Hacker News Reader

The criticism wasn't aimed at blockchain-as-a-technology, but at blockchain-as-a-panacea - at businesses who seem to think that slathering some blockchain on a problem is going to magically solve the problem.

How is the information verified in your system? Why does it need a blockchain?

You mentioned that the "criticism in the blog post at blockchain as a technology is misdirected", how so?

> How is the information verified in your system? Why does it need a blockchain?

The information is verified by creating an open system where anyone can point out wrong information (by providing oposing information). That is done with the help of a token in our system, which is issued to people who help establish consensus about what's true and what isn't. The token is the main reason we need a blockchain, and the gametheoretic aspects behind it dictate that there needs to be a token or similar in our system for it to work.

> You mentioned that the "criticism in the blog post at blockchain as a technology is misdirected", how so?

The "shit in - shit out" problem exists also with traditional systems and is not solely a blockchain-problem. That's what I meant to say with that.

I think that's about as much as I can share right now without angering my colleagues. The whitepaper which is coming out in a few weeks goes more into depth about the specifics of each part of the systems and provides some game-theoretic and economic foundations for the system.

> The information is verified by creating an open system where anyone can point out wrong information (by providing oposing information).

How do you combat "spam", were some users send thousands of fake registrations of well known art works?

How do you combat sockpuppet that send wrong opposing information to steal the art work of the original creator?

How do you combat sockpuppets that send wrong information to support fake claims?

If some piece of art has two claims, who decides which one is the correct?

> If some piece of art has two claims, who decides which one is the correct?

In the end I would say that all of your other questions can be reduced to this one, by adding weight to claims, which makes the quantity of claims and indentities irellevant.

We are not focused on the art market, but open to any kind of information. But for the example of the art ownership verification:

- Person A (true owner) claims that they own the Mona Lisa with a weight of 10 tokens

- Person B (non-owner) claims that they own the Mona Lisa with a weight of 10 tokens

- An additional actor X who knows about the ownership can add additional weight to the claim of Person A. If the consensus in the end decides (based on the weights) that Person A is the true owner, actor X will be rewarded. There are some specifics about how we construct the reward function that I can't talk about yet, but the core of it is that all participants in the network are incentivized to only submit true information, and also incentivized to combat any wrong information they see by providing conflicting true information.

If you are interested in the topic and would like to discuss more deeply feel free to reach out to me via the email on the website (I'm Max).

I can see and am grateful for your effort to answer the questions as helpfully as you can. But to me, when you say "the owner of the Mona Lisa is whoever has the most tokens in our system", it sounds like complete nonsense. The Louvre owns the Mona Lisa, and if your system disagrees with that, your system is broken.

I can definitely see where you are coming from since that mechanism of "most token wins", sounds very unintuitive and inherintly cheatable (and in a lot of places has rightfully bad connotations like "pay to win" in video games). When I first heard it I also had the same reaction. Sadly I don't think that I can offer any more information right now, but we are all very eager to publicize it and put it out into the open for critic ASAP.

>I can definitely see where you are coming from since that mechanism of "most token wins", sounds very unintuitive and inherintly cheatable (and in a lot of places has rightfully bad connotations like "pay to win" in video games).

It does indeed seem exactly that way. You said upthread:

>An additional actor X who knows about the ownership can add additional weight to the claim of Person A.

But how do you verify that X is someone "knows about the ownership"? What if X was someone A or B paid or incentivized to vouch for them so the consensus would be swayed? How do prevent bad actors from gaming the system in this way? If the asset in dispute is valuable enough it would certainly be worth it to a bad actor to pay a little to cheat and have the consensus fall on their side.

> What if X was someone A or B paid or incentivized to vouch for them so the consensus would be swayed?

If B were to pay X a bribe of amount N, so X enforces the statement with amount N, B could also directly enforce the statement with amount N, which would lead to exactly the same result. If X would receive less than N, to add a weight of N, they would be gifting some money to B, and vice versa. So between rational untrusted parties, a simple bribe would not make sense. There are some rationality assumptions relating to 51% attacks similar to PoS oriented systems that have to be met, but there are some mechanisms in place that should reduce the practical requirement regarding rationality among participating parties significantly.

> If the asset in dispute is valuable enough it would certainly be worth it to a bad actor to pay a little to cheat and have the consensus fall on their side.

Good question, but I'm afraid that's one of the questions I can't answer right now without leaking too much information.

> If B were to pay X a bribe of amount N, [...] B could also directly enforce the statement with amount N

Okay, better question: how do you deal with the (pathologic, and therefore approximately guarranteed to come up) situation where A's life savings are ten dollars and a horse, B has 100 million dollars, and B claims (with amount ≥ 11$) that B owns A's horse.

As long as the "100 million dollars" don't make up more than 50% of the tokens in the network, the incentives are supposed to play out in a way that all the other participants of the network contest the wrong statement made by B, and thus B isn't able to introduce information that is deemed wrong. So in general the situation should be preventable. There are some specific versions of that edge case that come to my mind that where that situation could indeed be unresolvable but in those cases that should be detectable and displayable to the users of the system.

As I said in my previous comment, there are some rationality requirements regarding some of the token amount.

> If B were to pay X a bribe of amount N, so X enforces the statement with amount N, B could also directly enforce the statement with amount N

You're assuming the bribe comes in your made up currency. B could bribe X with real money or out of band coerce/intimidate X with threats and violence.

> B could bribe X with real money

So a connected market with goods that can be bridged similar to the stock market.

> or out of band coerce/intimidate X with threats and violence

I'm not pushing for a world without governments and law enforcement.

I guess your idea is to have an ultimate veto power to revert any decision. So it's a centralized service that can run in a traditional database instead of a blockchain.

What will you do in polemic cases? For example, some people think that La Gioconda is own by the Italian people and should be returned to Italy (because Leonardo was Italian). What would you do if there is a huge campaign to claim that it is owned by some Italian museum? What about the Tutankhamen mummy? Alaska?

> I guess your idea is to have an ultimate veto power to revert any decision. So it's a centralized service that can run in a traditional database instead of a blockchain.

Not at all, and I don't know where you are getting that idea from. The system, as it is designed right now has no requirement for a single centralized or trusted party. It also has no requirement for any trust between the participants (which most systems in the same realm do).

> What will you do in polemic cases?

That's a good question, and one of the things we want to explore during iterations of the testnet. Without going into more specifics I would say that there is a part of the incentive mechanism that rewards a more clear-cut consensus (a 50/50 split on a boolean statement is the least favourable situation for participants on both sides). This unfavourable situation should then in turn fuel the contribution of more information so people can make a more informed rational decision which should then lead to a more clear consensus. There are also some higher level market mechanisms that we anticipate to develop which could be used to both stabilize and destabilize the consensus about a certain statement, like a off-chain PR campaign as you mentioned. We'll have to see how each of those can potentially impact a consensus and if that is something that needs a countermeasure in the core protocol.

I know it's not fair to critique your system without access to the code or the whitepaper, but it sure sounds like "truth is whatever the person with the most money to spend says it is."

It's almost like rule 0 of the modern internet. There will almost certainly be many cases in which there are more additional actors willing to pledge tokens stating that [arbitrary person] painted the Mona Lisa than there are additional actors willing to pledge tokens stating that Leonardo painted the Mona Lisa. In fact, given what so far seems like a tiny value in having information be correct, I would guess that the majority of works of art would end up attributed to the wrong artists given enough time.

I know you said your system isn't tracking this specifically, but the general principle remains true in any problem domain. The integrity of a common-consensus system depends entirely on how little trolls care about disrupting it.

"The integrity of a common-consensus system depends entirely on how little trolls care about disrupting it."

That is actually the key point why a blockchain (or ledger that allows for money-like transaction) is an important part for us. By binding what you say to a monetary reward people that act irrationaly and do it for the lulz will lose their money and ultimately have to leave the system by running out of funds.

Simmilarly, trolls could dimish the value of a PoW blockchain by creating and exploiting 51% attacks, but that hasn't happened with Bitcoin, since that is far too expensive to do it just for fun. Monetarily motivated attacks are of course a different discussion.

Of course, this depends on the cost and value of the truth being higher than the cost and value of a lulz.

The art example is a great demonstration of the mismatch. There is essentially zero value in the truth, so it was worth a bit of the author's time for the lulz.

With cryptocurrency, the cost to subvert the truth is pretty high, so the value of doing so, although high, doesn't exceed the cost. With anything for which the value isn't inherent, like a cryptocurrency, it's hard to see that working out.

I mean, if I'm using a hypothetical art-ownership blockchain system, and I see that someone has spent $5 to claim authorship of Mona Lisa, do I care enough to spend $6 to say no, it's Leonardo? Do I care enough to spend even $1, along with five other people? What if 200 4chan trolls have each pitched in a dollar to say Hitler painted it? Eventually, maybe, enough people kick in a combined $201 to say no, seriously, it was Leonardo. But then what? Am I expected to continue paying a regular fee to maintain the truth? Why would I even care that much?

Wikipedia is a great example of something working really well, but many of the reasons it works so well are contrary to what you've described. It is in fact often people with far more time than money that ensure the truth generally remains on wikipedia, and they use non-blockchain-like tools, including the exertion of authority and locking of article pages, to do it.

I'm not trying to crap on your super-sekret whitepaper at all! I'm mostly thinking this through as I type, and responding to the poor design, heck, the poor goal of the described art-authorship application. You might have solved all of this in a way that doesn't depend on trolls being poor or getting bored easily. I hope so.

> You might have solved all of this in a way that doesn't depend on trolls being poor or getting bored easily.

I would argue that most markets in existence are fundamentally based on that holding true. And somehow the world, for some reason inexplicable to me hasn't collapsed yet.

> Wikipedia is a great example of something working really well

In the general case, yeah I agree. However, there have also been cases where authority on Wikipedia has been blatantly misused to overrepresent personal views of privileged contributors and censor contradicting contributions. I'm not saying that permissioned information/knowledge gathering platforms like Wikipedia have no right to exist - as with any centralized system they probably have efficiency gains - but I think there are some properties of it that can be improved upon.

Most markets in existence are in meatspace, and trolls in meatspace risk life and limb. Online, the costs of trolling are much lower, which is why trolling threatens to destroy everything good on the internet.

If your solution is counting on "some reason inexplicable" that took me literally two sentences to easily and accurately explain, I'd worry.

If you think trolling hasn't caused collapses, well, I note that your site has four pictures of people who are unlikely to experience what I'm talking about.

Wikipedia definitely has issues, but most of those issues involve exactly the sorts of things we're talking about here: trust, and who has it. Most of their issues would be solved with more centralized control, not less, and still the issue of who to trust with that control remains.

> Most markets in existence are in meatspace, and trolls in meatspace risk life and limb. Online, the costs of trolling are much lower, which is why trolling threatens to destroy everything good on the internet.

Yes, since traditionally the cost of trolling is ~the cost of the time of the troll. Do you think the internet would look the same if there would be enforceable fines on slurs in Tweets or Youtube comments? More relatedly, have trolls managed to destroy Bitcoin and Ethereum, which are based exactly on those anti-trolling mechanics of rational players being in place?

Thanks for answering! I am very interested in prospect Blockchain solutions.

> the core of it is that all participants in the network are incentivized to only submit true information

How is this done? What is the incentive?

Yeah, these seem like impossible hurdles without a centralized identity system to fall back on. The good news is that we have several: FB, Twitter, etc. These centralized networks already pour tons of resources into identity verification- IE, the 'verified' tag on Twitter.

The combined reputation from many of these services could act as a badge of trustworthiness on the blockchain. You just need a system like Keybase to cryptographically associate accounts to a public key.

I have been shocked to see so few crypto solutions integrate with Keybase as a trustworthy identity network. I wonder why people aren't doing this?

> as you can see with e.g. Amazon reviews and fake news.

If Amazon wants to implement a stricter review check you can rest assured that it will be like YouTube demonitization all over again. So companies err on the safer side and take actions only when things get out of hand. Sure it is not perfect but it is better than system which leaves less leeway.

I read your below reply about how blockchain is required but you lost me at game theory, which frankly has become a go to explanation for everything blockchain. Why will this work in a blockchain? Is always followed by "because game theory and prisoners dilemma etc."

Still given that you are implementing such a system, I am curious what is your background and experience in game theory, maths, voting etc?

> If Amazon wants to implement a stricter review check you can rest assured that it will be like YouTube demonitization all over again.

Why would it? No one makes money from legitimately written Amazon reviews.

Because people rely on reviews and ratings to buy stuff? And stricter guidelines will mean lot of real reviews will be deleted along with fake ones? In which case some of the genuine items will be deprioritized because they have less reviews/rating?

> Because people rely on reviews and ratings to buy stuff?

There are people who try to make a living on YouTube. Demonetization hurt them and they have a ready-made audience to vent to.

But legitimate reviewers won't really care if Amazon implements stricter checks, because their livelihood doesn't rely on Amazon reviews. People won't have much sympathy for the people pushing crappy products using sketchy reviews who might get hurt, and they don't really have a high-profile forum to complain.

If Amazon wants to implement a stricter review check you can rest assured that it will not be like YouTube demonitization all over again.

Draft of our whitepaper is out now: https://medium.com/rlay-official/introducing-rlay-a-decentra...

your website has no details and you don't publish your whitepaper.

> Over the next few weeks, we will be publishing both code for our first testnet and our whitepaper

If you can be patient for a bit there will be something to see. If you want to you can also shoot me an email, and I'll send you the whitepaper ASAP.

Would you agree that this is a harder problem than say casino and gambling on the blockchain? There are great use cases for improving trust with smart contracts and ledgers that prevent the house from cheating or running off with the money.

Not sure what specifically you mean. Doesn't gambling on the blockchain usually work without a "house", e.g. smart contracts that are not in anyone's control? Those are the only gambling systems I've seen so far on the blockchain, but I must admit that I've never deeply looked at that topic.

The smart contracts have to be in someone's control. It's better if they aren't controlled by the house though. Check out FunFair.

Ah, the FunFair example makes it more clear what you mean. I think the main obstacle there is finding/creating a decentralized way for (pseudo-)random numbers to establish true fairness. From what I've heard that problem has mostly been solved, but I don't know for certain.

> The smart contracts have to be in someone's control.

They don't _have_ to be. Someone has to deploy the smart contract, but that doesn't necessarily mean that they get any privileged access to alter/use the smart contract after deployment. That's where a lot of the beauty of smart contract enabled blockchains comes from in my opinion.

There are various solutions for randomness. Funfair's is for the user and the house to generate chains of hashes, e.g. hash(hash(hash(secret))). Each posts the outermost hash to the smart contract. Then each can reveal successive preimages, and the contract can XOR each pair to get a random number.

Here's an article on an idea very similar to theirs, with sample code: https://www.blunderingcode.com/gamble-channels-fast-verifiab...