OK, I didn't realize that it's more than the localbitcoins model, but that actually raises even more questions...

If you're going to pitch this as an online alternative to the large exchanges, then you're going to have a HUGE trust issue to overcome. Not your trust, mind you, but the trust that buyers and sellers have with each other, and that's not always easily overcome with a trust/reputation rating. When bitcoin started, almost all transactions were either between people who knew each other or via small online traders that tried to make a reputation for themselves through volume and good service. Eventually, people got burned (or scared) enough by theft that they wanted to start meeting in person to overcome the problem with trusting anonymous folks online, and localbitcoins was born as a method for finding those local people (who had largely used sites like Craigslist to find each other - this still happens quite often, btw.) This was not a perfect setup either, so the market started to demand large exchanges to provide easy, fast, and trustworthy providers as opposed to the small, independant anonymous dealers online, and the major exchanges started popping up, providing exactly that. Today, due to the regulatory crackdowns on the larger exchanges, you're now offering this service as some sort of hybrid that allows buyers and sellers to more easily meet each other, either online or in person, without the overhang of regulation. A good idea on the surface, but that negates the single biggest draws that the major exchanges have: the speed of a transaction, and reliable trustworthiness over many millions of dollars' worth of transactions. It's no different from an individual trust perspective than meeting face to face, but at least with an in-person meeting you have the knowledge that you have a lot greater control over the transaction (you can get up and walk away if you feel uneasy.)

So really what you're proposing is more akin to a specialized Craigslist (or eBay, not really sure how you're going to present the data) to help bitcoin buyers and sellers match up with each other, who can then choose to meet up in person or not depending on their mutual needs. Fine - if you can brand it well enough somehow, you've got a fighting chance at success just from volume, which is what my original post was really getting at. But it isn't a novel solution, really, and reintroduces some of the problems already mitigated by other existing transaction methods.

There is a popular programming mantra, that you can pick exactly two of the following: Good, Fast, or Cheap. You can't have all three - if it's good and fast, it's not cheap; if it's good and cheap, it's not fast, and if it's fast and cheap, it's not good. (Yes, it's variation of a false dichotomy if taken literally, but the general pattern is sound empirically.)

To paraphrase that saying for this situation, pick two of: Trustworthy, Fast, or Best price. If you want trustworthy and fast, as either a buyer or seller, you're not going to get your best price. If you want trustworthy and the best price, it's going to take a while to look through all the listings finding your perfect trading partner, by which time the price may have changed. If you want the best price immediately, you're going to have to take the first person available and roll the dice on their trustworthiness. Additionally, by decentralizing the exchange, you're inherently lowering both the trustworthiness (the sum of trust of all transactions on the site does not imply trustworthiness of the individual with the best price) and the speed at which the transactions can take place. Best price is always fluctuating, so you can sometimes get lucky on that, but you're still limiting the market (compared to what's already out there) in two of three basic criteria.

That brings us back to the idea of localbitcoins and their (largely) face-to-face model as the alternative. The market for face-to-face sales is still quite limited compared to the major exchanges, and it seems from your last post that your primary market will be those users. The only real advantage you appear to have, then, will be the tracking of trust/reputation, which is difficult at best due to subjectiveness. Again, great idea on the surface, but it's not going to be the driving criteria for all transactions.

As for the risk of mugging, I think you missed the point I was making: it's not about the safety of the transaction itself, it's about the choice the buyer or seller has to make regarding the method used for the transaction; again, weighing trustworthiness, price, and speed... for the in-person transaction, the decision on trustworthiness would include trust that you won't get murdered.

I know this was rather long-winded, sorry - but I do have a specific purpose in saying all of this.