Trust and safety is the most important aspect of any marketplace with the the least amount of innovation.
If you look at eBay, the classic online marketplace, what monumental inventions/protocols (like what RSA did for encryption) have arisen that could be applied to each transaction ensuring they go smoothly without fail? If the item passes inspection, the seller gets the money and the buyer gets the item promptly. If the item does not pass inspection, the transaction can’t take place.
Looking at eBay and other peer-to-peer marketplaces, they give you an approximation for the experience you’ll have with each person by having users leave feedback. This is generally reliable because honest people continue to act honestly and a smaller percentage of dishonest people are filtered out and banned after displaying their bad behavior. The problem is now keeping bad actors out, while validating good actors and adding digital gold stars to their reputation for people to rely on. There are also a few other systems that come into play like PayPal claims, requiring tracking information after purchase, etc. that attempt to shore things up.
We don’t always think about it, but most transactions feel safe to us because of a middleman present. If we take a simplified example of physically depositing money at the bank, why do we feel safe doing that? Why do we trust that the teller is going to put that money into our account and not drop it in their wallet? Because we’re trusting that the bank has systems to keep their workers honest, and even if the teller did steal the money, we would require the bank to reimburse us. We require the bank to act as a good middelman and we (currently) have to put our trust somewhere.
A peer-to-peer marketplace has a completely different feeling because we are the ones taking on the risk (some places attempt to insure buyers, but usually buyers/sellers are interacting with surprisingly little between them). Internet marketplaces like StockX and Goat that address the clothing market try to take that risky feeling out by acting as a middleman. Adding a human middleman to a transaction is not a fullproof solution though. You’re taking the trust that buyers/sellers were placing in each other and moving it into another source that can still act poorly. There are terrible examples where just as much if not more damage was done to the users after the middleman made mistakes (Mt. Gox is a good example of this).
Going back to peer-to-peer marketplaces, let’s look at sellers where a feedback system has designated them trustworthy. They have 1000+ good reviews, gold stars, etc. Even though these are honest people, the transactions they enter into are still not designed in a way that ensures perfect outcomes. There will still be the odd bad transaction (item not in correct condition, late shipping, seller on vacation). They could even pull an exit scam and rip people off after building up a solid reputation before disappearing.
Removing the Human Factor
“Ride-sharing” companies seem to have decided that they will eliminate bad human behavior through driver automation. Most people view this automation as a way to reduce their labor costs (and it definitely is the #1 reason companies are pursuing self-driving). However, there is a positive benefit here because the chance for two strangers to do harm to one another has been removed (though you also lose positive interactions). People going home alone late at night won’t have to fear when the car picking them up doesn’t have a driver. I don’t want to imply it’s just drivers that are the bad actors here, passengers act terribly every single day (throwing up in cars, distracting the driver, being rude, lying to get ride refunds, etc.) and without the driver present there isn’t someone for them to act in opposition to.
Automating Online Marketplaces
Taking the solution from ride-sharing that I believe they will ultimately implement, how could this in theory be applied to an online marketplace? Perhaps we could imagine a closed warehouse where rows of machines await items from buyers. When they receive an item they could validate that it is authentic (one piece of this puzzle that some companies are already working on), rate it’s condition, wait until it’s purchased, then pack/label/ship the item to the buyer on the other end. This is a pretty large jump from what is currently happening online, but not dissimilar from the jump from human driven cars to driverless.
I do think automation has the chance to greatly reduce the harm people do to one another and is a benefit usually overshadowed by the negative impact it will have on the labor force. Unless other novel solutions arise, I see automation as the potential solution/middleman that will remove the risk factor and chance for people to behave poorly on marketplaces and in other contexts.