This. All of these ideas skim over the most important part, getting fiat into the system. That requires trust, and that always leads back to a central point. Who do you give $10k to in a decentralized system? Unless you meet them in person and bring your baseball bat and some friends, no one. Our TRUSTLESS bitcoin network relies on TRUST based networks to give it value, so much for independence.

That's how innovation works. There is no way around it. It will suck in the beginning. People will find flaws in current systems, then improve on them, version after version, in smaller and bigger bursts.

The constraint area hasn't been mapped yet, let alone that we can know how an "optimal solution" would work. A "design competition", especially with so many non-technical people, is not going to solve that. Who is a judge of what is going to work anyway?

I am not saying that peer review is not good. It's good. That's why I posted links to current projects and collected some information. Why don't you peer-review bitcoinx and darkxchange? Criticize their mathematics?

Why should we create another alt-coin? I suggest we use existing bitcoins to reward the exchange-miners.

Tradecoin isn't a bad idea. Let tradecoin be part of the contest. Any idea should be considered and voted on.
Tradecoin seems to be a high level abstraction layer protocol similar to devcoin and the like. The problem is if you can't buy coins at all due to the exchanges being centralized and hacked what good is tradecoin?

Enough words about this. I look forward to your perfect P2P exchange that runs without flaws from day one, has all the features people could want from day one, and won't have any scaling issues. In the meantime I'm going to look if I can concretely support some project. Sayonera.

It might be immune to DDOS because it's decentralized but then we still have to worry about timing attacks, possibly bots with algorithms designed to game the system. It's distributed but then would it be vulnerable to botnets now that it's distributed? How can we verify each node in the network isn't a zombie node or part of a dark pool to manipulate the network and how would we detect certain patterns or at least design the system to be modular enough that the protocol can continually update it's defenses via plugins or extensions?

I fear if anything is thrown together without much thought about design that the hackers could find flaws in the distributed design. The three main concerns are establishing trust, detecting malicious trade behavior or patterns and limiting the damage, and knowing for certain that every node is who they say they are.

Digital signatures via GPG/PGP can allow for a web of trust and can allow you to know each node is who they say they are. A system where I vouch for you, someone vouched for me, and we all are continuously reviewed can produce a web of trust (just off the top of my head). It still doesn't prevent a team of hackers from conducting malicious trade behavior and I'm not an expert on trading platforms of the possible attacks against them.

I just know when we looked at what happened with mtGox we saw all kinds of unexplained trading behavior that didn't look just like a panic sell but looked algorithmic and coordinated. It didn't look very natural, and I don't know enough to know what natural trading behavior looks like. This is why we would at least need someone in here who is an experienced trader who has seen all the attacks that happen in Forex or Wallstreet and how those attacks have been defended against. The protocol would have to be modular enough that as new attacks are discovered a plugin can be written to manage or detect it similar to how Firefox and Chrome have extensions and the like. Scaling is important because if mtGox really did crash because of too many users and if Bitinstant really is forgetting to pay us our coins because of too much load then that scale and load is still going to be a problem if you decentralize it.

If you're getting into GPG auth ect, we already have this kind of decentralized exchange it's called bitcoin-otc

GnuNet is the only candidate for such an endeavor and it still is prone to different kinds of ddos they're working out. For instance you can flood a peer til they disconnect and then impersonate them. Not good for an exchange unless you had cryptosigning, and if you're going to go that route why not just use IRC what's the difference. If you're competent enough to do multisig and pgp signing then you can use bitcoin-otc.

It would be trivial to write an application that plugs into #bitcoin-otc and allows you to authenticate with the bot easily instead of reinventing the wheel. The app can list the order book and check for gpg identify or make gpg identify easy for total idiots

Or, run Open Transactions inside Gnunet. Great success

An OTC client on top of RetroShare (http://retroshare.sourceforge.net/) would be another option... It's very user friendly as far as these tools go. And there would be sharing files, communication and trade in one application. It already has the web of trust, and already uses PGP just like OTC (but the user can remain oblivious of this). "Just" add in the reputation system and a way to post trades...

This sounds like it could work. I also think retroshare has potential. At least we can borrow code from it and solutions if those solutions are proven to work.