6 ms·

Details are scarce at this point but this is just another data point suggesting that DeFi doesn't solve a real problem. Yes, the existing financial infrastructu

Details are scarce at this point but this is just another data point suggesting that DeFi doesn't solve a real problem. Yes, the existing financial infrastructure is slow and costly but in a pretty good local maximum. Touted features of DeFi that are arguably worse than what they're replacing: 1. "Your contract is executed exactly as it is written! No loopholes, no shenanigans!" Clearly this is a double edged sword. Since code is a lot more complex than written human language, expressing intent can be very difficult. In a regular contract, if your counterparty figures out a clever way to cheat you out of your money, you can sue them in court and the judge will very likely tell them to give the money back and slap them with damages too. Apparently in Ethereum Smart Contracts, you can accidentally/inadvertently allow for a gas fee that 230x the transaction value and there's literally nothing anyone can do for you. 2. "No need for expensive lawyers/bankers!" True! But you now need to pay expensive computer scientists to write/review your contracts, who are probably at least twice as expensive as the bankers. Oops. 3. "Not centralized! Don't be chained by unjust government regulation!" True, but also probably not optimal for 99.5% of people. Anyone who has been banned from Venmo for making a North Korea joke knows how annoying AML/KYC is, but by and large the regs (in the US at least) are written in a way that protect the average joe and society as a whole.

j_walter 5y ago
Not just 23x...but 230x transaction value ($23M to send $100K). Not to mention it was 7,700,000x the going rate for the other transaction in that same block.
dilap 5y ago
This might be a very naive question, but would it not be the case that many contracts would be essentially similar in nature? So while weird loopholes would exist at first, over time there could be a large library of tried-and-true contracts which could be applied to the situation at hand?
- j_walter 5y ago
  You are correct and there is already a bunch of off the shelf contracts that are basically plug and play...however it takes little knowledge to actually publish said contract and a small typo can either make the contact never accessible or easily exploitable.
- ljm 5y ago
  That is arguably tending towards centralisation, as there will then be a market of brokers that know which contracts to suggest, and then there will be other places that will only accept contracts from a known broker. There will, of course, be the more expensive brokers that can exploit loopholes, but even as a stretch this isn't vastly different to a bank account in the Caymans. Of course, you don't have to participate - you can hold the 'cash' equivalent or concoct your own scheme, but you're not as protected from loss as you otherwise might be (provided there's an insurance package of sorts, in the absence of regulation).
  - leppr 5y ago
    Smart contract code is open-source[1], so the centralization you're talking about shouldn't be confused with the (de)centralization of power people talk about when they talk about the decentralized web or decentralized finance. Circling around a small set of trusted open-source technologies doesn't cause centralization of power, as we can see with GNU/Linux, HTML/JS, ... [1]: On many platforms smart contracts are stored in the form of high-level interpreted languages. On Ethereum, the blockchain stores EVM code (assembly), but contracts that haven't "verified" their source, typically by uploading the high-level code to etherscan.org, are seldom used (with some notable exceptions).
    - ljm 5y ago
      I contend that it doesn't cause centralisation of power, but centralisation and power are inevitable should the project hit the mainstream, proper: - GNU/Linux - GNU/Linux is open source but centralised. The userspace is the part that is distributed, via operating systems, and the source control is distributed, via git. But it's all for one Linux kernel. You can also build your own kernel, but that doesn't really make linux 'decentralized'. Similarly, Linux for a lot of people means 'Ubuntu'. - HTML/JS - this is centralised under WhatWG/W3C, etc. Arguably, these days, it's actually centralised under Chrome, because what Chrome does eventually becomes the spec. You can freely build your own implementations of HTML and JS/Ecmascript but most likely, you are using the centralised implementation via webkit, blink, or gecko. So, the fact that smart contract is open-source doesn't really mean anything. It'll grow big, then as a matter of convenience it will start to consolidate. git, for example, is decentralised, but git forges (github, gitlab, etc.) provide centralisation as a convenience.
- jonny_eh 5y ago
  Imagine a scenario where a popular contract is discovered to have a serious flaw years down the line, and they can't be patched.
- tschwimmer 5y ago
  The key difference (which is categorical) is that if there is a latent bug in one of these contracts that goes unnoticed and then gets exploited later, you're still out the money. Intent doesn't matter. There is essentially a 0% chance of such an outcome with a regular contract (at least in US contract law). Intent is a cornerstone of contracts in the real world. Just because someone came up with a clever "exploit" doesn't entitle them to rip you off. This is a very good thing.
- TimJRobinson 5y ago
  On Ethereum there's an organization called OpenZepplin that does exactly that. They have common code for making contracts that are ownable, upgradeable, follow NFT / ERC20 standards etc. Makes it much easier to write safe solidity code.
followLink 5y ago
Hard to say if this satirical or not ("Etherium") but will respond in good faith. 1. Same argument can be used in the opposite way, you can verify the counter party will act as expect, there is no need for costly litigation as backstop. 2. An engineer reviewing a contract once will not be more expensive than hiring a full time banker to review each transaction or a lawyer to pursue a legal battle. 3. Decentralization and regulation are not mutually exclusive.
- orwin 5y ago
  Agree with 1/, disagree with 2 and i have a different point of view on th 3rd point. A lawyer can review a contract, miss a loophole/backdoor, and then argue to the court that the backdoor was placed there in bad faith, recoup his client losses. You can't do that with code. If you load a backdored code, you can't argue "this is unfair" to a court of justice. On the 3rd point, i think "good" crypto (and by that i mean non-obvious shitcoins) tend to centralize a lot.
  - cco 5y ago
    > You can't do that with code. If you load a backdored code, you can't argue "this is unfair" to a court of justice. Why not? Assuming the miner and you are in the same jurisdiction, what prevents you from suing them? Seems relatively straightforward.
    - orwin 5y ago
      Because of limited responsability mainly. Did anyone attack microsoft for the catastrophic 0day ~2 years ago?
jrm4 5y ago
100%. I keep harping on the point that "smart contract" is such a bad name for these things. A real-life contract is not the execution of the task(s) itself, it's the legal recording and reinforcement of the promise(s) made by the parties if/when things go wrong. Smart executor does sound pretty grimy though :)
- unyttigfjelltol 5y ago
  In normal contracting the paper is the source code and you need to find some real live people to be the compiler. Sometimes it's everyday people, sometimes operations professionals, developers, finance or legal personnel. Ultimately you always can fall back on the court or arbitrator as the ultimate compiler and interpreter. So, different compilers maybe for smart contracts? ....
  - jrm4 5y ago
    I'm trying to roll with you on this idea, but I can't help but think that this is just a square-peg round-hole thing. I am a (mostly non-practicing) lawyer; but I teach IT full time. Like "A.I," a lot of people dream about the idea of revolutionizing or strongly improving on law via code, and I don't see it happening, ever. It's this well meaning and seductive idea that we can leverage the superior power of computery thought and ideas to "correct" the foibles of humans -- but I think "human disputes" are too slippery to be bound by code, and the beauty and triumph of "law" as a system is that it too is slippery enough to manage it reasonably. To me it's strongly related to the complete layman fantasy of "well, they should have just written the law more clearly and everything would be fine."
    - unyttigfjelltol 5y ago
      Yes, and the reverse also is an interesting thought exercise-- if this paper contract really was a piece of software, what would folks do to QA it, write it, interpret it. I mean, paper contracts are written using hopelessly buggy and ambiguous human languages theoretically capable of so many non-harmonious interpretations. The solution really is about friendly, trusted compilers approaching the contracts from a fundamentally cooperative direction. You write a contract with your trusted partner on a napkin at dinner. And when trust fails, courts interpret in an explicitly prosocial way.
leppr 5y ago
This accident actually proves the exact contrary. Despite these obvious flaws, people are still using DeFi. There must be something of value in there that you're missing. And it's not speculation, as there are way better ways to speculate in crypto than by using DeFi (shittokens, NFTs, ...).
- Lazare 5y ago
  Lottery tickets aren't a good investment, but people still buy them. It's certainly true that there is activity in the DeFi space, and it's certainly possible that this activity is because DeFi is just so gosh darn amazing at solving people's need for basic financial products. ...at least in theory. But if you're going to make that rather extraordinary claim, I think you need to provide some extraordinary evidence.
  - leppr 5y ago
    I'm disputing the logical conclusion from the facts at hand that no real problem is solved, not providing a value judgment about DeFi. It's not an extraordinary claim but common knowledge in the start-up investment land that when you see an ugly, flawed product get an unexplained amount of engagement, you become interested. > Lottery tickets aren't a good investment, but people still buy them. Lottery tickets solve the real problem of hopelessness about improving one's financial situation through work. It might not be a worthwhile investment in financial terms, but it has its use just like entertainment media or alcohol.
- tschwimmer 5y ago
  People inject heroin, drink liters of soda and gamble their life savings away too, but none of that is evidence that these are healthy or socially valuable activities. My explanation of the prevalence of DeFi is that it's currently at or near the peak of the hype cycle and there's a ton of money to be made either by trying to legitimatize the ecosystem or else via run of the mill pump and dumps.
- roca 5y ago
  "There must be something of value in these tulip bulbs because everyone is investing in them."
herval 5y ago
> Since code is a lot more complex than written human language, expressing intent can be very difficult. Isn’t that the other way around? Natural language remains an unsolved problem, and formal verification of software is pretty much a requirement for compilers. Your example is evidence of that: intent in natural language is easy to distort and hard to prove, thus making it much more complex
- Traster 5y ago
  Well let's put it this way - in natural language, we say stuff and implicitly assume we've done our best to express intent, but it may not be perfect (hence lawyers). In smart contracts you express your intent into a contract and then we pretend there was no intent. There is only the contract. And that's a massive floor.
  - herval 5y ago
    Right. Natural language is more complex and nuanced.
asciimike 5y ago
James Mickens' "Blockchains are a bad idea" talk [1] is a great explanation that discusses exactly that. [1]: https://www.youtube.com/watch?v=15RTC22Z2xI https://www.youtube.com/watch?v=15RTC22Z2xI

j_walter 5y ago

Not just 23x...but 230x transaction value ($23M to send $100K). Not to mention it was 7,700,000x the going rate for the other transaction in that same block.

dilap 5y ago

This might be a very naive question, but would it not be the case that many contracts would be essentially similar in nature? So while weird loopholes would exist at first, over time there could be a large library of tried-and-true contracts which could be applied to the situation at hand?

j_walter 5y ago
You are correct and there is already a bunch of off the shelf contracts that are basically plug and play...however it takes little knowledge to actually publish said contract and a small typo can either make the contact never accessible or easily exploitable.
ljm 5y ago
That is arguably tending towards centralisation, as there will then be a market of brokers that know which contracts to suggest, and then there will be other places that will only accept contracts from a known broker. There will, of course, be the more expensive brokers that can exploit loopholes, but even as a stretch this isn't vastly different to a bank account in the Caymans. Of course, you don't have to participate - you can hold the 'cash' equivalent or concoct your own scheme, but you're not as protected from loss as you otherwise might be (provided there's an insurance package of sorts, in the absence of regulation).
- leppr 5y ago
  Smart contract code is open-source[1], so the centralization you're talking about shouldn't be confused with the (de)centralization of power people talk about when they talk about the decentralized web or decentralized finance. Circling around a small set of trusted open-source technologies doesn't cause centralization of power, as we can see with GNU/Linux, HTML/JS, ... [1]: On many platforms smart contracts are stored in the form of high-level interpreted languages. On Ethereum, the blockchain stores EVM code (assembly), but contracts that haven't "verified" their source, typically by uploading the high-level code to etherscan.org, are seldom used (with some notable exceptions).
  - ljm 5y ago
    I contend that it doesn't cause centralisation of power, but centralisation and power are inevitable should the project hit the mainstream, proper: - GNU/Linux - GNU/Linux is open source but centralised. The userspace is the part that is distributed, via operating systems, and the source control is distributed, via git. But it's all for one Linux kernel. You can also build your own kernel, but that doesn't really make linux 'decentralized'. Similarly, Linux for a lot of people means 'Ubuntu'. - HTML/JS - this is centralised under WhatWG/W3C, etc. Arguably, these days, it's actually centralised under Chrome, because what Chrome does eventually becomes the spec. You can freely build your own implementations of HTML and JS/Ecmascript but most likely, you are using the centralised implementation via webkit, blink, or gecko. So, the fact that smart contract is open-source doesn't really mean anything. It'll grow big, then as a matter of convenience it will start to consolidate. git, for example, is decentralised, but git forges (github, gitlab, etc.) provide centralisation as a convenience.
jonny_eh 5y ago
Imagine a scenario where a popular contract is discovered to have a serious flaw years down the line, and they can't be patched.
tschwimmer 5y ago
The key difference (which is categorical) is that if there is a latent bug in one of these contracts that goes unnoticed and then gets exploited later, you're still out the money. Intent doesn't matter. There is essentially a 0% chance of such an outcome with a regular contract (at least in US contract law). Intent is a cornerstone of contracts in the real world. Just because someone came up with a clever "exploit" doesn't entitle them to rip you off. This is a very good thing.
TimJRobinson 5y ago
On Ethereum there's an organization called OpenZepplin that does exactly that. They have common code for making contracts that are ownable, upgradeable, follow NFT / ERC20 standards etc. Makes it much easier to write safe solidity code.

followLink 5y ago

Hard to say if this satirical or not ("Etherium") but will respond in good faith. 1. Same argument can be used in the opposite way, you can verify the counter party will act as expect, there is no need for costly litigation as backstop. 2. An engineer reviewing a contract once will not be more expensive than hiring a full time banker to review each transaction or a lawyer to pursue a legal battle. 3. Decentralization and regulation are not mutually exclusive.

orwin 5y ago
Agree with 1/, disagree with 2 and i have a different point of view on th 3rd point. A lawyer can review a contract, miss a loophole/backdoor, and then argue to the court that the backdoor was placed there in bad faith, recoup his client losses. You can't do that with code. If you load a backdored code, you can't argue "this is unfair" to a court of justice. On the 3rd point, i think "good" crypto (and by that i mean non-obvious shitcoins) tend to centralize a lot.
- cco 5y ago
  > You can't do that with code. If you load a backdored code, you can't argue "this is unfair" to a court of justice. Why not? Assuming the miner and you are in the same jurisdiction, what prevents you from suing them? Seems relatively straightforward.
  - orwin 5y ago
    Because of limited responsability mainly. Did anyone attack microsoft for the catastrophic 0day ~2 years ago?

jrm4 5y ago

100%. I keep harping on the point that "smart contract" is such a bad name for these things. A real-life contract is not the execution of the task(s) itself, it's the legal recording and reinforcement of the promise(s) made by the parties if/when things go wrong. Smart executor does sound pretty grimy though :)

unyttigfjelltol 5y ago
In normal contracting the paper is the source code and you need to find some real live people to be the compiler. Sometimes it's everyday people, sometimes operations professionals, developers, finance or legal personnel. Ultimately you always can fall back on the court or arbitrator as the ultimate compiler and interpreter. So, different compilers maybe for smart contracts? ....
- jrm4 5y ago
  I'm trying to roll with you on this idea, but I can't help but think that this is just a square-peg round-hole thing. I am a (mostly non-practicing) lawyer; but I teach IT full time. Like "A.I," a lot of people dream about the idea of revolutionizing or strongly improving on law via code, and I don't see it happening, ever. It's this well meaning and seductive idea that we can leverage the superior power of computery thought and ideas to "correct" the foibles of humans -- but I think "human disputes" are too slippery to be bound by code, and the beauty and triumph of "law" as a system is that it too is slippery enough to manage it reasonably. To me it's strongly related to the complete layman fantasy of "well, they should have just written the law more clearly and everything would be fine."
  - unyttigfjelltol 5y ago
    Yes, and the reverse also is an interesting thought exercise-- if this paper contract really was a piece of software, what would folks do to QA it, write it, interpret it. I mean, paper contracts are written using hopelessly buggy and ambiguous human languages theoretically capable of so many non-harmonious interpretations. The solution really is about friendly, trusted compilers approaching the contracts from a fundamentally cooperative direction. You write a contract with your trusted partner on a napkin at dinner. And when trust fails, courts interpret in an explicitly prosocial way.

leppr 5y ago

This accident actually proves the exact contrary. Despite these obvious flaws, people are still using DeFi. There must be something of value in there that you're missing. And it's not speculation, as there are way better ways to speculate in crypto than by using DeFi (shittokens, NFTs, ...).

Lazare 5y ago
Lottery tickets aren't a good investment, but people still buy them. It's certainly true that there is activity in the DeFi space, and it's certainly possible that this activity is because DeFi is just so gosh darn amazing at solving people's need for basic financial products. ...at least in theory. But if you're going to make that rather extraordinary claim, I think you need to provide some extraordinary evidence.
- leppr 5y ago
  I'm disputing the logical conclusion from the facts at hand that no real problem is solved, not providing a value judgment about DeFi. It's not an extraordinary claim but common knowledge in the start-up investment land that when you see an ugly, flawed product get an unexplained amount of engagement, you become interested. > Lottery tickets aren't a good investment, but people still buy them. Lottery tickets solve the real problem of hopelessness about improving one's financial situation through work. It might not be a worthwhile investment in financial terms, but it has its use just like entertainment media or alcohol.
tschwimmer 5y ago
People inject heroin, drink liters of soda and gamble their life savings away too, but none of that is evidence that these are healthy or socially valuable activities. My explanation of the prevalence of DeFi is that it's currently at or near the peak of the hype cycle and there's a ton of money to be made either by trying to legitimatize the ecosystem or else via run of the mill pump and dumps.
roca 5y ago
"There must be something of value in these tulip bulbs because everyone is investing in them."

herval 5y ago

> Since code is a lot more complex than written human language, expressing intent can be very difficult. Isn’t that the other way around? Natural language remains an unsolved problem, and formal verification of software is pretty much a requirement for compilers. Your example is evidence of that: intent in natural language is easy to distort and hard to prove, thus making it much more complex

Traster 5y ago
Well let's put it this way - in natural language, we say stuff and implicitly assume we've done our best to express intent, but it may not be perfect (hence lawyers). In smart contracts you express your intent into a contract and then we pretend there was no intent. There is only the contract. And that's a massive floor.
- herval 5y ago
  Right. Natural language is more complex and nuanced.

asciimike 5y ago

James Mickens' "Blockchains are a bad idea" talk [1] is a great explanation that discusses exactly that. [1]: https://www.youtube.com/watch?v=15RTC22Z2xI https://www.youtube.com/watch?v=15RTC22Z2xI