SR cut the shit imo it is pretty obvious you are just looking for more money. Which is perfectly fine, running a site like SR is a full time job and it isn't worth it to do it for free. But you are going to do damage to the site by implementing this method of getting money. Because nobody is going to do bulk deals through the escrow system. I have seen deals for 10 grams of crystal LSD, hundreds of thousands of tablets, kilos of MDMA etc from legit vendors online, even on fairly well known drug forums. There is no way in fucking hell the people doing deals like that are going to risk losing to a reverse scammer. The risk has historically been on the customer and even though escrow is fine for small deals and really a good model, it ensures that there will never be bulk sellers. And many sellers with bulk don't want to be bothered changing to small orders either. I suggest you tax big vendors on their orders, but don't make them risk losing out to reverse scammers. BTW there is an open source bitcoin mix software being developed that is pretty cool, it decentralizes trust by requiring multiple nodes to sign off on the block chain (this is a little known feature built into the bitcoin spec, but that hasn't afaik been taken advantage of yet). This will allow for decentralized bitcoin mixes where no single mix operator can steal bitcoins from a client who uses it. It also uses blind signature mixing so protects from evil mix operators colluding together to deanonymize coin flows. It allows for mix nodes to charge a price for their services too. I think it will quickly turn into a commercial network of mixes when it is done and really be a great way to get truly anonymized bitcoins. But anyway the reason I bring this up is that apparently escrow can be decentralized and still in line with the bitcoin spec, because this software (Open Transactions...look it up) decentralizes trust with bitcoins in a mixing context. Soon I think people will automatically store their coins constantly mixing through a distributed trust network of mix nodes / addresses. You know what you really should work on though, SR? Implementing more systems to protect from the feds doing name flooding attacks. They already have the software to aid in name flooding attacks, where small teams of agents aim to control many personas online to maximize human inteligence 'surface area' in pseudonymous networked environments. This is scary. The feds could easily mass register accounts and I am sure that they are doing this. Charging for seller accounts at least puts a price tag on the attack instead of making it essentially free to engage in. However, the feds have a lot of money to spend on intelligence sources so this isn't perfect protection by any means. I think you should go with somewhat of a vendor recruitment strategy. You don't need a trust rank system like OVDB although a decentralized web of trust would be nice. But what I have been thinking of recently is using a blind signature system to verify the presence of people on a public drug discussion forum prior to some date in time. The logic with this is that public drug discussion forums have very large amounts of registered users with many years of history, and I think that the feds only recently started fucking with persona management software and name flooding attacks. The chances that the DEA has a lot of registered nyms on bluelight that registered in 2011 is a lot higher than the probability that they have a lot of bluelight accounts from 2009, although they probably do have some I don't think they mass registered that early. Some sort of system that can verify a persons account on bluelight while giving them anonymous credentials would be nice. The verification can be done automatically by having the user submit the hash of a message they havn't made yet to the verification server and then giving the verification server the message hash and link it to the thread it will be put into. Since only the person with the account will know what they are going to post before they post it, the verification server can verify that they own the account in question. Then the server can issue them a blind signature certificate that can be redeemed for a 'verified' credential on SR. Only allow people who were on public drug forums from some point in time a prior to get verified status. Now a verified user can prove that they were a member on a public drug discussion forum at some point in time, but they will still have a large crowd size to blend into, particularly if enough people use the system. This means their verified name wont be linkable to their name on the public drug discussion forum. The main reason you want to use blind signatures like this to create unlinkability and a large crowd size is to protect from having IP address records from public forums used to deanonymize users. If this were done, verified users would be in the crowd of everyone who obtained a blind signature (to the verification server) or the crowd of all public drug forum users (to any other potential attacker). Of course being verified status would narrow you down to someone on one of the forums that can get verified status in this way, but people on those forums are already probably targets already, talking about illegal drug use openly I hope they all use Tor but I doubt it. If they all used Tor all the time when using bluelight it would be much easier to recruit and give them verified status lol. Not just bluelight though but also forums like drugs-forum uk etc. Really you could even verify people had nyms on other popular forums in the same way, to increase potential crowd size substantially and probably protect even more from nym flooding attacks (after all I bet DEA has more nyms on bluelight than they do on some random popular forum, on the other hand I bet more silk road users have accounts on bluelight than any other forum too so it goes both ways). I think protecting from human intelligence name flooding attacks in as many ways as possible is the most important thing that needs to be focused on right now, we are on the verge of making huge progess in the communications / computer security / product transfer - interception detection / financial anonymity etc departments but the one place we are always going to be weak is human intelligence. And human intelligence has a lot of potential to do a lot of damage because once an infiltration or compromise takes place ALL of the nodes networked to the malicious one via the postal system that get packages from it are at serious risk of being compromised, and in the mid-distant future this is going to be the only way the feds have to keep intelligence flows on the scene going (especially once undetected interceptions are a thing of the past).