3 ms·
This raises the "oracle problem" which is that blockchain-recorded assertions about things outside of the blockchain are not necessarily correct (unless a commu
by schoen 8y ago
This raises the "oracle problem" which is that blockchain-recorded assertions about things outside of the blockchain are not necessarily correct (unless a community has agreed to treat them as definitionally correct).
If you have a trusted oracle for people's identities, you might just be able to trust that oracle's assertions (maybe that's what you're referring to as "cross signing") without also recording them in a blockchain, unless you're concerned that the oracle will tell the truth initially and then lie later on.
- jacobush 8y agoI meant on a very low level - a central database can be manipulated to effectively rewrite history. If you have a web of signatures, similar to PGP, Keybase and a few other systems, you can't do that. The same goes for a blockchain. What you put in there is susceptible to the oracle problem, but that is another level. Both these things need attention.
- schoen 8y ago> a central database can be manipulated to effectively rewrite history That's true, and that's an important issue. But I might still ask, for each application, is there someone who's trusted to introduce records into the system but not to update them later? (We can definitely envision applications where the answer might be "yes" -- often related to concerns about censorship -- but it doesn't seem like this is the case for every application.)