As another user said there is also the problem of scalability, for little items it is fine to make an escrow of a few dollars, but not all the buyers will have 2 times the money of what they are buying when we speaks of thousands $, for a company it will be also a hell for accounting, i don't see small companies having enough money to adopt the system either...
My point is that it will be hard to put the system mainstream in these conditions for sure.
Not that I'm suggesting that BitBay will become SR3 or anything like that... but another possibility might be that a seller is selling something that he'd be perfectly fine making substantial escrow deposits simply because they want the transaction to be completely anonymous and will likely transact business with a buyer who is of a suspicious nature. The reason the seller is fine with the escrow deposits is because they item they're selling actually costs almost nothing... despite having funds held during the escrow period - the funds themselves are all that is risked.
The other aspect that seems to be missed with many of these scenarios is that of time - which can be managed via contract as well. If you (in unusualfacts30's example) are selling $10K worth of headphones - unless you are selling the whole lot at once - you may require more or you may require much less for the deposits. For example say that $10K represents 2000 headphones... so your material risk per contract is $5 and you are charging $20 for them. If you sold all of them in 2 weeks it might require as much as $40K in deposts!!! However, on the other hand, if you average 10 sales per week... then you will only have $400 tied up at any given time... with returned deposits available for new sales regularly.
On the other hand you stand to make $30K in profits... for which you will not have to worry about charge-backs, deadbeat buyers, etc. Sure you might have a couple losers in the group... and say that costs you $50 in material losses and $200 in deposit losses (assuming a 5% fraud rate). You will still have paid less in those losses than you would have in merchant fees and other fraud. Now you have $39.95K to buy more headphones... and you got $39.8K of your deposits back... (assuming worst-case scenario) so on the next round you are only risking profits from the last container of headphones (well $50 extra if you have another 10 deadbeats... but we'll call that a 'rounding error').
Not saying this is important to all the 'honest vendors' in the world - but there's also no centralized organization reporting to your government what your actual sales numbers look like at all. So depending on what your tax laws and/or personal ethics dictate... it could be tax-free as well as fee-free and virtually risk-free on top of that.
Edit: fixed math fart
Addendum: In actuality I see far less problem with scalability of sales from a B2B standpoint than from a direct-sales standpoint. It could be that the model will indeed prove too cumbersome for someone wanting to sell their car for quick cash... but it could be argued that those transactions are better handled elsewhere regardless. Direct sales would require a large capital outlay on both sides... but this is commensurate with higher risk factors. On the other hand if I am the manufacturer of those headphones and I'm selling a container of them for $10K... then I'm only interested in buyers for whom $20K is a reasonable deposit amount in order to guarantee the transaction completes.
There's a reason that Best Buy can charge $25 for a cable they buy in lots of 100K units and pay less than $.50 ea for... because no one shopping at Best Buy is interested in buying 100K cables, and because Best Buy has to offset the inventory/shrinkage/etc... on those units. On the other hand, the company that made those 100K cables... spent less than $.25 on them - but they're fine only making 100% profit and not the 5,000% BB makes because they have almost no risk, limited inventory demands, and a single transaction to deal with.