I predict a lot of strain on the Bitcoin network soon due to Mastercoin

[coinrevo]

d'aniel. you need clearing. that's hard to do. mastercoin and others don't do clearing. you can't really transfer goods this way. say I want to sell you 1 ounce of gold. how do you make sure you know I have it, and they you are going to be in possession after the transaction? somebody is going to safely store the gold in a vault. then I have to physically hand it over to you, or someone you trust. also you want to make sure that nothing goes wrong during the transaction. in bank-less societies this is part of the counterparty risk.

frankly, not even the largest commodities exchange in the world, CME, has a perfect answer, as they are e.g. gaming their silver holdings. and some of the largest central banks in the world are lying about their gold holdings. this shows you its a highly non trivial matter, although it might seem easy.

with synthetic markets (swaps, prediction markets), you have the problem you don't have really natural supply & demand, as goods never change hands. its a betting market on real exchanges. what we need is a real exchange of goods and services, not gambling markets (which have their use cases).

[d'aniel]

You actually don't need any kind of clearing for the predictions market I mentioned; it can all be done with Bitcoin scripts, and one or more reliable oracles who don't even need to be active participants in transactions.  Here's a nice explanation: https://bitcointalk.org/index.php?topic=260898.msg2804469#msg2804469

Sure, a decentralized predictions market would rely on external infrastructure for actual price discovery of assets, but being able to gain financial exposure to these without needing verified accounts on centralized exchanges seems useful to me.

[coinrevo]

I know you don't clearing for swaps. but swaps doesn't produce much meaningful economic activity and there all kinds of problem with them. if prices move too quickly there can be negative overhang which breaks the system (margin calls). to prevent this you can do these as binary options. btcoracle does this, and guess what - there system is being gamed, and the price you get are horrible.

for "oracles" you need to trust servers anyway, so this is kind of pointless (and completely inefficient at scale). there is no such thing currently, and to make them work you need a lot of things which haven't been invented yet. too much talk, too little code that actually works.

[d'aniel]

The oracle scheme I linked to is "m-of-n" compatible, and has O(1) scaling with the number of participants in a given bet.  It's just simple publishing of secrets and their hashes.  Furthermore, you could build a transaction based on this today using the rawtransactions API and Eligius to get it mined (it's currently non-standard).

[coinrevo]

1. all that an exchange does is: a) establish a price by aggregating orders in a continuous auction b) exchange goods after orders are matched. what Peter is suggesting doesn't solve either of those problems. he has a theory, but not the code. same as all the other attempts in this direction (mastercoin, bitshares, ethereum, OT). otherwise, prove me wrong: I bet 1 BTC that BTC will close under 1000$ this month on bitstamp at 1.8x. any bids? see, exchanges without price discovery are not exchanges.

2. without external inputs smart contracts are worthless. you can't run this on p2p nodes.

3. this is not O(1). order matching is time-dependent.

again, there are some very basic concepts missing which are obvious to people who have worked in the field. a good start is a good book on exchanges, i.e. auction systems. usually it also takes a couple of years to acquire the necessary concepts in practical microstructure of auction, to understand the issues such as latency, reference prices, etc. etc.

I can tell you what P2P scheme works for certain asset classes, but its simply impossible to do this for liquid public assets as of now.

[d'aniel]

All I was focusing on is a way to enable people to make bets with strangers with minimal trust.  After that, people can find each other via whatever channels they like.

But I agree with you that a decentralized exchange will be terrible for efficient high volume trading and price discovery - it would do best to rely on centralized transaction servers for this.  There I think the focus for improvement should be low-trust operation (e.g. Open-Transactions (I thought it did have functioning exchanges?  Haven't checked personally.), and MPC*).

* http://fc09.ifca.ai/papers/15_Secure_MPC_goes_live.pdf