Topic: Coinffeine - (P2P bitcoin exchange) code now available! (Read 4005 times)

Great project ..! Looking promising. Reading through the exchange algorithm

This is moving forward...

http://www.coinffeine.com/download.html

The Technical Preview version is here!

Nice to see this project advancing! Good job!

It was an interesting project, but not decentralized and the idea of mutual destroyed collateral suffers from a blackmail risk.

Multiple micro payment is too slow for casual payment, some kind of escrow/credit rating and dispute resolve system is much more convenient than each one managing their risk by themselves (much smaller transaction each time)

Establish many of these escrow points so that escrows can be distributed around the globe, using M of N signature to release the bitcoins. But anyway, when there is a dispute, some kind of manual work is still required. The buyer can simply claim he had already made the fiat transaction but the seller never receive it, the buyer can even provide faked transaction record. How to discourage the buyer to make fraud is the key

NASHX used a model of both party tying up same amount of risk capital as a guarantee before the trade, if something went wrong, the risk capital of both party is destroyed. I found this concept to be very interesting, but it seems their site did not get a large attention:

http://nashx.com/HowItWorks

Nice approach, but I see a few fundamental flaws with that concept:

- The trade time is very slow: if you have 100 rounds it is 100 times the payment processors tx time (with might be a few minutes for OKPay and hours or days for normal bank tx). What is the actual tx time for OKPay?
- To keep that trade time acceptable you are restricted to those payment processors with open API (Okpay,...). There are not many of those available, so it is very limited to a few companies. If the government wants to stop that kind of exchange, they just have to make pressure to a handful companies. How many payment processors like OKPay exist? Can you name them? I assume not more then 10.
- It is very easy to track those tx as coinffeine tx due the clear tx pattern (repeated micro payments).
- It is questionable if the payment processors will support/accept that, as they normally dont take a fee assuming a client average behaviour of a few tx per month. If you make 100s or 1000s of tx a month it produces more costs to the bank as they will earn with that client, so a economic orientated bank will kick out that client. If they take a fee the exchange becomes too expensive.
- In case of higher price volatility (very common in btc) there is an incentive to change the strategy and take the loss of the collateral in account to start a new trade taking the advantage of the new better price. That is more risky the lower the collateral is/the more rounds you use. For 100 rounds it means 1% of price change which can lead to a strategy change. At localBitcoin that kind of "future trading" is a common problem.
- If you reduce the rounds to a lower number and increase the risk value per round than you get a serious blackmail risk. E.g. if you trade 2 btc for 1000 EUR and make 10 rounds you have a risk of 100EUR/0.2 BTC. As there is an asymmetric risk stake (one lose more then the other), a scammer could use that weak point to blackmail the other party with a signed tx where he the change the output to his favor. That way the other peer can decide to accept the blackmail and lose 0.05 btc or not accept it and lose 0.1 btc. A economic rational trader will accept, so game theoretical that will lead to a failure of the exchange. You can only mitigate that problem with lowering the risk stake which means more rounds which comes with the usability problem of longer time and depending on the payment processor with too high overall fees.
- There is no protection against an system attacker (competitor, nation state, banks...) who takes in account financial losses to ruin the trust/usability into the exchange system.
- I assume (not sure about that, have not read all the details) that there is a double spend risk, as the first deposit tx has a timeout the BTC seller could have spent with another tx the inputs makeing the following tx invalid. If the other peer has already done the bank tx he would lose those payments.
- I assume (also not sure about that) that there is a malleability risk as well with the dependency of the payout tx to the pay-in tx. You cannot use reliable a tx id as long that tx is not in the blockchain.
- The case that the other peer cannot continue due unintended problems (accident, disc crash, death,....) cannot be covered by a 2of2 Multisig locked tx.

And my last concern:

- Who is behind that project? You got some funding from investors but there are no details open to the public. What is the business model behind it? An investor wants to earn money some day. If you dont take fees how they should earn money?
The only speculative idea which comes to my mind is that some parties might be interested to get an pseudo decentralized exchange. In such an exchange the information flows over a few payment processors, so it would be very easy to gather the information about all the trades of such an "decentralized" fiat-btc exchange. So it could be used as honeypot attracting people who avoid centralized exchanges due the regulation and control they have about their clients. The promise that the privacy is only leaked to the other trader is not true if there is not a huge variety of banks you can use and if the bank tx is easily trackable. And as stated above normal banks will lead to unusable slow trading times, so it will be limited to those few payment processors.

Looks slow and expensive.  min transaction fee x N?  10 min block time x N?  no way...

NXT seems way better.

Truly amazing if this could work.


Yet, I Wonder one thing..

Bitcoin can ofcourse be traded to another person using p2p but really, hoes does it work With the fiat? Can't you break that Down a bit more, not sufficient info on that yet.

Fees for bitcoin are already low-to-non-existent and you can find payment processors with fairly small fees that have an absolute maximum.
For our first version we are using a model with several steps of the same amount (limited by the initial deposit) but we have an idea of how to use non-uniform steps to minimize fiat fees and number of fiat payments (and therefore fees).

Interesting concept, but it will work only if transaction fees for sending multiple batches of fiat & bitcoin are low enough/ security deposits are large enough to cover big % of actual transactions so each buys/ sell is done in only few steps.

After the bad scam history of exchanges, it would be refreshing to have a nice P2P exchange up

The amount you can buy is proportional to the amount of bitcoin you have. In principle, with very small initial amounts you can initiate a series of exponentially bigger exchanges. You can find those initial amounts from sources like faucets and, the authors of Coinffeine are considering different ways of providing this small amounts.

BTW, I'm one of the developers

I agree. If you are going to an exchange to purchase bitcoins and have no bitcoins yourself. How do you put up the deposit?

How are you going to take electronic dollars? Don't banks have to allow you to run an exchange? That would mean it's regulated.. which doesn't really sound like the bittorrent of exchanges.

The Coinffeine team all just left their regular jobs to work full time on bringing you a distributed Bitcoin-Fiat exchange within the next few months.

They have also raised a first round of seed capital... I encourage you to take a look at their work so far: http://www.coinffeine.com/

And, if I've understood it correctly, I think their exchange algorithm is inherently susceptible to (accidental and/or malicious) failure due to bitcoin transaction malleability.

Their github wiki shows a worked example of the bitcoin transactions http://github.com/Coinffeine/coinffeine/wiki/Exchange-algorithm#transaction-definitions.

It shows that their algorithm is based on creating and signing transactions to spend outputs from other transactions that haven't even been broadcast yet.

So if, for example, Sam's Tx0.1 is subject to a malleability modification it's hash will not be the same as Tx0.3 assumes thus Tx0.3 is useless as a means for Sam to get his (potentially large chunk of) bitcoins back.

If Bob refuses or is no longer around to help Sam create and sign a new version of Tx0.3 referencing Tx0.1's actual, confirmed hash then Sam's bitcoins are permanently locked in Tx0.1.  

@flix I re-read their wiki but couldn't see anywhere that the 'one side has no bitcoin deposit' bootstrapping issue was dealt with...can you give me a more precise pointer?

Where do the bitcoin/fiat safety deposits get stored during the transaction?

Actually, it looks like they have a solution for the bootstrapping issue... see their wiki:  http://www.coinffeine.com/


These guys are advancing.. however it is going slower than I'd like. I'm convinced a p2p exchange is key to Bitcoin's stability, future growth and persistent penetration in countries with unfavourable regulations...

Very interesting concept. Is it correct that deposits will end up as a mining fee in case they are forfeited?