Here is an model for commerce on ZeroNet which involves three types of players: consumers, sellers, and trust brokers. First, let us look at the role of trust brokers. Each trust broker offers a combination of arbitration and ratings services. For a competitive fee, a trust broker will list a seller as "approved" on their site along with some ratings information if the seller meets whatever criteria the broker defines. The ratings information might be a combination of consumer reviews, number of times arbitration was involved (per month, for example), and possibly even the result of some investigation into the quality/security of operations.
The trust broker is also a certificate signer. The arrangement with consumers might be something like: if you maintain a good reputation with our sellers and you fill out our satisfaction surveys, then the service is free (but if you do not fill out the surveys in a timely fashion, there is a fee). In fact, the fee might be proportional to how much of a detailed survey the consumer wishes to skip.
Now, consumers are paid for their feedback with an endorsement by the trust broker and the vetting efforts of the trust broker are paid for by the seller. Sellers would only sell to approved (i.e. those with a valid certificate) consumers and consumers would only buy from trusted sellers.
In this arrangement, it is advantageous that pseudonyms are used since there is currently no fully trustless way to transact goods other than information. Also note that there is a response to the objection that trust-brokers would rubber-stamp their paying sellers. Indeed, it is easy to detect a cheating trust broker since other trust brokers would rate the seller (or consumer) worse. Then there is a huge incentive to discredit the cheating trust broker and steal their customers. Plus, on ZeroNet, it is easy to keep tabs on competitors ratings and compare them.
ZeroNet offers an incredible innovation: cloning. With easy to deploy, clonable site templates for stores and brokerships, the barriers to entry into the market are incredibly low. So we can expect the provision of trust brokering to become pareto optimal within a short time. Furthermore, once there is a competitive market for trust brokers, sellers (and consumers) would likely wish to gain multiple endorsements.
Also, with the advent of merger sites, we can imagine the emergence of meta-stores which organize the contents of many individual sellers in an easy to use way and includes their scores. The cost of generating such meta-stores is likely to be around zero, especially if clonable ones emerge.
What is needed now is the following:
Easy, clonable stores.
It needs to be easy to set up such a store without doing any coding, like a new ZeroBlog. This means there must be a way to add new products, manage the list of trust brokers, view and clear pending sales. There also needs to some way to keep sales private. The nature of the transaction can be hidden in ZeroNet, but not the fact that it took place. (Correct me if I'm wrong.) One way to accomplish this is to handle transactions outside of ZeroNet. Another way is to transmit transactions to a trust broker. Then logic on the seller's site could scan the transaction logs at their trust brokers, looking for new transactions. This would not anonymize the user, but would hide the store attached to a given transaction.
Easy, clonable trust brokerships.
Like the easy, clonable stores, model trust brokerships need to be configurable after cloning without going into the code. There needs to be a way to approve users as consumers and to add and remove sellers and monitor their status regarding fees. There also has to be baked in logic to get new transactions and generate surveys for the customer and the seller. In fact, this makes the second alternative above for handling transactions seem more reasonable.