it's acceptable some some you know ideological backing plus um not too much uh of a of a sort of reduction in physical resources or in comfort or something like that does that exist now i want to talk a little bit about what this kind of mechanism needs to look like so what are we

doing with money when we're engaging in these transactions well the first thing we're doing is that we're expressing our preferences we express preferences by making purchases when you buy something that's transmitting information up the supply line to whoever's producing stuff about you know the fact that people want this thing and the way in which you know money is spent uh sort of means that that information is accurate because like uh uh you wouldn't you the action of buying something requires that you spend money so you can't like fake it or so to speak so that's transmitting information about what you want then we also need some way for um all of these like we need some way for people to find

each other who wants to potentially trade we need some way to uh like computationally i guess well not only communicate information about preferences but also settle the actual transactions so when you money has some kind of some kind of rule set right it has some kind of you know you can only spend to find what you have and you have to conserve supply and we need to find some way to whatever this hypothetical mechanism is uh where something like something like it it's less clear in something like language but it still exists right like the state of language is in our minds and all it's not exactly you know language doesn't have an exact state we don't have exact consensus on it but most of us in the room probably speak english

some of us me excluded probably speak french and one can say that english or french as a set of rules and a set of meanings is some uh you know fuzzy sum of all of the states and the minds of the people who speak the language right french is not what's written down in the you know in the dictionary nora's english it's like it's it's the language is that which people speak in the case of a economic mechanism we need a little more precision and especially in this case where we want to more accurately reflect preferences we need computational precision i argue that the way that we can address this computational reductionist problem is to have preferences be computational instead of just having instead of people

just choosing to purchase or not to purchase something we need to be able to write down as a program what it is that we care about so that when we're engaging in these economic transactions because we can't think about this every day when we're engaging in these economic transactions we can engage in those in a way which conveys our more complex preferences about the world uh while the computation to figure out whether those preferences are actually being realized by the transactions in which we're engaging is outsourced so it's outsourced to uh some kind of computational system which will end up being a blockchain go figure um that does most of the heavy lifting for us while still allowing us to choose

this needs to be transparent you need to be able to transparent in terms of rules at least you need to be able to see what it is that's uh going on and you need to be able to change your preferences but most of the time you don't want to be privy to all of the exact details i argue that the system also needs to be private and my argument for privacy or for privacy of interaction is going to be a little bit different or perhaps a little bit more abstract in the current system of say monetary commerce and coordination there are a lot of um asymmetries or like uh you could you could treat this you could say wealth inequality represents at least a form of this

but there are cases where some people would benefit from keeping the existing system as it is and without privacy this argument is quite general it doesn't it doesn't even depend on this particular case you could have another case of trying to switch coordination mechanisms in which this would also apply without privacy those who start to use the coordination mechanism can be specifically targeted so what we want to protect here is this if i go back a few slides what just one slide is this second question we want to predict protect the feasible path from where we are at the moment to where we want to get and in order for that path to be feasible those who start using the new coordination system which might otherwise be targeted because they might represent a threat to

the dominant position of those at the top of some hierarchy in the current system uh you know without privacy they could easily be targeted so privacy protects this kind of nascent emergence of some alternative equilibrium so i went boy how many minutes 25 minutes through this talk and i've not yet said the word blockchain except once and saying that i was going to say the word blockchain now i'm going to say the word blockchain accepted i don't even like that word because it's just a data structure i like the word distributed ledger so i work on enoma which is a distributed ledger and a set of associated protocols uh which as you might have guessed aims to provide a alternative coordination mechanism uh in the as a sort of part at least of the third

approach as i outlined earlier finding another form of coordination that doesn't have the same problems that money does so we can continue to engage in something like these current networks of global commerce which we like without externalizing harms uh you know repeatedly what is a nova so anoma is first and foremost a system for digital preference expression users using enoma can write what are called validity predicates which are a little like smart contracts and that they're programmable but they're designed to express the kinds of state changes that the user wants or the user prefers so as a user of enoma i encode my preferences you could say in a validity predicate and the exact mechanism which i do that with which i can do that

can vary in part because my preferences are going to change in time and sometimes they're going to change very rapidly sometimes they're going to change very slowly i probably you know don't want carbon to be emitted that's a very slow changing preference maybe i want an energy bar now and i don't want one an hour from now that's a very fast changing preference so some of this preference expression will be done by sending transactions to a ledger some of it will be done by citing things but either way validity predicates allow users of inoma to express computationally that which it is what they want the ledger itself then provides a kind of settlement system which settles state changes as long as everyone involved is happy with a new state so if we go back to our earlier example of alice bob and charlie with what was it eggs milk and bread

then alice bob and charlie would write validity predicates that express that uh they have eggs milk or bread and they want eggs milk or bread no explicit no no money is required here and no explicit coordination between alice bob charlie is required here they just need to write down which states they would prefer right so there's a way of representing eggs milk and bread on the ledger and alice bob and charlie say that they would prefer a situation in which they had slightly fewer eggs and slightly more milk something like this and if there exists a state which is better which is like pareto efficiently better then the ledger allows that state to be changed so all these validity predicates return yes uh alice bob and charlie accept the change and

that state changes act as soon as they do but first in order to especially for more complex trades in order to even figure out what they should be talking about or what kinds of state changes are possible these counterparties need to find each other so the other thing which this and it's not just money that does this is money and meshed in a whole uh superstructure uh of legal contract and of uh commerce and advertisements and lots of you know good and not so good things but alice bob and charlie need to find each other and say were they trading not physical goods in which case they probably have some kind of local connection uh but rather digital goods or something more abstract they need to discover each other so

in order to facilitate this inoma has what we call an intent gossip system where users craft pieces of data that express what it is they want and what it is they have what kind of trades they're willing to execute and they send those around so unlike transaction gossip which aims to gossip all the transactions everyone this is like a sparse gossip network so users are writing down their intents and they're sending it around to anyone who might potentially be interested in being a counterparty those intents then are assembled by entities we call matchmakers into transactions so if i am paying attention to what alice and bob and charlie are telling me i might see that in fact i can create a transaction which sends uh you know eggs milk and bread around

appropriately and then i can use those intents which are which are bindings they contain digital signatures i can use though into those intents to craft that transaction and i can settle it against the ledger so this is how we facilitate counterparty discovery earlier i talked about how we want these interactions to be private uh in part uh so that participants in this new coordination mechanism can't be asymmetrically targeted in order to allow these interactions to be private and nomi uses zero knowledge proofs and the the basic conceptual way in which i think it makes sense to think about zero knowledge proofs is that the ledger simply verifies adherence to the rule set it doesn't need to know anything about

content so the content and here uh you know the content is something like as adrian mentioned in stock the content of something like messages versus metadata is fairly obvious uh where in financial transactions it might be a little bit blurry but when i say content here i mean all of the details of the actual transaction so that which the ledger verifies is just like the rule set in the case of money is something like was supply conserved in the case of say trading eggs milk with bread it's like did in fact uh you know alice bob and charlie approve the state transition but the contents of that state transition even the assets like eggs milk and bread and which went to whom those are all private to the ledger this is also important because in a sense the ledger is something that is far away

from the actual commercial transaction occurring especially if the commercial transaction is local it's an exchange of eggs milk and bread between three people in the same town so we want to send as little information far away as possible shelling binding so one question someone might have is okay so let's say let's say this ledger enforces these rules and let's say people can write down preferences that say that they don't want carbon costs to be externalized they don't want pollution to be dumped in the water whatever what prevents me from just you know ignoring it what if i uh say lie about this or what if i just choose to use another system specifically when i want to engage in these kinds of transactions or specifically when i'm a producer and part of the supply chain

the answer to this question is a little bit tricky because we also don't want to create we don't we don't want to create too much binding right i mean the the system of legal contracts and money uses force as the binding mechanism so if you don't have if you don't pay for a ticket on the subway or you don't follow the rules of capitalism you can be eventually put in jail there are several steps in between but ultimately force is what backs it up um if we can avoid it i think we should uh we don't necessarily i would argue need that kind of um because force also comes with risk and like armies and nuclear weapons and accumulations of essential accumulations of power and generally bad stuff so we would like a binding mechanism that doesn't require that much

uh oomph i think we can get this because participation in the economic system itself is valuable to people so what we want to do is tie so if i want if i want to particip think of it as like playing a game so when i play a game of chess i sit down with my friend alberto and we play a game of chess and we start moving pieces on the board and i'm losing and i really hate losing a chess right so i just i'll bet i've exchanged queens and i just take my i take my queen and i put it back on the board and then we keep playing and i win the game why alberto lets me keep playing i don't know if i want a game of chess i think the answer is no because the moment i change the rules the moment i put the queen back on the

board i'm really not playing the same game anymore this is the kind of binding that we can get with distributed letters as soon as i change the rules i'm no longer in consensus with everyone else i'm no longer participating in this system and i argue that that's enough that if we have producers and consumers who want to participate in some economic network together choose to use the system and in virtue of doing so they have to agree to these rules to say internalized carbon costs as people have expressed their preferences they can always choose to use some other system but then they would no longer be able to interact with these other participants just like i can always choose to speak some other language but that i'm no longer going to be able to communicate with people who speak the language that i don't want to speak anymore

of course in order to actually uh influence these real-world transactions we need to get data about the world into and out of the letter just as if you have some kind of language which is only written down as a set of arbitrary symbols without connections to the world as it acquires say in people's minds for natural language uh that language you know doesn't really refer to anything it is meaningless you could say similarly we need ways for the uh ledger or the preferences expressed upon the ledger the assets or physical goods represented in the ledger to acquire meaning in the real world and we need this in two directions the first direction is usually what's termed as termed as the oracle problem so we need to get data about the world about say

the carbon that produces or emitting or about uh even the supply chains themselves the structure of the supply chains themselves onto the ledger the best way to do this is to embed say the the commercial structure of who's paying whom uh in this complex supply chain uh directly by having the ledger be like that which they use to settle it because then it's hard to fake uh and for stuff like uh oracle reporting of pollution um perhaps you need some kind of you still may need some kind of trusted party you need some kind of uh uh shelling point game like commit reveal reporting you need some way of getting that information on the ledger the second direction i don't i haven't heard another name for it so i've been calling it the actuation problem

but if there exists an existing term i am i'm always trying to avoid unnecessary terms so please let me know the actuation problem is the problem of getting data about the ledger into the real world mostly this just consists of people choosing uh to treat uh state on the ledger as as uh representing something and every time you accept payment so if i accept payment uh or i accept a state transition to ledger as a reason to give someone else my physical eggs or if if we're still using money if i accept payment and give someone coffee that is actuation in the sense that i'm talking about because the state chained on the letter is affecting the world but you could think of a more direct version of this albeit one that should be treated with great caution where there's some kind of

like robots which fly around which run like clients for the ledger and which enforce the rules or which are you know scanning stuff or something like this or where there are say gates at a border which are maybe they're actually maybe they're taking uh maybe there's some kind of nft uh for a proof of of coin vaccination and these gates deporter wants to verify that you have uh you've received this vaccination uh while you want to protect your own privacy so you could create a zero knowledge proof that you own an nft which proves that you received a vaccination with you know this particular vaccine blah blah blah and present that to the gate at the ledger and it would check that state it would check the state on the blockchain maybe check some other

proof of your identity something and allow you through so that's also a case of actuation but in all of the instances where we want and what we want you know we're not trying to you can build a ledger or a language you can invent a language as some kind of academic exercise in designing something interesting which is cool but in order to have this affect the real world in order to perhaps alleviate the subtractive divergence we need semantic embodiment so we need in all of these cases where we want data about the world to affect the ledger and data on the ledger to affect the world to figure out a way to craft an oracle to get the data onto the ledger and craft some kind of actuation to get the data into the world finally i want to go back to or

penultimately i want to go back to an earlier point where i talked about alignment of control so in this utilization of a distributed ledger to enact this kind of preference expression and settlement um we're making a sort of uh a slight approximation here uh the consensus mechanism say in b of t consensus or nakamoto consensus can fail and even if we try to build a resilient system say if there's some kind of proof of stake token there there are there are technical requirements of the system right and if the enough participants choose to for whatever reason not follow the canonical rules whatever those might

be uh they can cause the system to do something else in my view that means that it's very important that we have a kind of out of band incentive alignment that the participants the validators or the people who are even match making transactions or the people who are in some way enforcing the rule set of the system even even if it's decentralized it's important that they also have a stake in the future of the community itself so that they're also members of this group which is trying to build a school that they're also farmers in this community which wants to trade because only if they have this other out-of-band incentive uh are they going to you know want this want this system to uh you know continue to

uh operate in the best interest of this community and as soon as it's controlled somewhere else we have the same problem that i talked about with the us dollar uh having you know the bank in the us and users in south america where as soon as there's a divergence in the best interest of those groups the ones controlling rules may do something else which causes the coordination mechanism to no longer work as the participants who wanted to use it aim to use it anoma is part of a whole in order to get to a better equilibrium in this design space we don't just need like a distributed ledger we need a bunch of experiments i would argue and we need in particular we need a

carefully crafted transition because we exist as part of this this system uh system of monetary interact i mean all of us we paid money for this conference we probably bought some lunch uh certainly the conference organizers bought us lunch right so there's another layer of interaction we are even in building these systems now we are in part sustained by our existence is sustained by this monetary system of economic coordination whose abstractive divergence we want to alleviate so we need to craft some kind of path that does not immediately require that we say give up using money while also allowing us to actually explore and test whether this

alternative equilibrium exists and that will require not just blockchain protocols especially not just blockchain protocols deployed in contacts where the people who are using them you know uh could also just use money and would continue to live just fine but also local experiments we need and by local i don't mean geographical i also mean say local and commercial space a game could start using this for example instead of money and in many cases those kinds of experiments may be more self-contained or maybe easier to test than isolation because we want to answer again two questions we want to answer the question of whether this better equilibrium exists and what a mechanism for it would look like as we're trying to design and then we need to figure out some way to get there from here but if we can do this then we we have

our cake and eat it too uh if we can craft this kind of alternative mechanism then hopefully we can continue to come to paris and meet people from all around the world who i would have never been able to meet if i lived 100 years ago and continue to play with computers and continue to occasionally launch rockets and make electronic music and do lots of fun things which depend on you know even the current flawed global supply chains but they depend on that depth of complexity and specialization and often in some cases genuine freedom if we want to retain that we need to find a way to alleviate the subtractive divergence and i think only the only trajectory i see

um the neuron is not the unique solution to this it just tries to be solution that's the only trajectory i see which allows us to do that is one in which we design this kind of alternative coordination mechanism so i think this is the question we should be interested in as architects of protocols can we design this what does it look like how do we get there thank you um any questions yes how does fractal scaling work so uh

in fractal scaling uh users can launch anoma instances for their local i should have explained this better good question users can launch anomaly instances for their local communities be those uh geographically local or commercially local as in some kind of game and those local instances are controlled by the communities so they're not uh you know it's it's a it's sovereign uh maybe there's some kind of proof of stake token for your community or just some set of people who are responsible for operating the blockchain but all of the transactions uh which you're interested in settling amongst members of that community then settle on that local instance and you can have an intent gossip instance for your local community when you want to interact across fractal instances using enoma you can use

interchange protocols to send assets and even sometimes to settle trades atomically when you need to just as the nature the the the content or the nature of commerce itself is fractal that there are more interactions within local communities if you live in paris most of your transactions are with people and or who live in paris we think it makes sense to scale the protocols both in the perspective of latency and the sort of alignment of control which i talked about by having local instances which settle most of the transactions between members of community and occasional cross-instance transactions when you need to trade with someone far away so this system sounds very complicated and expensive to

run so it's kinda has uh it sounds like it has adverse uh effect on liquidity for most things and the expression of preferences in in private forms not even not in private but private especially it sounds like uh it's computationally hard and uh and probably slow as well so um that's not the question percep because i like don't know how to formulate it but it sounds like a lot of obstacles uh what what what's uh what what you can do there to make it work that's a good question um i think there are two parts to that question that's and it's important to distinguish between them one part is what is the cost sort of

technically like what's the computational cost of doing all of this stuff the good part about that question is that the computational cost of doing all this stuff goes down as our computers get faster and although human population continues to increase it does not increase anywhere near as rapidly so in some sense the complexity of the coordination problem is growing but it's growing slowly and the speed of the hardware to which we're outside or outsourcing this computation and say the cost of storage or something uh uh that uh well the inverse the cluster that is growing uh still quite fast or even maybe expedite you know low exponentially so there's reason to think that even if zero knowledge proofs are slow now they will get much faster in five years and we don't need to you know we're not going to need to in a global sense settle that many more

transactions the second part of this question is is this expensive in time in order to express their preferences accurately are people going to have to spend a lot of time describing them interacting with the system thinking about it if the answer to that question is yes then it won't work either because you know just like the 600 page binder i mean maybe it's a little different maybe it's working in code but if i need to spend all of my time thinking about these transactions i can't go about my everyday living so here it's very important that the components of preferences that validity predicates are themselves redistributable just like open source code and that it's important that say they are aligned with existing networks of trust so maybe i have some friend who spends a lot of time researching uh environmental impacts

what i really want to do is like delegate mostly delegate my preferences about environmental impacts as i'm engaging in complex economic interactions to this friend right and we want to be able to do that in chains so the um this is why i said enough as part of a whole we'll also need infrastructure for sharing preferences between people and delegating sort of this kind of computational crafting or delegating the creation of programs which capture the kinds of preferences which people have in ways which align with expertise but i think if we can do that the good news is that all of these components are sort of open source fungible because we're using zero knowledge proof so you

can share like circuits or something without sharing the actual data there's reason to believe that we can uh craft some kind of system that will make that relatively inexpensive in times that most of the time i don't think about it i just need to you know check every year or something and readjust my dials and talk to my friends just as i would normally do and when they make some suggestion occasionally i go home and change the settings on an app on my phone to incorporate it thanks that was a really interesting talk um so i mean you're you're talking about this as being an existential risk which and given the example talking about you know monocropping things relate to you know the destruction of the planet

that sort of first is a big problem and secondly puts a timeline on how how much time we have to one build the problem but then two and this is where our curiosity perspective is we need to onboard as many people to participate in this system as possible and and i always sort of think about these sort of things in from the perspective of like what my mom do and my mom's like kind of just got to whatsapp so she's sort of proper boomer type person so adenoma or any other projects you're trying to do to solve the problem of attractive divergence do you have like a timeline and are you sort of also thinking about translating all these you know quite you know very valid but complicated concepts in a way that makes it easy for

mass adoption to happen before we screw the planet up well i mean that's that's the problem we're working on solving now right you should join us i think part of our approach is to try and um release part of this relates also to fractal scaling that we want to release software that people can start to use in their local communities and that will allow frankly the benefits like the the you know maybe the commercial surplus or something generally to accrue to their local communities not to accrue to us even though we're going to launch an instance of innova but part of it i would say is very dependent on

broader societal awareness of which anoma is only a part but i think the broader societal awareness needs to be very focused on a frankly dismantlement of the illusion that states are going to solve these problems because they're not nation states even something like the paris climate accords are operating with the wrong abstractions the paris climate cour if you look at the legal theory behind the paris climate accords uh the lawyers are reasoning about how to make things fair between nation states but as soon as they do this as soon as even if you you know you go through something like okay there have been different historical emissions of pollution relating to levels of development maybe we should compensate for that in our sort of emissions caps we apply to countries

something something as soon as you do that you immediately run into huge uh like incentive for game theoretic problems because if you try to apply emissions caps differently then okay industries are going to migrate between quantities right this is happening because the layer of abstraction is too coarse that if you uh deal with trying to create fairness between countries or enforce these things between countries and rely on some kind of hierarchy to enforce this within the country while people are moving between countries um you know it's never going to solve the problem even if even if the nation-state apparatus moved faster so uh the we need to solve this problem at the like level of the actual supply chain which is the level of our everyday interactions um which means that in some sense it needs to be kind of ubiquitous but also in the background so i think

broader societal awareness benefits according to local communities and also some kind of wake-up call maybe or sense that this this problem isn't i mean we still have some time the good news is that the the you know technology parts of the stack don't take a very long time to deploy so if you look at like apps which and most apps don't change the nature of commerce they're not getting rid of money but apps can be can spread very quickly um still if we want to change the trajectory of say carbon emissions we're operating you know if we don't operate on a timeline of years uh it's not going to do very much so yeah it's urgent other questions okay no worries thank

you you