On chain credit scoring… how do we verify who is who?
A high volume of transactions could be just a retail p2p trader without any verifiable business that’s worth a bankable score.
Onchain data mainly shows volume not intent.
For banks to verify the source of money to give a person a good credit score using onchain data, there has to be a lot of identity and business verifications, so as to know if for sure they worth the score, and that brings up the issue of privacy for some people.
Infact most people, cause now their money can be fully tracked and taxed.
Now, let me address each one. On the identity question, you are right that we cannot just point at a wallet and call it a borrower. That is why I framed the identity layer as the foundation that the entire stack sits on, not a feature you bolt on later. BVN and NIN already exist in Nigeria. What does not exist yet is a clean mechanism that ties onchain activity to those identities without dumping someone's full transaction history into a lender's database. Zero-knowledge proofs make this possible in principle, in such a way that a borrower can prove they meet a creditworthiness threshold without revealing every transaction that produced it. Again, the building blocks are there, but nobody has assembled them for this market yet.
On volume versus intent, that is honestly the sharpest challenge in this whole conversation, and you are right to raise it. A real scoring model has to go deeper than volume. It looks at counterparty diversity, whether flows are cyclical or accumulative, settlement regularity, and repayment history on any prior credit. Those signals should together tell a better story, even though it is significantly harder to build than a volume score. I guess that difficulty is precisely why it has not been built yet.
On privacy and tax exposure, I will be honest: this is the hardest adoption barrier in the piece, and I did not give it enough space. A meaningful chunk of the Nigerian informal economy stays informal on purpose, and an onchain credit identity creates visibility that was not there before. The trade-off is real. Access to credit at 15% annually versus 20% monthly is a compelling enough offer that many people will accept that visibility cost. But not everyone will, and some of the people who will not are the most financially excluded to begin with. But then again, privacy-preserving verification design is not optional if this stack is going to achieve real adoption.
Please also remember that I wrote this piece not to claim that these problems are solved, but to claim that they are worth solving.
My mentor is back on a roll
How about adoption issues for them
On chain credit scoring… how do we verify who is who?
A high volume of transactions could be just a retail p2p trader without any verifiable business that’s worth a bankable score.
Onchain data mainly shows volume not intent.
For banks to verify the source of money to give a person a good credit score using onchain data, there has to be a lot of identity and business verifications, so as to know if for sure they worth the score, and that brings up the issue of privacy for some people.
Infact most people, cause now their money can be fully tracked and taxed.
Yess , I agree with you that they are problems worth solving.
There would be a lot of barriers to solve.
Privacy and worthiness.
Taxation and privacy.
The beautiful thing about this is and why I really love the article is because it’s forward thinking.
Once this issue can be solved, scaling on this foundation would be extremely big for financial institutions.
Very well put response btw.
Lovely question, and a fair pushback!
Now, let me address each one. On the identity question, you are right that we cannot just point at a wallet and call it a borrower. That is why I framed the identity layer as the foundation that the entire stack sits on, not a feature you bolt on later. BVN and NIN already exist in Nigeria. What does not exist yet is a clean mechanism that ties onchain activity to those identities without dumping someone's full transaction history into a lender's database. Zero-knowledge proofs make this possible in principle, in such a way that a borrower can prove they meet a creditworthiness threshold without revealing every transaction that produced it. Again, the building blocks are there, but nobody has assembled them for this market yet.
On volume versus intent, that is honestly the sharpest challenge in this whole conversation, and you are right to raise it. A real scoring model has to go deeper than volume. It looks at counterparty diversity, whether flows are cyclical or accumulative, settlement regularity, and repayment history on any prior credit. Those signals should together tell a better story, even though it is significantly harder to build than a volume score. I guess that difficulty is precisely why it has not been built yet.
On privacy and tax exposure, I will be honest: this is the hardest adoption barrier in the piece, and I did not give it enough space. A meaningful chunk of the Nigerian informal economy stays informal on purpose, and an onchain credit identity creates visibility that was not there before. The trade-off is real. Access to credit at 15% annually versus 20% monthly is a compelling enough offer that many people will accept that visibility cost. But not everyone will, and some of the people who will not are the most financially excluded to begin with. But then again, privacy-preserving verification design is not optional if this stack is going to achieve real adoption.
Please also remember that I wrote this piece not to claim that these problems are solved, but to claim that they are worth solving.
And thanks for the proper explanation.
Yess , I agree with you that they are problems worth solving.
There would be a lot of barriers to solve.
Privacy and worthiness.
Taxation and privacy.
The beautiful thing about this is and why I really love the article is because it’s forward thinking.
Once this issue can be solved, scaling on this foundation would be extremely big for financial institutions.
Very well put response btw.