Envision a scenario where an anonymous person, in a randomly selected country of the world decides they need a $6,500 USD loan (or 2 ETH for purposes of making location random). They open a crypto backed platform, type "2 ETH" - "Enter", and within minutes creditor bids begin rolling in - all of various maturities, various interest rates, and possibly more exotic terms like equity conversion.
We must cover why anyone would consider crediting an anonymous user as it breaks everything the current system relies on. On an individual level we use a "credit score" to determine the risk, on a corporate level a "credit rating". In essense, these things capture 2 data points - the borrowing history of the lendee, and their future ability to pay it back, quantified. Both of these data points will impact the terms of the lender offers the lendee.
We can use a blockchain to track the history of a lendee, but it is much more difficult to quantify their ability to pay it back in the future. Yes, we could call it day, and set extreemely high interest rates to compesate for this unknown, but I propose that the credit system is permanently broken from being anonymous. We can use the following examples for illustration of it's shortcomings.
Let's define an example payback ability metric, or PAM. This PAM will simply take inputs of work history of the lendee and output a binary decision of yes or no. In this example, if a lendee has held any job for the previous 365 days, they are a yes, otherwise they are a no.
How can we determine employee history through crypto alone? Well, that's the missing piece. It would seem that the only solution is to break decentralization and anonymity at the same time. A trusted entity must verify that a specific person has been employed like they claim to have been.
At this moment, I am uncertain that credit can be a completely anonymous process. A belief I hold loosely is crypto is a beautful abstraction that suffers from it's very nature of being anonymous. In making the "transaction" the only solidified concept, it suffers from data loss. When you create an application layer on top of crypto which only associates with a transaction, the actual blockchain has no awareness of the critically important metadata that the application is tracking.
Futhermore, a lendee's blockchain history is also extremely gamable. To build "credit history" a lendee could simply move money through wallets over a period of time that makes it seem like it was a loan. Now, a blockchain must be able to determine if a wallet holder is secretly transacting with themself or a friend to boost their history. To solve for this, it would appear a trusted 3rd party is the only way to verify the legitimacy of a transaction being accepted as "credit history" worthy.