Autonomous Ecologies #1 was organized by Black Sky (https://www.blacksky.network/) & Anoma  (https://anoma.net/) and took place March 4th, 2023 in Denver, CO. For more information visit https://www.autonomousecologies.xyz

I'm 1 of the co-founders of Enoma and Nemata.

I also used to work on a protocol called IBC before, which you might have heard of.

Today I'm going to talk about scale-free consensus for scale-free money.

I know you're going to say, Christopher, you're introducing another new term.

You know, to me, honestly, it looks like Denver.

And Denver in fact is on USD, the word on, so I chose to use the verb on, like on drugs, like it's very intentional, like you know, Denver is on USD and this is this is what is pretty much like what your world on USD looks like.

And then there's this other, there's this like, yeah, there's this word scale-free money.

Haven't told you yet, a super, super secret.

But I'm going to argue that basically, whether your world looks like this world, looks like this world, is like a consequence of the topology of your accounting system, at least like in large part.

So I'm going to talk a little bit about what that means.

But first, first, first, first, first, we must start with a history lesson.

So there's this whole discipline called economics.

It kind of turns out that it's mostly wrong.

There's some crazy assumptions once you start to read it.

Risk-free interest rate is a standard assumption.

Let's just take as a standard assumption that you can get money for free.

We don't need to, you know, we don't need to invent any fancy terms.

This just came into my head when I was making the presentation.

I think it's actually like a rhyme from a children's song.

I don't know if, you know, like is the ratio of butchers to bakers to candlestick makers usually 1?

Probably not, but they are all like pretty helpful things.

So there's this theory that's part of the economic canon that money emerges from trying to solve the double coincidence of wants.

The theory basically says that communities start out bartering.

So someone, you know, the butcher chops the meat, the baker bakes the bread, the candlestick maker makes the candlesticks, and like they just barter with each other to get the stuff which they want, and that's kind of annoying because the butcher has to find a time when he wants to cut meat and the baker also wants to give him bread and he wants bread and the baker wants meat and that seems much less likely to randomly emerge than if you had some great kind of unit of money just to intermediate all of this so you don't need to solve double coincidence at once.

And so the economic canon postulates, I think Adam Smith particularly is the 1 who originated this, that this barter economy leads to money.

It turns out that this like, when you go read history as opposed to just like your economics textbooks.

A bunch of anthropologists spent a lot of time, like 10 years and PhD thesis and all these other things, looking for a barter economy that turned into money and it just didn't happen.

Because people are smarter than economists and they did credit accounting.

So credit accounting is pretty straightforward, in fact.

In a small town everyone knows each other, people do need the stuff that everyone else is producing otherwise they probably wouldn't be producing it.

They just don't need it all at the same time.

So credit basically integrates over time.

So in a simple credit economy in a town, you know, the butcher just goes to the baker and he actually just takes the bread, right?

He's not bringing meat, at least not most of the time.

The baker keeps a tally sheet and says, okay, well, you came on Tuesday, and you came on Wednesday, and you came on Thursday, and where's my meat?

of this accounting scheme is that it's actually kind

of like purposefully somewhat approximate, like it doesn't need to be exact.

You just need to identify, oh, there's someone in this town who's not producing anything that anybody needs and they keep taking too much, right?

So you still want to identify gross imbalances, but you don't need to exactly track everything.

The other advantage of discredited accounting is that it keeps everyone's incentives pretty aligned.

Because if I'm the butcher and you're the baker, I want you to be able to continue to produce bread.

You want me to be able to continue to slaughter meat.

It's in our best interest, just because we're in this very tight-knit community To be understanding of like you know things not always exactly matching up to equal right because there's a lot of fluctuation Maybe some years the harvest is better some years You know your wheat fields are ruined by locusts or whatever.

You can tell that I've never grown wheat.

And in particular, 1 thing that's great about it is that this kind of pre-modern money is actually something called scale free.

And by scale free, I just mean that anyone can issue money.

So here the definition of money being used is just like whatever it is that we are using to account, right?

So I'm not saying like this is not dollars, you know, they're not bills going around, right?

But When I say money, I'm talking about the accounting system.

And in this pre-modern monetary system, anyone can issue credit.

Basically, I issue credit by going to the baker and taking some bread, and the baker writes something down.

And that something is the identity who's issuing credit.

That could be individuals, could be businesses.

The town probably has a claim on resources if they're doing whole-town improvement projects, whatever, building stuff.

This credit accounting is not in a database system.

There's no spreadsheet, there's no central bank, there's no...

Probably not in at least the instance which I read, there wasn't really a central accounting facility.

You know, people would just compare notes sometimes, but basically, like, people would individually keep their own credit ledgers.

The baker would keep his credit ledger, the butcher would keep his credit ledger, etc.

I think you can consider this a monetary system, but it's virtual.

And in particular, the reason why, you know, maybe the primary reason why we don't have this nice system anymore is that it does not scale.

This only works when everyone knows each other and everyone can kind of keep approximate track of who's taking what, who's producing what, you know, is this system sort of balanced?

And it also requires that everyone basically needs what each other is producing, even if not at the same time.

So as a butcher, as long as I need bread sometimes, then it's fine to me if you, the baker, come and take some of my meat.

Sometimes we don't need to match up in time, but we need to have the expectation that we'll match up eventually, right?

And that works when you have a very dense network of just economic, real economic or physical economic interdependency between the participants, but not so much in

the modern world where you don't at large scales.

So in the modern world we live with fiat money.

Fiat money is like also credit, it's just limited to a few dozen governments who can issue it, and the rest of us plebs trade obligations with regards to those governments.

This does, like, solve a certain sense of

Strangers don't need to trust each other, and they don't need to have mutual interdependency in order

They just need to trust the same fiat, and they need to trust the accuracy of the counting mechanism.

a lot of preaching to the choir here, right?

I'm going to focus on 2 problems which I think are perhaps somewhat un-Fiatized.

It's not just that governments do the wrong thing or something.

In my understanding, the 2 principal problems with fiat are that Centralization of trusts results in a loss of fault tolerance, and I actually think here the second and third and fourth You know etc order effects are worse than the first order effects that the really bad thing about the centralization centralization of fiat issuance is not just that like I disagree with what the government spends money on It's that the competition over that money supply generates large public externalities Which are usually negative because people are like you know need to come up with good propaganda To get them get the public to elect their side or to like even before elections to not, you know, not riot whatever.

And that especially when amplified by digital technology and sort of the modern complexity has large negative, usually negative consequences for public sense-making.

But the other fatal flaw of fiat that I want to talk about is a pretty 1, and I think it's 1 that you might not think about so immediately because it's not like, it's not in the political discussion sphere.

And that's that fiat relies on measurement in the present.

So there's a big difference between transacting with kind of personal credit money and transacting with fiat, in that when you transact with fiat, you're only trading obligations to the government.

You're actually not like, this is a kind of decontextualized exchange.

So even when, you know, sometimes maybe you're trading with, you're paying someone who you don't know, but sometimes you're paying someone who you know, right?

But even when you're paying something, someone who you know, when you use fiat, you calculate exactly what it is that you owe the other person or what it is that they owe you So that you can walk away from every interaction and basically be happy with that, right?

Whereas when you're dealing in more personal credit denominations of individuals or organizations, the parties walk away having some kind of mutual interest in each other's success, right?

If I hold government debt and I am like responsible for mentoring you, I, you know, I have no real reason to care if I succeed at doing that because I still have the same government debt, right?

But if I hold your credit and I'm mentoring you, then I want to do a good job.

Because if I do a good job, your credit, which I now hold, is going to be valuable to me in the future because other people will want what it is that you're now able to do, right?

So this reliance on measurement in the present, you know, I think works fine for like physical stuff that basically mostly exists in the present, like sandwiches,

and it just doesn't work so well for things of the value which is mostly in

So, what is this concept, scale-free money, and what would it look like?

A world of scale-free money is 1 where we kind of like try to take this thing that's happening in the village where everyone is issuing credit and figure out some way to like solve the scaling problem, right?

So everyone can issue money wherever they want.

in a monetary system design for only institutions at very large scales being able to issue money, right?

You know, independent of what I think like the structural analysis is more important than the particular analysis.

So independent of like what the governments are, governments are large.

They're large and typically they have you know monopoly on physical force, but in a

fiat system only those large institutions can issue money.

a scale-free monetary system everyone can issue money.

Individuals can issue money, Individuals can issue like multiple denominations of money.

There's different friend groups can issue money Organizations can issue money governments can issue money now.

This is not like UBI these are different denominations of money So then the challenge is how do you like well in a world where we're all using different denominations of money, we don't actually agree on everything, anything, we just have like 5 bajillion different denominations of money, so how do we match that up?

And that's basically what I'll spend the remainder of the presentation on.

But the base assumption, if you try and design a system this way, is that everyone can issue money.

You just use cryptographic signatures or something, and voila, money.

So in particular, also, I want to note that I'm using the word money, and money is a very loaded word.

It's possibly the most loaded word in the English language at this point.

But money in this system, it starts to become something else.

So I would argue that money kind of already is a distributed voting scheme.

But we're getting the results of like running this terrible voting

know scale free money tries to like change how

In particular it changes it in a way that that sort of walks a tightrope between scarcity and abundance.

So we do want probably an accounting system that has the ability to track things which are physically scarce, because we have some physically scarce resources that are not, not everything is

a public good, and we need some amount of scarcity to

On the other hand, we have a lot of things that are not scarce resources, and if we optimize only for tracking scarce resources, we're probably only going to succeed at coordinating those, which is roughly what we see.

from the perspective of an individual denomination, there's still like a finite amount of my credit, right, whatever it is that I've issued, but it is not scarce in terms of number of denominations.

So you can always like create a new denomination of money or combine with a different friend and choose to create your temporary DAO and that can start to issue money.

So there is kind of a distinction there in the scarcity of the scheme.

There are a lot of cool things you can do.

Again, I've bracketed the hard problem of, OK, now we have 5 million denominations.

But first, I just want to note 2 cool things you can do with skill-free money.

As soon as you start, basically, it would just fall out for free when you start thinking about the problem this way.

1 cool thing you can do is mutual credit issuance or sort of you know to map it on to existing conceptual lingo heterogeneous UBI So that just means that we have 2 2 denominations of credit right issued by different parties could be different people Friends could be different organizations and those parties agree to issue each other, you know, some probably like the same for people or in a proportional amount of credit per unit times.

This could be people, friend groups, companies, nations, but this allows you to build up these kind of, basically it allows you to build up interdependency that you don't have to measure.

You can view that from like the macroscopic economic perspective, you can view it from the public goods perspective, you can view it from the quantum physics perspective, measurement is not a free thing.

So we want to design economic systems which allow people to choose that they aren't going to measure stuff, but express some kind of interdependency, right?

So like, you know, families are communists is another way of putting this, because it's would be terrible and dystopian and also just really time consuming to exactly measure all of the resources that you're sharing in a close-knit social group, right?

So we want to build systems which have the ability for people to do this kind of mutual issuance,

And if you have a bunch of mutual credit issuance going on in some network graph, it starts to look something like heterogeneous UBI, but where there's no central decision maker, where all of the decisions are pairwise.

on top of pairwise decisions over time, you can build up this very resilient network of mutual credit relationships.

The other thing you can do, or that's sort

of as an interesting corollary of thinking about a scale-free money, is that with scale-free monetary systems, most of the issuance of money, both of existing denominations and particularly of new denominations, is in the future.

And for denominations at large scales, like countries are probably still gonna issue money, For denominations at large scales, those denominations in a world that kind of is actually compatible with scale-free money are very competitive with each other,

as in they're like competing to be used as money at a large scale.

And if you have denominations which are competing to be used as money, the next question you might ask is, well, what determines whether 1 denomination gets chosen over another?

And I would argue that if people are sort of choosing what money to use and that choice is pretty free, then people have

the ability to choose on the basis of how fair they think the distribution is.

you know, you think a distribution is, in some ways could be seen as like well who is being distributed that money what was kind of the initial supply and if you're starting a currency for some sort of large-scale institution like a state you need to like You probably want to distribute it pretty widely, right?

So you might assign some of the initial supply as basically retroactive public goods funding for stuff that's been going on relevant to your community in the near past.

And if people choose on the basis of how fair they think these distributions are, then hopefully it creates some kind of competitive pressure for new distributions of money in order to be seen as fair to try and accurately, at least somewhat accurately, compensate projects which have been helpful to their community in the past.

Right, okay, so let's get into the hard problem.

Oh, well, conveniently I work on a project called Anoma.

Right, so what is Anoma, and why is it relevant to this very theoretical question of scale-free money.

Anoma, just to greatly shorten things, is a protocol stack for what we call intent-centric private coordination.

And I will give you a diagram to explain that instead of trying just with concepts.

But first, let me note, for the sake of terminological clarity, Enoma consists of 2 main components, Typhon, which handles peer-to-peer networking, Mempool consensus execution engine, and Tyga, which is kind of like a, Tyga's like the EVM closest thing, but it's mostly, it's more like an operating system for privacy preserving distributed applications when you combine Typhon plus Tiger equals Anoma Right.

So first, let me talk about this word intent centric So Anoma is designed as an architecture to solve 2 problems The first is counterparty discovery and the second is settlement.

Settlement is what blockchains do, as in you have a

bunch of transactions and you want to order them and you want to like execute them, you know, maybe store some data, although I would argue that parts, you know, could be kind of auxiliary.

And then you care about reading the execution result afterwards.

And in particular, settlement happens when you have parties who already basically, in order to do settlement, you must know what it is that you want to settle.

So if there are multiple parties involved in a transaction, those parties need to know who they are, they need to have crafted that transaction, they need to have agreed on its contents, and they can settle it.

And the hard part in these systems is counterparty discovery.

So counterparty Discovery is the part, basically our word for everything which has to happen in order for these parties to find each other.

So usually the advantage of using a kind of like programmatic distributed system for multi-party interactions is that you can do some amount of automated counterparty discovery.

So we could have like a world where all counterparty discovery is done by like texting people on signal That's not a bad method, but it doesn't scale right so we want to be able to find often counterparties on on the basis Not of who they

are But what it is that they can do right counterparties who have some particular asset that we want Counterparties who are you know in

in sort of overlapping trust circles stuff like this, right?

So Enoma tries to solve programmatically counterparty discovery and settlement and it tries to do so privately So Let me walk through just a little bit of the data flow of how this works.

So we start with a user, and we say the user has a preference, and I think this is a very safe assumption to make, because if a user didn't have a preference, they would have no reason to interact with the system, right?

So user creates what we call in our architecture something called an intent.

And an intent is like a generalized form of an off-chain signed message that says, that describes what it is that the user wants.

It could say like they want to swap A for B or they have, you know, some more complex preferences or looking for a potato which is in fact blue russet in which they can pick up within 50 miles of Oregon.

They send that intent around to what we call the intent gossip network.

And the intent gossip network is doing, there's some kind of, should be a circular arrow here.

The intent gossip network is trying to match.

So it's sending around these intents, all in kind of this sparse gossip, so not everyone gets all of them, but everyone gets some of them.

And Solvers are looking for 2 intents which can be matched.

Matching is possible both in partial and full.

So in full matching you have like, you know, Alice wants to trade X for Y, Bob wants to trade Y for X.

Now you have a balanced transaction which can be settled.

But even if Alice wants to trade X for Y and Bob wants to trade like Y for Z, that's not yet a final transaction, but it is simplifiable, right?

And Noma's designed to support this kind of partial settlement, which we think is essential to scaling these sort of complex off-chain counter-party discovery interactions.

Right, so then let's assume that solving has worked, you have some kind of transaction that gets sent to the, you know, more typical mempool consensus executor, and then you have something which you can read back afterwards.

So How is this relevant for scale free money What do we need to do in order to implement scale free money?

To again simplify things somewhat we need to support issuance exchange We need to support cycle clearing cycle clearing is how you deal with counterparties who don't directly have trust relations.

So Alice, Bob, Charlie, Alice and Bob know each other, Bob and Charlie know each other, Alice wants to pay Charlie, but Alice and Charlie don't accept each other's credit or whatever.

In order to scale, you need to be able to use these larger paths in the graph to settle, such that in this case, like Martin, I guess, gave the great presentation on circles earlier, basically the same thing is happening.

So, Tyga is the part of Anomis architecture that handles issuance exchange, cycle clearing, and mutual credit.

Tiger provides full privacy for everything but solvers, as in not your identity, but the fact that you want to trade A for B or B for C is visible to the solver who's matching the intents.

But you can kind of forward liquidity along the trust graphs.

So maybe it doesn't matter so much, but if you want solver privacy, you can also add TEE or simple MPC just for that part and then settle with CKPs later, which probably always makes sense.

Gossip, settlement, state synchronization can be handled with Typhon.

And in particular, Typhon provides something that we've started to call scale-free consensus.

So Enoma is 1 way Enoma is kind of different than other blockchain systems is that it is designed to leave the trust graph as a free parameter.

So architecturally, often systems start by making some trust assumptions.

Like they say, okay, we're going to have a validator set, and it'll be big, but you have to trust

And there's like 1 validator set basically, right?

Or they say that, okay, we're gonna have like these parties and you have to, you know, we'll start there.

Or even in the peer-to-peer layer, they say that, oh, well, we're going to assume that like we have such and such trust model and like either it works or it doesn't, it's kind of atomic.

And users who are using these systems have to atomically take or leave this like 1 trust model.

And you know, in practice many, you know, most blockchains Have like other more practical problems that are more of a security risk and their trust models failing like their software has bugs But on the other hand this does you know, we think this does come with consequences not only the users have to exactly agree, but also that like basically protocols and trust models are tightly coupled, and users, when they switch between protocols, also switch between trust models and tend to associate them with each other, which is not necessary from a technical perspective.

So Enoma runs through the whole protocol stack, which is kind of annoying, it takes a long time to actually build, but runs this concept of heterogeneous trust, where the trust graph, as in what assumptions users want to make, is a free parameter and is even assumed to be dynamically changing.

And where users are making sufficiently compatible trust assumptions, particularly for settlement, they can interact, and where they're not, you know, they can't.

So Enumware provides this using something we call scale-free consensus.

And scale-free consensus doesn't mean like sort of magically scaling.

What it does mean is that you could also call it scale invariant, maybe slightly more technically accurate.

But scale-free consensus is based on a compositional identity abstraction.

So a Compositional identity abstraction means that the protocol deals with identity in a way which is Compositional and independent like the protocol deals with identity from the perspective only of an observer So there's some internal logic for each identity as to how signatures are produced, how messages are decrypted, typically that's based on knowledge of private information such as a private key, but to the rest of the protocol stack we only care about verifying that, we only care about knowing how to encrypt things to an identity and We only care about knowing how to encrypt things to an identity and we only care about knowing how to check signatures, basically check that an identity sent a message.

And this abstraction in Anoma is compositional.

So if I know, like, 2 people, Alice and Bob, I can create the identity Alice and Bob, from which messages can only be sent by both of them agreeing, or I create Alice or Bob, from which either 1 can send a message.

See all Black Sky transcripts on Youtube

Christopher Goes - Anoma: Scale-free Consensus for Scale-free Money [Autonomous Ecologies #1]