Hacker News new | past | comments | ask | show | jobs | submit login
Decentralized trust graph for online value exchange without a blockchain (2017) (archive.org)
123 points by spolu 46 days ago | hide | past | favorite | 39 comments



>Conversely, cryptocurrencies such as Bitcoin or Ethereum have decentralized the process of issuing and managing a currency. But while the operations of such currencies, based on blockchains, have been fully decentralized, the trust graph of these cryptocurrencies have remained entirely centralized. Everyone need to trust Bitcoin to transact in Bitcoin, and everyone needs to trust Ethereum to transact in Ethereum or assets issued on the Ethereum blockchain. The centralized nature of the trust involved in these cryptocurrencies being actually at the core of how these currencies operate, as the only viable way to properly incentivize a proof-of-work system.

I don't think it's useful to equivocate the trust one places in an open decentralized system with the trust someone puts in a custodian of their money to not run off with it. That's like equivocating trusting end-to-end encryption to keep your messages secure with trusting Facebook to keep your messages secure. Sure, in both cases there's a risk, but one is based on your understanding of an open system, and the other is based on a guess that the custodian you're trusting your data to won't expose it by choice or accident. Just because the word "trust" can be used for both doesn't mean there aren't significant differences between them.

I think the system described is interesting, but for regular human commerce purposes, I think the downsides of it (managing IOUs from many different parties, needing to have a trust-path between users that could steal from you, needing to keep a node online always or trust your money to someone else who does so) make it less useful than cryptocurrencies. Having a common currency between many different parties is much easier to work with than trying to figure out a workable way to value IOUs from many different parties. Needing an established trust-path between users seems like a large obstacle in the internet age where I might suddenly want to transact with anyone on the planet.

The scalability of this system seems interesting, but I think the scalability upgrades that cryptocurrency is getting from sharding and rollups will make cryptocurrency more than scalable enough for human commerce purposes. I wonder if details of this system would be more useful in some kind of scenario of machine swarms bartering with each other, maybe in a situation where a common currency specifically doesn't make sense and is undesirable, or where there's no reasonable-latency access to the common blockchain because of the isolation of the network. Or maybe there's a technique here that's useful for cross-currency or cross-blockchain transactions.


That quoted paragraph also caught me up upon first read. After I thought about it I do agree that there is some level of trust in the protocol itself and actions of other actors using that protocol can influence that trust. But I suppose that the trust needs to be applied at some level. Trusting end-to-end encryption is a real trust. If someone comes along and finds a way to factor large prime integers the system would break and trust would dissipate.

It seems to me that its about realigning trust in this case. And its easier to trust a process that uses mechanics that force behavior to conform to the properties you desire than to trust a process that uses mechanics to enable behavior to conform to the desired properties. You still have to place trust in both implementations nevertheless.

In the case of Ethereum what is the trust? The protocol, the implementation? What is the trust in a bank? The same but with a bunch of human factors around the enabling the bank to function as you desire.

Settle appears to be a step back and try to solve the problem currency has in general by reverting to a barter system. The purpose of a currency however is to use a common valuation system.


> factor large prime integers

factoring large prime integers is very easy =P


Didn't Ethereum very early on push an update that undid a $50 million hack?


Yeah, they figured at that early point in the chain's life, having one hacker control so much of all ETH in existence was worth saving it.

It was a hard coded contract change to fix the exploit. No rollbacks though, the funds that were already stolen, stayed stolen. They had to be 'stolen' back, recovered, using the same hack the hacker was using.


Yep, the "trustless for thee but none for thanks" model.

On the plus side the DAO hackers are doing very well for themselves economically, the old chain, Ethereum Classic, is having a moment [1].

[1] https://coinmarketcap.com/currencies/ethereum-classic/


Do you mean equate instead of equivocate?


I think the idea of money as credit is very interesting (look at my HN comment history for my ramblings on the topic!).

> I think the system described is interesting, but for regular human commerce purposes, I think the downsides of it (managing IOUs from many different parties, needing to have a trust-path between users that could steal from you, needing to keep a node online always or trust your money to someone else who does so) make it less useful than cryptocurrencies. Having a common currency between many different parties is much easier to work with than trying to figure out a workable way to value IOUs from many different parties. Needing an established trust-path between users seems like a large obstacle in the internet age where I might suddenly want to transact with anyone on the planet.

You are right, these are disadvantages but other than "trust paths" the complexity can be mitigated with:

1. Market makers who buy IOUs at a discount and swap them for _their_ IOU which can be more trusted. Essentially you want a little bit of centralisation around parties which are known to always make good on their liabilities. These parties could be fully automated (think like Ethereum DAO) and transparent. In some ways these market makers are like banks of today but instead of lending their credit into existence, like banks do, they require people to buy it (usually) at a premium.

2. Insurance or a credit derivatives market so people can hedge against counterparts defaults.

3. If everyone agrees on a common numeraire and wallets are sophisticated to show an aggregate balance in that numeraire taking into account credit risk then that could remove much of the complexity for users. Integrated with market makers, you could "auto swap" IOUs to issuers which you prefer. Wallets can also provide a credit risk break down on all counterparts. Agree this is more complicated than current notions of money.

4. Securitisation markets. Package up IOUs into tranches. Traders can speculate on various levels of credit quality.

5. Sensible reputation management. Non-invasive performance tracking of debtors. Did they meet margin requirements? Did they meet all coupon payments? Etc. Can also take into account degree of co-operation with other users, for example do they accommodate restructuring and help others meet their liabilities?

I don't like crypto because there's essentially zero accountability for issuers and that kind of environment is optimal for scammers. Crypto behaves like synthetic equity/commodity instruments with no accountable issuers. In contrast, with a credit based system, issuers are responsible for their issued liabilities and are expected to make good on them. In crypto, no-one thinks in terms of liabilities and so you get ridiculous projects like "Synthentix" who are collateralising a USD Stablecoins with their own issued tokens. For anyone with a basic understanding of financial risk and accounting, their approach is mad.

A credit based system is fairer because anyone can issue credit and the onus in on issuers to ensure they have the necessary reputation for others to accept their credit. Furthermore, credit-like instruments are natural Stablecoins - providing all agree on a numeraire, the value of credit is a function of credit risk which can be successfully managed in most cases.


>1. Market makers who buy IOUs at a discount and swap them for _their_ IOU which can be more trusted.

Doesn't this go against decentralization? Why would anyone want rojeee IOUs when they can instead trade US federal government IOUs (aka. US dollars)? What's the advantage in managing IOUs from a bunch of different entities and having to pay market makers every time you transact?

>These parties could be fully automated (think like Ethereum DAO) and transparent

This is a bit handwavy. How does the system know how much rojeee IOUs are worth? Your ability to repay is based off a multitude of factors that can't be captured on the blockchain.

>2. Insurance or a credit derivatives market so people can hedge against counterparts defaults.

>4. Securitisation markets. Package up IOUs into tranches. Traders can speculate on various levels of credit quality.

All of this is going to increase complexity exponentially, and for what benefit?


Dollars aren't an IOU. Their advantage is all creditors and the government itself is required to accept it as payment, and practically speaking, all businesses will also accept it even though they aren't legally required to. But owning them doesn't mean the government owes you anything.

The advantage of an actual IOU is revenue and expense streams, even when they are very predictable, may not happen on the same schedule, or one is seasonal but the other is not. So allowing the buyer of a good or service to pay with money they don't yet have (and allowing the receiver to count that as revenue according to accrual GAAP) is critical to how businesses function.

Not forcing them to trust each other is the only thing that makes this possible, and that happens via banks and other market makers. This arguably even requires some level of centralization because many factors make this easy for a bank but pretty difficult if not impossible for some arbitrary third party, i.e. banks are extremely well capitalized, heavily regulated and audited, might have multi-century histories of reliably making good on their guarantees, have extremely specialized departments dedicated to assessing default risk. Paying for this reliability and trustworthiness has been deemed a worthwhile cost of business for effectively as long as businesses have existed.

Making this actually free would be nice, but I don't see how it's possible. Settle doesn't seem to solve that from reading this. It just trades explicit cost for risk, which you can already do if you prefer risk. Cryptocurrencies don't solve it either. Something like lightning is fine for microtransactions, but anything large enough that you would otherwise involve a bank or escrow agency can only be handled by the blockchain itself with its massive transaction costs, which happen to be presently masked from buyers and sellers due to the ability of miners to get rich off of speculative frenzy, but when that stops being the case, suddenly you're going to need to pay explicitly to move large amounts of value, more than you would have been paying a bank.


> I don't like crypto because there's essentially zero accountability for issuers and that kind of environment is optimal for scammers.

Please expand on this as I'm almost certain this doesn't apply to BTC, ETH, and other reputable chains.


Uhh... if you define reputable by market cap and purported volume then... [1]

[1] https://ag.ny.gov/sites/default/files/2021.02.17_-_settlemen...


Sure, it's more an observation of the mechanics underpinning cryptocurrencies. My statement was actually incorrect in that there are not actually "issuers" per se in crypto and that can be a problem.

Cryptocurrencies like BTC, ETH and whatever else are not "money" in the sense that we generally understand. This money is either a credit on a commercial bank or a credit on a central bank. These entities are responsible for managing the liquidity and solvency of their balance sheet such that the credit you hold retains value (Yes, I know there is inflation but that's an orthogonal concern, I would argue). There is a governance and legal framework underpinning how money works in this "mundane World". Everything in the mundane world is a legal agreement. Financial agreements are always someone's liability and that's a good thing because someone is always accountable if something goes wrong.

Cryptocurrency behaves more like some kind of synthetic commodity. By that I mean that it inherently has no value and has no use for anything other than being a "token". When cryptocurrency is created, it has no issuer. There's no accepted legal framework or explicit governance framework for holding people accountable if something goes wrong (other than code is law) and things go wrong all the time. There are rug pulls, exchange rate crashes and project failures are ten a penny etc... and this is all happening while the crypto world is in a mega bull market! What happens when the next bear market comes around? All the projects that _seem_ viable now will suddenly become unviable. All the debt positions collateralised with sketchy crypto will unwind en masse. I suspect the tokens for many projects out there will trade close to zero. People who invested in various projects from algo stablecoins to lending protocols will suddenly find that they hold worthless tokens with zero recourse. By this point the insiders would have exited into BTC/ETH/fiat ready to start the next round of "projects".

Maybe this is OK. I mean, it is what it is and clearly some people are fine with that. I think we can do better though. That's why I'm interested in credit based monetary systems because reputation is a key part, so participants are accountable for their actions. This is fundamental for any significant real world adoption. Furthermore, credit based instruments do have intrinsic value which is a function of issuer credit risk. Such instruments are more stable than "synthetic commodities" and have more utility for real world uses.

Having said all that. I work in the crypto world and quite enjoy it. I just don't think it's as good as some would have you believe.


Cryptocurrencies can still be subject to hard forks which allow seizure transactions. Likely in the future governments will demand such changes, if they don't outlaw cryptocurrencies entirely. Furthermore, bitcoin's value is highly volatile. So your initial investment of $100 might be worth $5000 or $0.10 in 3 years. You might not have to trust the network itself, but you have to trust the value of the asset to remain stable enough. Which it isn't due to multiple factors.


Not sure how you think governments could demand changes from major cryptos. They're not currently beholden to governments, and it seems unlikely they'll become more beholden as they grow larger.

Volatility is ameliorated by stablecoins pegged to fiat currencies. Furthermore, the long-term goal is to not need an "interface to the real world", because you will be paid in crypto and you will pay for things in crypto. Even then, who exactly do you mean when you say "the people providing the interface"? I don't need a middle-man to agree to exchange crypto for fiat with someone. You can use one, sure, but it's not required.


Cryptocurrencies are in the process of becoming more beholden, or at least more outlawed: https://www.trtworld.com/magazine/what-do-turkey-s-cryptocur...

Governments have the guns. I doubt that cryptocurrencies will win in the long run, unless they provide hooks for governments to project their power, which includes asset seizure transactions.

As for the people providing the interface, I've since edited my comment. My point used to be that exchanges like coinbase engage in massive market manipulation to aid their own goals, but it's not the only cause for the volatility of cryptocurrencies like bitcoin, so I edited it out.


Several governments can trivially do a 51% attack and take over Bitcoin if they so wish, for example:

- The Chinese government can get all the main pool/farm owners (all self-doxed) in a room, offer them the choice between being executed or send all blocks their way for central validation. If they choose execution, send government employees to reboot the farms with new code.

- The Taiwanese government can cease the next batch of mining kit as it comes out of TSMC and lol all the way to the (central) bank.

- The US government can tell Coinbase "regulate this thing as told or die". Then all the suits who now own crypto via suit-friendly custodians will in a panic coordinate and fund a 51% attack to save their "investment".

Just a few ideas :-).


A government could easily mandate block lists. All legitimate businesses dealing with cryptocurrencies would be required to not consider "blocked" coins as valid. I can even realistically envision broad support for such a change, when some "character" who is sufficiently evil becomes known (via non-stop media targeting I suspect) to be empowered by a certain cryptocurrency.

I suppose there are some blockchains where it is not possible to follow balances between transactions. I'm not sure what would happen with those.


Coinjoins and whirpooling bitcoin is becoming cheaper, easier, and more widely adopted...chain hopping is also becoming easier thanks to decentralized exchanges and atomic swaps. If the government wants to censor transactions they have to expend many times more energy than any person with a free wallet and 5-10 minutes. Enforcement becomes cost prohibitive as the network grows and more people use these tools.


I think this idea has been implemented in Offset https://www.offsetcredit.org

HN thread: https://news.ycombinator.com/item?id=23438241


It's true. In addition, spolu gave some very useful pieces of advice during the design of Offset. Too bad settle.network is not live anymore.

I am not sure whether https://settlenetwork.com is actually affiliated with the original settle.network written by spolu.


> I am not sure whether https://settlenetwork.com is actually affiliated with the original settle.network written by spolu.

See also https://news.ycombinator.com/item?id=27082751


Read early versions of the Ripple white paper, before they decided to make XRP a store of value. The compulsion to cash in on a settlement layer by establishing your own scarce medium of exchange is too great.


Just to be pedantic, what happened is the original founder sold it to some skeezy guys who wanted to cash in on the digital currency crazy, and those skeezy guys added XRP as a way to cash in.

So it's not quite that the original founder felt pressure to create the token, but he did feel some pressure to make money, and eventually that led to a token.


Ripple didn't exist when XRP was created. ripplepay now rumplepay.com was probably what you meant.


Aspects of this sound similar to Stellar and Ripple. Could folks with more knowledge elaborate on the differences?


It seems so. Its the notion of "Trust Lines"[0]

All this thing does is limit potential losses from fraud. Not eliminating.

The main value proposition of a blockchain is to solve the "principal-agent" problem [1] and gis simply reduces the risk but doesn't remove it entirely.

Plus some of the requirements on a node having to be online (in a decentalized, byzantine environment) all the time are unrealistic.

This seems more like ripple.

Stellar uses a more sophisticated notion of "quorum slices" and is resistant to byzantine faults [2]

[0]: https://trustlines.foundation/faq.html

[1]:https://as1ndu.xyz/2021/04/clarifying-the-blockchain-proposi...

[2]: https://youtu.be/vmwnhZmEZjc


The Stellar Network relies on nodes declaring quorum slices, which can be understood as trust relationships, but it implements a single global blockchain. So I don't think it's very similar.

Edit: one similarity is that token issuers in Stellar can remain authoritative on their token.


I didn't know this project but if you visit their website today you can see that Stellar is listed there as a reference.


This is a curious post. There is apparently a Settle Network and a github repo. But S. Spolu is not listed among the execs running it.

https://settlenetwork.com/

https://github.com/spolu/settle

https://settlenetwork.com/settle-network/ (Scroll to bottom for the mug shots)


"settlenetwork.com" seems to be unrelated to this, other than having the words "settle" and "network" in it.


github repo points to it. and the archive.org page footer points to the repo.


Where?

I see a link to https://settle.network which isn't the same as https://settlenetwork.com


Did you click that link? It ends up in settlenetwork.com.


Ah, okay. My guess is he probably let "settle.network" expire, and the owners of the unrelated "settlenetwork.com" bought the domain and redirected it for anti-brand confusion, SEO or latent traffic reasons. I don't see anything technically related across them. The original settle.network was "without a blockchain", settlenetwork.com is very much "with a blockchain".

Edit: The "settle.network" domain as it sits now was first registered in 2019, well after the last activity on the github repo. So yeah, he let the domain lapse out.


Well, as I said, it's all "curious". Since the OP of this thread is in fact the owner of that repo (thus the README), possibly he can chime in and clear up the confusion. /g (I am just rubber necking here!)


Sorry, late reply. Indeed settle.network expired and is now owned by another party. Sorry for the ensued confusion.


I am working on something like this. Having a money in the system is a pretty hard problem (quantitative type theory-ish / lifted inference).

You can get around that with culture and using existing systems as settlement layer (for the time being).

Even then there is only one cryptocurrency that has a proper mechanism design for oracles (amoveo) - which you need if you want a trustless layer 2.

Datalisp (@ for telegram .is for binge-written PDF) is this project (that I just started) it's basically a vector clock for wrapping interfaces in authenticated data structures and Bayesian inference with logic programming for estimating / inferring trust.

By giving a useful framework for refining reproducibility we can build trust. Trust we need if we want a system to serve as a foundation for digital societies.

Francis Bacon said knowledge was possible and science could establish trust. Now we need that, automated.


Aren't gas fees and decentralization inversely correlated?




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: