Winds of Change: The Case for New Digital Currency(imf.org)
imf.org
Winds of Change: The Case for New Digital Currency
https://www.imf.org/en/News/Articles/2018/11/13/sp111418-winds-of-change-the-case-for-new-digital-currency
35 comments
You can't remove the need for trust - that need is inherently beyond the reach of math.
If I want to use money to pay for some goods, I need trust that I can get that money back if the goods won't arrive. Math alone can't provide that, it needs something extra, e.g. a trusted escrow service in the case of crypto.
If I want to use money as a store of value, I need trust that the value is going to stay stable and maintain liquidity in the long term. Math alone can't provide that, that trust is inherently social. I also need reasonable trust that the money isn't likely to get stolen - math isn't sufficient for that, and our current experience shows that crypto assets are more vulnerable in practice than e.g. bank deposits, maintaining appropriate "opsec" is tricky and people fail in that too frequently.
And for your other point, the economy existed before a monetary system and currently works off of many monetary systems - however, for a monetary system the only criteria that matters is if it's good for what the economy needs; if some system does not fit these needs, then it won't get used much and is not particularly relevant.
If I want to use money to pay for some goods, I need trust that I can get that money back if the goods won't arrive. Math alone can't provide that, it needs something extra, e.g. a trusted escrow service in the case of crypto.
If I want to use money as a store of value, I need trust that the value is going to stay stable and maintain liquidity in the long term. Math alone can't provide that, that trust is inherently social. I also need reasonable trust that the money isn't likely to get stolen - math isn't sufficient for that, and our current experience shows that crypto assets are more vulnerable in practice than e.g. bank deposits, maintaining appropriate "opsec" is tricky and people fail in that too frequently.
And for your other point, the economy existed before a monetary system and currently works off of many monetary systems - however, for a monetary system the only criteria that matters is if it's good for what the economy needs; if some system does not fit these needs, then it won't get used much and is not particularly relevant.
This is the elephant in the room, and the key component of cryptocurrencies which is yet to be solved. Cryptocurrencies can trustlessly manage intangible tokens which were themselves issued on the blockchain. But when you want to interact with real world, tangible assets, blockchains can only manage proxies of the real asset. There has not yet been a trustless way to ensure the link between the tangible asset and the blockchain proxy.
An alternative solution to requiring a 3rd party escrow is to use the concept of "mutually assured destruction". The general idea is that each party deposits twice the value into a shared 2-of-2 multisig wallet. Then, two transactions are generated: 1 for completing the purchase, and one for completing a refund. Both parties need to sign one of the transactions, or all their funds remain locked up, so there is an incentive to cooperate and come to an agreement without needing a 3rd party involved. It isn't perfect, because there are other drawbacks such as needing a large amount of money up front, but it is a trustless alternative.
This page explains it better: http://nashx.com/HowItWorks
This page explains it better: http://nashx.com/HowItWorks
>If I want to use money to pay for some goods, I need trust that I can get that money back if the goods won't arrive. Math alone can't provide that, it needs something extra, e.g. a trusted escrow service in the case of crypto.
No need for extras, you can do this today on Ethereum with Hashed Time Lock Contracts
No need for extras, you can do this today on Ethereum with Hashed Time Lock Contracts
Could you please expand on how does the Ethereum blockchain get to know whether the goods arrived if the buyer and seller are saying conflicting things and no third parties are involved?
> Could you please expand on how does the Ethereum blockchain get to know whether the goods arrived if the buyer and seller are saying conflicting things and no third parties are involved?
It's possible to have some mutually assured destruction here, e.g. the contract is for $100 so the buyer puts in $150 and the seller puts in $100. Then when the buyer releases the money, they get $50 of their money back and the seller gets $200 ($100 from the buyer and their original $100). If the buyer doesn't release the money, nobody gets anything.
It gives the buyer a way to punish the seller for non-delivery, but without receiving any personal benefit (and in fact at a personal cost) for using it.
It's possible to have some mutually assured destruction here, e.g. the contract is for $100 so the buyer puts in $150 and the seller puts in $100. Then when the buyer releases the money, they get $50 of their money back and the seller gets $200 ($100 from the buyer and their original $100). If the buyer doesn't release the money, nobody gets anything.
It gives the buyer a way to punish the seller for non-delivery, but without receiving any personal benefit (and in fact at a personal cost) for using it.
If the only recourse for a defrauded buyer is either to simply eat the loss or lose even more money, that's not what I'd consider an acceptable solution that's competitive with other, non-crypto means of payment. It's a solution that does provide some disincentive against fraud, but it's significantly worse for most consumers than the current status quo.
> If the only recourse for a defrauded buyer is either to simply eat the loss or lose even more money, that's not what I'd consider an acceptable solution that's competitive with other, non-crypto means of payment.
In the existing payment market you have a seller with no stake and have to correct for it by drafting the payment processor to post a stake instead. Then if the seller doesn't send the product (or sometimes even if they do), the buyer makes a claim with the payment processor and the payment processor is stuck refunding it because the alternative is having a larger cost imposed on them by the government. Then they try to mitigate the loss by not paying the seller, if they can.
This is just unnecessary indirection when the seller posts a stake instead. If the seller has $200 on the line over a $100 value item, they're going to deliver it because the alternative is a $100 net loss. It's the same reason the payment processor refunds your money -- because the alternative is worse for them.
The buyer doesn't actually lose the extra $50, they just sit on it until the seller makes good, which happens eventually because the seller's alternative is worse. The end result of MAD isn't that everybody gets nuked, it's that nobody does.
In the existing payment market you have a seller with no stake and have to correct for it by drafting the payment processor to post a stake instead. Then if the seller doesn't send the product (or sometimes even if they do), the buyer makes a claim with the payment processor and the payment processor is stuck refunding it because the alternative is having a larger cost imposed on them by the government. Then they try to mitigate the loss by not paying the seller, if they can.
This is just unnecessary indirection when the seller posts a stake instead. If the seller has $200 on the line over a $100 value item, they're going to deliver it because the alternative is a $100 net loss. It's the same reason the payment processor refunds your money -- because the alternative is worse for them.
The buyer doesn't actually lose the extra $50, they just sit on it until the seller makes good, which happens eventually because the seller's alternative is worse. The end result of MAD isn't that everybody gets nuked, it's that nobody does.
there's an endless supply of replies to comments like yours that go like "it will never be possible to fix every problem so the safest move is no move", and I don't wish to be that kind of commenter, i.e. I think it is good to at least try and propose mechanisms -even flawed ones- so that at least the discussion of the problems continue and hopefully the mechanisms can be improved.
it is in this mindset that I am asking if we can adress the following issue: suppose the government or its postal system selectively withholds the goods payed for through cryptographic protocols, then buyer and seller lose.
again, I'm not trying to be destructive, just wondering out loud if we can generalize the kind of protocol you describe to include the transporter as an attacker in the attack model.
I think we could split up in 2 cases: transporters that are paid on the same cryptographic protocol/platform, and transporters that are paid outside. I believe it will be easier to solve the problem by restricting to transporters who get paid by the same contract/platform, since we could have the transporter deposit an insurance of same or higher value as the good to be transported, and upon arrival his pre-agreed transport cost and deposit are released?
Obviously the traditional postal system can not be forced to be paid through this platform, so this hole in the market would effectively create demand for crypto postage and hence create jobs.
I would like to see a more rigorous breakdown of such a protocol, and a censor-ship free listing of issues remaining with the protocol, so that the discussion on improvement can continue.
Also it seems like the first transporter does not need to be the same person as the last transporter, i.e. intermediary packet handoff could also be handled by the protocol such that the first transporter gets paid back his deposit after doing "his part" of the packet journey...
it is in this mindset that I am asking if we can adress the following issue: suppose the government or its postal system selectively withholds the goods payed for through cryptographic protocols, then buyer and seller lose.
again, I'm not trying to be destructive, just wondering out loud if we can generalize the kind of protocol you describe to include the transporter as an attacker in the attack model.
I think we could split up in 2 cases: transporters that are paid on the same cryptographic protocol/platform, and transporters that are paid outside. I believe it will be easier to solve the problem by restricting to transporters who get paid by the same contract/platform, since we could have the transporter deposit an insurance of same or higher value as the good to be transported, and upon arrival his pre-agreed transport cost and deposit are released?
Obviously the traditional postal system can not be forced to be paid through this platform, so this hole in the market would effectively create demand for crypto postage and hence create jobs.
I would like to see a more rigorous breakdown of such a protocol, and a censor-ship free listing of issues remaining with the protocol, so that the discussion on improvement can continue.
Also it seems like the first transporter does not need to be the same person as the last transporter, i.e. intermediary packet handoff could also be handled by the protocol such that the first transporter gets paid back his deposit after doing "his part" of the packet journey...
There are two parts to that. One is insurance -- 8% of packages don't make it to the destination and it would be preferable to pay an 8% premium than to lose 100% of the delivery 8% of the time. There is obviously an insurance market for this.
The second part is that it's often more efficient for the insurance provider to be the delivery provider, because then they have the right incentives to actually deliver your package. Then it's their problem whether they farm out the delivery to the local postal service and take the risk of having to pay claims, or operate some kind of underground railroad themselves in authoritarian countries and pay a little higher cost to reduce the losses they have to eat.
The second part is that it's often more efficient for the insurance provider to be the delivery provider, because then they have the right incentives to actually deliver your package. Then it's their problem whether they farm out the delivery to the local postal service and take the risk of having to pay claims, or operate some kind of underground railroad themselves in authoritarian countries and pay a little higher cost to reduce the losses they have to eat.
It seems that you're describing a solution to the problem of delivery failure. However, the larger problem where consumers need the ability to revoke deals is not failure of delivery as such, but failure to deliver the right goods, possibly maliciously. And vice versa, the possibility by consumers to exploit the system to maliciously refuse paying for goods.
Censorship resistance is an edge case that's nice to have for some people, fraud resistance is the mainstream need that's mandatory for most users. Current crypto approaches don't handle fraud resistance as well as the traditional payment systems, so censorship resistance is a moot point until/unless that gets solved.
Censorship resistance is an edge case that's nice to have for some people, fraud resistance is the mainstream need that's mandatory for most users. Current crypto approaches don't handle fraud resistance as well as the traditional payment systems, so censorship resistance is a moot point until/unless that gets solved.
I am making no claims about priority of problems, so I certainly welcome identification of other issues and proposals to improve them.
If we mentally subdivide all commercial activity between traditional/centralized/blind trust commercial activities and novel/decentralized/cryptographic trust commercial activities, then (even if it is not occuring yet) it can desirable for the supporters of decentralization to be able to assure themselves that the status quo can not undermine the economic value of the cryptographic trust community.
Even outside of cryptocurrencies we have debates about net neutrality, priority of packets etc... One could similarily wonder if current delivery services are able to say profit by investing in specific companies within a sector of products, and differentially prioritizing the delivery of their goods (or gentleness of delivery for fragile goods, or even non-delivery).
I don't believe in moot points, people can work on different issues, design solutions for them and then they can look at how to intersect/generalize their protocols so that it displays both or more desirable traits...
If we mentally subdivide all commercial activity between traditional/centralized/blind trust commercial activities and novel/decentralized/cryptographic trust commercial activities, then (even if it is not occuring yet) it can desirable for the supporters of decentralization to be able to assure themselves that the status quo can not undermine the economic value of the cryptographic trust community.
Even outside of cryptocurrencies we have debates about net neutrality, priority of packets etc... One could similarily wonder if current delivery services are able to say profit by investing in specific companies within a sector of products, and differentially prioritizing the delivery of their goods (or gentleness of delivery for fragile goods, or even non-delivery).
I don't believe in moot points, people can work on different issues, design solutions for them and then they can look at how to intersect/generalize their protocols so that it displays both or more desirable traits...
Okay, in that regard the assumption that "the government or its postal system selectively withholds the goods payed for through cryptographic protocols" seems questionable. Is this a real problem that we're seeing in some markets? As far as I'm aware it seems that in general delivery services are decoupled from the payment for goods; the delivery system, no matter if it's government postal system or private parcel delivery, does not know and can not know how (and if) the goods were paid for. They may require some declaration from the sender about the value for insurance and customs purposes, but that's different than the payment data.
Of course he can't.
Bitcoin has HTLCs too...
Well there are smart contracts which can act as an escrow service. I don't know how that works in reality, though. If smart contracts have to be deterministic how can they be used to guarantee that, say, a delivery was made? I want to belieber.
The way how that's usually done within smart contracts relies on a trusted (the key word here) "oracle" that is considered to hold the truth about reality.
E.g. you may have a smart contract that's conditional on the price of some stocks in the future or on the result of some sports game. It'll get the information about the price of these stocks from some place which is assumed to tell the truth, that's the oracle - for example, it could be some NASDAQ service or the sports league website. However, it's obviously not guaranteed to tell the truth about the reality, you need to trust it, because if it lies, there's no recourse afterwards.
If I make a "real" bet on the superbowl results and bribe the superbowl website admin to publish a wrong result on the official site for 5 minutes at the right time while my bet is settled, then the bet will be revoked afterwards by the courts; however, if it's a smart contract looking at that site, then once I've cheated the escrow service, then I've won because it was falsely trusted.
E.g. you may have a smart contract that's conditional on the price of some stocks in the future or on the result of some sports game. It'll get the information about the price of these stocks from some place which is assumed to tell the truth, that's the oracle - for example, it could be some NASDAQ service or the sports league website. However, it's obviously not guaranteed to tell the truth about the reality, you need to trust it, because if it lies, there's no recourse afterwards.
If I make a "real" bet on the superbowl results and bribe the superbowl website admin to publish a wrong result on the official site for 5 minutes at the right time while my bet is settled, then the bet will be revoked afterwards by the courts; however, if it's a smart contract looking at that site, then once I've cheated the escrow service, then I've won because it was falsely trusted.
What if the oracle stops working or gives inconsistent results?
Do you think having to rely on oracles is a major flaw in that they will become trust authorities?
Do you think having to rely on oracles is a major flaw in that they will become trust authorities?
This is where consensus comes in. Projects like Blocknet have the ability where you can request an answer from N amount of nodes and the response will be that of the majority.
Yes, just as internet messaging and videoconferencing have made travel for F2F meetings more important, "trustless" cryptocurrency increases the value of "mathless" trust mechanisms.
> You can't remove the need for trust - that need is inherently beyond the reach of math.
That hides more than it reveals. Yes, trust is beyond math.
Measuring value is also beyond math. However there are ways of measuring value that aren't amenable to math (eg, how much you like a diamond) and ways of measuring value that are amenable to math (how much you are willing to pay for the diamond). The first worked method fine in a subsistence economy. The second way is how our current economy works. We are not completely and utterly dependent on being able to measure and transfer value with math. If we could not do it capitalism would not exist, nor would our society or way of life.
And so it is with trust. You can trust a banker not to embezzle your money because he has a reputation, mouths to feed, or he's your brother, or he will be perused by government agents armed with big guns, or any number of the other reasons we do now - none of which are easily quantified. Or you can trust him because he can't undo his decisions without corrupting a while pile of others and do to that he had to expend the net energy output of New Zealand for days. One is amenable to being manipulated with math, and one isn't.
Because the trust provided by crypto currencies can be quantified and manipulated with math we have smart contracts. Smart contracts mean the example you gave about trusting someone to deliver the goods means a computer can be the escrow agent and it is all done automatically the 99 times in 100 there is no dispute.
Yes, the OP was wrong in that crypto currencies allow us to replace trust with math. But only a little wrong. Crypto currencies allow us to manipulate and transfer trust with math, which is not so different.
That hides more than it reveals. Yes, trust is beyond math.
Measuring value is also beyond math. However there are ways of measuring value that aren't amenable to math (eg, how much you like a diamond) and ways of measuring value that are amenable to math (how much you are willing to pay for the diamond). The first worked method fine in a subsistence economy. The second way is how our current economy works. We are not completely and utterly dependent on being able to measure and transfer value with math. If we could not do it capitalism would not exist, nor would our society or way of life.
And so it is with trust. You can trust a banker not to embezzle your money because he has a reputation, mouths to feed, or he's your brother, or he will be perused by government agents armed with big guns, or any number of the other reasons we do now - none of which are easily quantified. Or you can trust him because he can't undo his decisions without corrupting a while pile of others and do to that he had to expend the net energy output of New Zealand for days. One is amenable to being manipulated with math, and one isn't.
Because the trust provided by crypto currencies can be quantified and manipulated with math we have smart contracts. Smart contracts mean the example you gave about trusting someone to deliver the goods means a computer can be the escrow agent and it is all done automatically the 99 times in 100 there is no dispute.
Yes, the OP was wrong in that crypto currencies allow us to replace trust with math. But only a little wrong. Crypto currencies allow us to manipulate and transfer trust with math, which is not so different.
There is still the need for trust... in the implementation of software!
Even if you read the source code carefully and ocmpile the software yourself, you trust the compiler and the operating system to do the right thing.
Even if you read the source code carefully and ocmpile the software yourself, you trust the compiler and the operating system to do the right thing.
No, you dont need to trust the software per say; there is a need to understand how software works in order to participate. That knowledge gap for most people will result in them trusting a 3rd party auditing of the software. You do much the same when you allowed your bank to share information with equifax... what happened there? One of many issues with trusting corps to do the right thing [EDIT: spelling]
> the purpose of cryptocurrencies is to remove the need for trust. Trust in the math, not the banker
By exchanging trust for energy.
Cryptocurrencies are the best example of trust being a tangible thing - it can be quite literally converted to Kilowatt hours. This shows how incredible trust is as an energy use optimization device. We should be finding ways to be able to safely depend more on trust, not less.
By exchanging trust for energy.
Cryptocurrencies are the best example of trust being a tangible thing - it can be quite literally converted to Kilowatt hours. This shows how incredible trust is as an energy use optimization device. We should be finding ways to be able to safely depend more on trust, not less.
Because of bitcoin, cryptocurrency has a strong association with "free money," which is basically a dissent theory of macroeconomics that doesn't want central banks to have control. This was written by a central banker, so obviously there's a belligerence here...
In any case... Idk what I think about free money. Macroeconomics is hard.
OTOH, I think cryptocurrencies have other disruptive potential even if we want to keep our current macroeconomics, central bank control over interest rates.
It's the kind of technology that turns a £100bn industry into a £1bn one. There are several clunky inefficient financial services sub-industries that could potentiall be digitised by good implementation of blockchain tech.
The problem is too-big-to-fail. Besides actual central banking, does Christine Lagarde see her job as protector of the financial services industry? If so, she's going to build bureaucracy that maintains the status quo.
In any case... Idk what I think about free money. Macroeconomics is hard.
OTOH, I think cryptocurrencies have other disruptive potential even if we want to keep our current macroeconomics, central bank control over interest rates.
It's the kind of technology that turns a £100bn industry into a £1bn one. There are several clunky inefficient financial services sub-industries that could potentiall be digitised by good implementation of blockchain tech.
The problem is too-big-to-fail. Besides actual central banking, does Christine Lagarde see her job as protector of the financial services industry? If so, she's going to build bureaucracy that maintains the status quo.
I feel like one day, central banks are going to be attacked by private banks and the means of attack will be cryptocurrency. The aim will be to get rid of central bank deposits and seize control of the payments system - once the central bank deposits are gone and we're cashless, this will become easy. They could even base it offshore to avoid consumer protections. If we're lucky when this happens, it will be a test of the remaining power of nation states. If we're unlucky, they'll roll over as the most basic means of exchange is privatized.
This is suprisingly good and even refreshing speech after reading tons of very opinionated pro- and contra-crypto articles.
Financial system design still have this chicken or the egg problem – we need to design fintech systems aligned with offline social interaction patterns as much as possible, but their (current) design changes our social interaction patterns in return as well.
It's still fascinating to see how the world reimagines and reconsiders the nature of money. And not only in cryptopunks community, but in such a huge powerhouses as IMF as well.
Financial system design still have this chicken or the egg problem – we need to design fintech systems aligned with offline social interaction patterns as much as possible, but their (current) design changes our social interaction patterns in return as well.
It's still fascinating to see how the world reimagines and reconsiders the nature of money. And not only in cryptopunks community, but in such a huge powerhouses as IMF as well.
If cryptocurrency is like bittorrent for finance, then this proposal is like downloading DRM-encumbered music from a walled garden marketplace. I think I'll pass.
> This currency could satisfy public policy goals, such as (i) financial inclusion, and (ii) security and consumer protection; and to provide what the private sector cannot: (iii) privacy in payments.
If financial inclusion and privacy in payments were actual public policy goals, cryptocurrencies would be a lot less attractive.
In reality, the real public policy goals are:
1) Financial exclusion: using denial of access to the state run financial system as a political, law enforcement, social control and tax collection tool.
3) "Know your customer" and "anti-money laundering" (AML/KYC) rules that remove privacy in payments to make it easier to use the state run financial system as described in 1). These regulations in practice exclude the the poorer people in society - the people Lagarde like means when she says "financial inclusion" - that can't meet requirements like "proof of address" and can't meet the minimal balance / month fee requirements that have come about from these AML/KYC regs.
If financial inclusion and privacy in payments were actual public policy goals, cryptocurrencies would be a lot less attractive.
In reality, the real public policy goals are:
1) Financial exclusion: using denial of access to the state run financial system as a political, law enforcement, social control and tax collection tool.
3) "Know your customer" and "anti-money laundering" (AML/KYC) rules that remove privacy in payments to make it easier to use the state run financial system as described in 1). These regulations in practice exclude the the poorer people in society - the people Lagarde like means when she says "financial inclusion" - that can't meet requirements like "proof of address" and can't meet the minimal balance / month fee requirements that have come about from these AML/KYC regs.
I don't disagree with your points around the public policy goals of currency, but suggesting that minimal balance / monthly fee requirements come from AML/KYC regulations is quite the stretch. I'd wager those come from greed, pure and simple.
The banks charge those fees because they can. If we got rid of AML/KYC regulations and reduced the banks' costs around regulatory compliance, do you think the banks will pass the savings onto customers or pocket the difference? I know where I'm putting my bet.
The banks charge those fees because they can. If we got rid of AML/KYC regulations and reduced the banks' costs around regulatory compliance, do you think the banks will pass the savings onto customers or pocket the difference? I know where I'm putting my bet.
> If we got rid of AML/KYC regulations and reduced the banks' costs around regulatory compliance, do you think the banks will pass the savings onto customers or pocket the difference?
The existing banks, on their own? Surely not. But complicated and expensive regulations like that are the main reason we don't have more, smaller banks, and they would be happy to gain customers by charging lower fees. Then the existing banks would have to compete and do the same.
The existing banks, on their own? Surely not. But complicated and expensive regulations like that are the main reason we don't have more, smaller banks, and they would be happy to gain customers by charging lower fees. Then the existing banks would have to compete and do the same.
We need two things to operate productively without the State:
Cryptocurrency is one. It does the job of the mint.
The second piece, is we need a set of contracts that allow us to never take ownership of assets anymore.
Instead of "here's a dollar, please may I have a taco", we'll say "please may I have a taco, here are my contracts" and the chef will ask the computer, "are this one's contracts aligned enough that I give them a taco?"
There will be times you need to create revenue, which will be taxable, but it won't be the default.
I'm not talking about tax evasion here. I'm talking about operating in a way that human activity can't be modeled as a transfer of assets. Essentially widening your friends and family circle to temporarily include arbitrary individuals.
Cryptocurrency is one. It does the job of the mint.
The second piece, is we need a set of contracts that allow us to never take ownership of assets anymore.
Instead of "here's a dollar, please may I have a taco", we'll say "please may I have a taco, here are my contracts" and the chef will ask the computer, "are this one's contracts aligned enough that I give them a taco?"
There will be times you need to create revenue, which will be taxable, but it won't be the default.
I'm not talking about tax evasion here. I'm talking about operating in a way that human activity can't be modeled as a transfer of assets. Essentially widening your friends and family circle to temporarily include arbitrary individuals.
I don't think they understand that the purpose of cryptocurrencies is to remove the need for trust. Trust in the math, not the banker. AKA the economy should be built off the monetary system rather than the monetary system being built off the economy.