Ripple anyone?

The difficult problem is the one involving fiat. Matching buyers and sellers is trivial at the volumes mtgox handles, even on days when trading spikes.

What is NOT trivial is the broker aspect. Handling money, interfacing with banks, paying out money (a large part of the problem), and so on. Know Your Customer regulations and similar only exacerbate the problem.

It's also critical to remember that trades must be QUICK. Sending cash or gold is not a valid solution.

One main reason an "exchange" exists in a one place is: speed.
If a buy-sell takes more than few seconds (or milliseconds), why bother with the concept of exchange at all?

On secondary points: Think about why an exchange exists: You trust your Bitcoins to the exchange. A p2p exchange is founded on mistrust. While it works for Bitcoin as an absolute currency, mistrust kills it for an exchange.

The broker aspect is important but its not the most crucial part. The most crucial part is the exchange, that is essential.

But could a peer to peer exchange really function given the volume and flow of transactions going on? Plus this volume is only going to increase.  Could it keep up in real time? Surely there are going to be lags.

With a system to track credit obligations between parties, and a reporting system for settlements between those parties, who needs a broker?

The centralized inferstructure surrounding Bitcoin is fundamentally broken.  And NOW is the time to fix it.  

Here's how...

Here's what needs to happen for Bitcoin to be massively adopted and not killed by its own exponential growth:

- decentralized, scalable exchanges.

- stupid simple, one-click LOCAL wallet solutions that a newbie could not mess up if they tried.

- decentralized encrypted cloud back of local wallets.

- decentralized Bitpay-like solutions.

No central anything.  No central wallets.  No central exchanges. No central payment intermediaries.  No central wallet backup locations.  Nothing central period.

Here's a solution to all these problems:

A decentralized Bitcoin user/minor driven network...

...A darknet that's fully powered by the Bitcoin users and miners.  Simply take existing darknet code and retrofit miners to it.

On top of this network we build a decentralized exchange that quickly matches buyers to sellers.

In addition to that we build a decentralized cloud storage solution.  Using this cloud storage, local wallets are backed up across the network automatically.  Along with every bitcoin client is an encrypted backup of the wallet in the network.  No more novices losing their coins.

All users get enough cloud storage to store their wallet.

Web services could also be purchased on the Bitcoin network too.  The fees would go to the Bitcoin miners powering the network.  The fees would be based on a market bid/ask.  Effectively allowing a fast, fully decentralized tor-like network with no central point of attack.

This could be dovetailed in nicely with the existing Namecoin DNA system.

Users could also purchase additional Bitcoin cloud storage based on the market driven by the miners/users.

An encrypted email-like system could also easily be built on top of the network that would allow fast communications that can't be shutdown by an Internet kill switch.

Who knows what else could be build on the platform and offered as services.

These additional services would encourage more mining and diversify the network.

Miners/users get the fees....  The free market governs the price.... No central authority possible.... Everyone wins.

That would be a very robust network that would solve most the Bitcoin centralized inferstructure problems in one fell swoop.... And create an entirely new Internet that can't be snooped or shut down.

...and would make Bitcoin completely independent of the Internet.  All transactions could be roughted through the Bitcoin network.

We already have more then enough hashing power.  There is already open source darknet code to build on .  The groundwork has been laid.  

...Now we just gotta create it!

Create it and finally we are free of central attacks, central control, and Bitcoin can scale.... And no longer beholden to any central authority.  

...and idea who's time has come.

Techies, tell me...

...the viability of the idea...
...any improvements to the idea...
...the obstacles that need to be anticipated and overcome...
...and the next step to make it happen.

No trolls reply.  No non-intelligent replies.  Only those who know what the hell they are talking about chime in here.

A de-centralised exchange cannot work.

When you buy BTC from Fiat you need to send the money somewhere, claerly this cannot be a different person in a different country everytime you buy any BTC, this is not practical on far too many levels and is why BTC exists in the first place!

When you sell BTC you require to receive cash. Same as above applies only this time you are relying on an individual sending the money having received a coin that cannot be traced = 'Licence to Scam'

Exchanges exist to pool liquidity and also to ensure that settlement risks are borne by a central company, usually with some financial depth and credibility. This is the part that is lacking currently.

This would be way easier if you just created "Darkcoin" and allow users to choose whether they want BTC or DKC. (Hmmm... Maybe that's a bad name)

Then you don't need to deal with problems like a seller being matched up with 30 partial orders and needing to figure out how to pay them all through fiat infrastructure.