Topic: A decentralized exchange! (Read 4461 times)

There are various ways of working around this, though not all choices would be suitable for quick trading. One would be for each escrow agent to potentially offer accounts and hold money. Even though they then act more centralized they still settle and share trade info through the network. Another would be to have some way to verify funds transfer and if not within some time frame then cancel the trade. Obviously only trades completed are treated as "price movers". Many aspects of this would need to be figured out and I doubt it could handle trading as fast as an independent exchange. A decentralized network will never handle trading as fast as a single server processing SQL transactions.

Adding an escrow option to OTC would indeed be a good idea, but it doesn't really solve any of the bigger issues around it. In your system the buyer essentially has a buyer's remorse option for about half a day before the bank transfer is final. He/she would just cancel the transfer in case the price development during those hours is not satisfactory, while the seller has the coins locked up in escrow. It is a system that is good when price is long term stable but it is very far from a trading platform.

The problem of a decentralized exchange could be reduced the the establishment of an escrow and trade recording network.

The trade messaging and recording could be done with a modified copy of Bitcoin where the escrow agents use their keys to sign messages about trades. Call it BitTrade.

A trade would involve the bitcoin seller sending coins into escrow (via Bitcoin) until the fiat is paid directly to the seller. The recording of settlement is handled by the escrow agent who then signs messages for each stage onto the BitTrade network which are recorded into it's chain.

Stages might be: Offer (Bid/Ask), Accept Escrow, Complete, Cancelled. So those messages with qty/price and other details would get encoded into the chain. The escrow agent signs them to validate each stage is genuine and builds up a trust level based on past transaction count.

With escrow signing built into Bitcoin this type of escrow can be handled easily and the only step that needs to occur in the "real"world is transfer of fiat to the seller. This could be handled in numerous ways from meeting in person, postal mail order, WU, ACH transfer, or bank transfer.

A BitTrade client would view the BitTrade blockchain and always be able to present a market view based on activity globally at any moment. Info would consist of trades initiated, in escrow and completed or cancelled. Some nice charts could be drawn up too.

This could be seen as an automated version of bitcoin-otc with it's own validating recording blockchain. Outsiders need not even know what any money transferred was for. Could be anything much like buying/selling on eBay could be anything.

That's where we disagree. I think trading mtgoxusd-vouchers have obvious trust issues and that that is the reason they are not used more, and not that traders are somehow unwilling to use IRC/PGP. There are issues around divisibility, MtGox' availability, and the sheer paper work required to get bank transfers going (not to mention the two week waiting list) that needs to work a lot smoother before the user interface turns out to be the primary problem.

various fiat gateways around the world would release their own vouchers/coin pegged to fiat who had otr ratings and were trusted enough to use, then in the exchange users would just p2p trade between themselves through escrow nodes/bots run by anybody with trusted rating, or between users who trust each other enough to trade and are gpg identified like what happens everyday on #bitcoin-otc.

mtgoxusd/vouchx is the same thing basically. if you had one of these vouchers or exchanger coins you could cash directly from whoever issued it or trade it p2p with somebody else looking to buy in and get coins. ripple sort of works this way too, except it's closed source .

now you can write a bot and have it buy/sell automatically for you across the network using escrow bots. jack up the anonymity level on it (choosing your level of anonymity is actually a gnunet feature), configure stop loss, and it could trade safely since nobody could find it. it would only be identified by it's gpg authentication handle/ratings and resolved GADS name through dht, which other traders could look up, which hopefully had some sort of contact info if there was a problem. because bitcoin does multisig cold wallets you could run a separate security bot to watch and verify trades with a second signature to move the coins, either automatically or it sends a signal to user to manually initiate trade if they are satisfied.

on the users end they would just enter a price/market/limit and click sell and bot would do it's thing until it found a match based on pre defined trust levels and prompt you if you wanted to proceed or configure it to auto trade and let it go for awhile trying to find the best price.

That's a promising idea. I like it, but I am skeptical on how you could possibly bootstrap a coin pegged to USD. It would not be an altcoin based on a blockchain, just signed by an authority who guaratees the value. There are some obvious trust issues here.

But I'm all for it in principle, as it would empower the decentralised exchanges. This is the core problem to solve. Focus on this, not on how to build yet another useless peer-to-peer web framework. A thousand of those would bloom if there was a medium of fiat exchange that could work decentralised without the risk of chargebacks etc.

A bitcoin miner driven darknet network solves this problem.

Ripple is unfortunately more complicated than bitcoin, namecoin, litecoin and ppcoin together.
And even bitcoin is to complicated for the most people than how should they use ripple ?
We need a more simple and understandable solution. Like Liberty Reserve for EUR and USD but decentralized and based on bitcoin technology.
Lets say Cryptodollar and Cryptoeuro. 1 cryptodollar= 1 dollar, 1 cryptoeuro= 1 euro.
Then you can change bitcoin for cryptodollar by a cryptochange like bitparking was. This type of exchanges are less regulated than fiat exchanges because all are virtual.

You could actually make a pretty decent p2p IRC group of servers over gnunet that would be resistant to ddos with pseudo anonymous communications resistant to timing analysis which is Gnunet's strength. You can also run Tor IRC servers and Freenet/i2p servers easily.

IRC is the solution. It already exists, and is being used to bootstrap most p2p node propagation.
http://ieeexplore.ieee.org/xpl/articleDetails.jsp?reload=true&arnumber=5159226

paywall but there's copies around

Users don't use reg IRC clients they use a front end for it that automates everything including identifying with gribble (the bot(s) that would run on the decentralized network, not flooding the existing bitcoin-otc channel) and have a chat or trollbox they could use

If the brokers use the OTC the end users dont need to even know about OTC. They will trade happily and easily in the frontend as they do today without knowing whats happening in the backend.

Yes, OTC is the obvious starting point. It's the only working decentralized exchange at this point. But I the current interface with IRC and PGP is not very usable for most people. Is there a project for a more friendly frontend? (I mean software, not so much a website, as that's again a single point of failure...) Also, though it is peer-to-peer, it currently completely leans on freenode maintaining the channel isn't it? And the gribble bot?

I have described a complete exchange-broker-client system in github based on what Ive read in bitcointalk, and own experience from trading stocks/futures.

Im sure the best model is to have the exchange 100% decentralized and let the brokers work as they do today, having their own website and their own system handling clients funds. but instead of working separately they must start to collaberate and trust eachother and settle dollars & bitcoins regulary.

https://github.com/p2p/bitcoin-exchange

You would create your own altcoin called Derp's USD coin and peg it to USD. Then you are now the gateway to cash fiat out of the network. They pay you to get Derpcoins and you hold money until somebody cash's out. People trade your altcoins against bitcoin and it acts as a voucher but secure. You would obviously mine these yourself. Could also use Open Transactions to do this. Whoever wanted to trade Derpcoin with you then could.

Problem with this is it's highly illegal with bullshit FinCen rules

Sure, but what would you be exchanging _against_?

Perhaps USD, but paid how? Bank transfers between peers? Those take a day, during which the buyer can simply cancel if the bitcoins drop in value. So that's skewed to buyers. Credit cards? Good luck processing those. Paypal? Chargebacks.

There is already a decentralised exchange. It's called OTC. It has all these problems. If you want to reimplement it, at least look into solving them.

Great!

I think one of the issues that needs to be addressed with the current projects is that they're unapproachable by normal users (and even normal developers). Bitcoin would likely not have lifted off if Satoshi hadn't made a simple Windows GUI for bitcoin.

Also, for example the opentransactions description and features list is kind of difficult to parse if you're not already into financial cryptography.

I believe currency substitutes are foreign currency specifically. There are many domestic alternative currencies that aren't regulated.

A PM dealer issuing his own funny non-negotiable scrip to a consenting party is a-ok unless the sale of gold and silver for cash is being regulated again.

Ive listed them all in github

I forgot that one, I'll add it to my list! Though it's not completely decentralized, you don't need to trust any single server or company, which is already better than Mtgox etc.

Open Transactions when it's ready will be good enough. If enough federated servers can just create our own currencies and act as exchanges for the fiat in and out of the system. These can be run over Gnunet and come with the bonus of crypto verification instead of relying on GnuNet DHT which is unreliable and possibility nodes can be faked (like faking an escrow node to scam..uhoh).