Bitmarkets: Private decentralized marketplaces based on two party escrow
voluntary.net
voluntary.net
If the buyer approves the transaction stuck in escrow, they get their 1x the price back, the seller gets their 1x deposit and 1x payment.
If the seller gives a refund, everybody gets back what they put in.
If neither party does anything, the buyer is out 2x and the seller is out 1x. This encourages the buyer to pay up and the seller to deliver.
This is bolted onto Bitcoin, but it could be bolted onto any payment system - Visa, PayPal/eBay, Venmo, Snapcash, MMORPG in-game currencies, etc. It would probably be more useful on Venmo than on Bitcoin.
It's a service -- if you want it and can't afford it, you don't get it. It doesn't work any differently than anything else you can't afford.
It could be bolted onto any payment system that has scriptable multisig transactions. Visa, Paypal, and friends would not work without a trusted third party escrow.
If neither party does anything, the buyer is out 2x and
the seller is out 1x. This encourages the buyer to pay
up and the seller to deliver.
Wouldn't buyers choose (b) because getting back 1X is better than nothing?
And if you're going to release payments even when you're the victim of fraud, you might as well not bother with the escrow?
I wonder if it is theoretically impossible to create a protocol that punishes cheating when the cheating occurs outside the system (when the physical goods change hands).
And even if it isn't you would also somehow have to consider the fact that things get lost or damaged in the mail and assign blame to one party on the protocol level.
Say I am a malicious seller and I am offering a brand new XBox for 1BTC. You the purchaser start the order put up your two BTC, I put up my 1BTC and send you a used XBox only worth 0.75BTC. Now you as the buyer have two choices. You can either a) Refuse to confirm in which case you'll be out 1.25BTC or b) you can accept your loss confirm the order and be out 0.25BTC. Very few people will chose option (a) to the point that it would almost certainly be profitable for the malicious seller.
The fall back is reputation systems which require centralization to avoid gaming/having enough data points to be useful.
The profitability for the malicious seller is a function of the probability a given buyer in the system will choose to punish the seller or recoup some money.
Also, is there a time limit? For example, what if the seller ships and buyer receives but is lazy / not incentived by the deposit..
The challenge here is this is an attempt to create a "Justice Algorithm".
I think a reputation based system is probably the solution here, and it doesn't need to be centralized. You could have anonymous reputation nodes -- which would become valueable in the system over time the more trust you have built. Trust seems to be the responsibility of the seller to manufacture by delivering goods as promised. It would be less useful for 1 off transactions with individuals.
You'd also want to calculate some trust score based on volume and value and the non-trivial challenge of preventing manufacturing trust via buying / selling to your self.
Also you have the situation with your nodes of then requiring an additional layer of trust on the network to determine which nodes are to be trusted.
You can't even do things like lower the trust ratings of people who trust people who end up being dishonest because you don't know that they were dishonest in their trust of them.
Reputation systems don't work with anonymity. They rarely even work without it.
But most importantly, the idea of making "trust" graphs using a weak form of transitivity needs to die. Reputation networks should encode sign statements of facts, from which trust can be estimated by participants, using a variety of approaches - in particular non-local approaches which aren't fooled by cliques.
The way you avoid clique is by staying within your clique. That is requiring that the entities you extend trust to be trusted by people close to you, who themselves trust each other.
But it's not just cliques. Say you have a system like eBay, based on buyer feedback. eBay can be fooled by playing a long game, and selling a lot of cheap items for a while before selling a big item.
If instead of eBay you have a distributed store of transaction report, you can constantly refine the trust algorithm. For instance, you might pay attention to the price of the items sold, or to the trustworthiness of the reviewers themselves, etc.
The more data goes in, the better your inference. Truth is coherent and entangled. The more data there is, the harder it is to convincingly lie (but the harder it is to remain anonymous as well).
>For instance, you might pay attention to the price of the items sold, or to the trustworthiness of the reviewers themselves, etc.
Ok so how about I just sell a bunch of high value "items" between my own accounts? Now my account has participated in a number of valid high value transactions. Since it is distributed there is no way to ensure the transaction is real.
>trustworthiness of the reviewers themselves
Who build their trustworthiness the same way. This is easy to game look at pagerank for an example of a similar rating system.
>The more data there is, the harder it is to convincingly lie
This isn't actually true though because you don't verify the data in any way(nor can you) so an attacker can craft the input to tell any story they want.
Because as you increase the radius of your search, in a small-world netwok, you reach a very large number of real people while being minimally impacted by Sybil cliques. Check out the work on SybilGuard or SybilInfer for instance.
>Ok so how about I just sell a bunch of high value "items" between my own accounts? Now my account has participated in a number of valid high value transactions. Since it is distributed there is no way to ensure the transaction is real.
Ah, but you're looking for transactions reported by people who are themselves close to your circles of trust.
> Who build their trustworthiness the same way. This is easy to game look at pagerank for an example of a similar rating system.
A random surfer model is a naive model. With better probabilistic model, you become more resilient to gaming.
> This isn't actually true though because you don't verify the data in any way(nor can you) so an attacker can craft the input to tell any story they want.
This isn't arbitrary data, it's data connected to you in some way.
http://www.sitetruth.com/doc/socialisbadforsearch09.pdf
So, anonymous reputation systems don't work.
Full two-party escrow ensures that only these two parties can decide what to do with the funds and no one else could decide for them. You must use Bitcoin to ensure that. If you still have some 3rd guy to decide, you don't need such a setup in the first place.
(Mnemonic: possessive its uses the same number of apostrophes as his and hers and theirs do.)
Many people discuss extortion scenario, so let me reply to all of you. Extortion is limited when both parties are anonymous or have little knowledge about personality of another person. If I try to play dirty with you, how do I know that you are in a desperate position and will submit? Maybe you will turn out a better player in this poker? Or maybe you will simply send me a message "no, please pay as we agreed initially, period" and we won't be able to discuss this longer.
This Nash Equilibrium approach is particularly great for automated agents. E.g. two apps paying each other for measurable service and owners of both apps know perfectly well how another app will behave - there's no one to extort. Either you follow the contract, or you don't. So you can troll someone by simply losing your own funds, all alone in a sad silence.
I think its a great approach. The reputation system and efficient private search which doesn't require copying the entire database are the hard parts!
Think of it this way:
1) Bitmessage ensures that FBI has to catch merchants or buyers individually instead of catching them in one single server like SR. 2) Tor makes it harder for them to locate a party by tracing where message came from.
Tor is not perfect, but Bitmessage is not better, but worse than Tor in terms of tracing the origin of the message (because BM does not use onion routing with layers of encryption).
But when Alice sends a message to Bob (who is an undercover TLA agent), Bob could monitor traffic and statistically figure that certain IP address sends messages addressed to him earlier than other nodes. So he goes and busts the owner of that IP address.
Onion routing attempts to make this discovery harder for Bob because Bob will always receive messages from different IPs that supposedly do not belong to him. They have little value to him because they themselves are connected to relaying nodes, not to initial senders. Bob would have to bust too many IPs before finding the sender, which could be too expensive.
Onion routing begins to fail when Bob has a lot of his own nodes in the Tor network, so for a large enough number of messages he can trace route back to the sender's physical address.
To achieve real anonymity when chatting with strangers (e.g. blackmarket merchants), one needs to use a combination of these factors:
1) Bitmessage or alike to avoid evesdropping.
2) Tor to make it harder for recipient to find location of the sender.
3) Low-latency network to make statistical analysis less efficient. Every relaying node (both Tor and Bitmessage) should delay broadcasting messages randomly.
4) Infrequent communication, so it takes time for recipient to gather data. (This is a variant of #3)
5) Change physical location frequently, randomly and rarely reuse them. E.g. connect from various free wi-fi points in cafes, parks, shops, Apple Stores etc.
6) Never reuse identity between people you communicate with. Merchants must have separate Bitmessage and Bitcoin address per invoice (once item is sold, post another item with different identity). Buyers must use different Bitmessage and Bitcoin address for each purchase. This way amount of information available to an adversary will be strictly limited to just one deal. And that deal will be limited to one unique location and a few exchanged messages that hopefully won't be enough to locate the person. And even if that happens, person couldn't be charged with more than one sin.
If you communicate with people you trust (friends, family members), you only need #1 and that would be enough.