[hiq]

If you use tokenized versions of things, don't most of the 5 benefits listed previously break down?

If you have an entity selling some tokens representing GOOG on some blockchain, then this entity can (will) be regulated, is a central point of failure, and so on. This problem of interfacing a blockchain with the real world seems like a common theme that's swept under the rug, and I've never seen a convincing answer. What am I missing?

[vmception]

oracles. people are using APIs to get pricing data of real world access, and create entire nodes of networks that have to agree, in the blockchain space they call these oracles. for tokenized assets this just makes them synthetics, indeed one of the biggest services you can use for creating and trading synthetics is called "Synthetix"

a centralized asset backed model is only one possibility and that does have the central point of failure.

it is extremely lucrative to build these API and API networks right now as an 'oracle provider', and decently lucrative to be a highly available oracle.

In the crypto space, both the oracle provider is tokenized with the tokens being typically used to get access to the API network (friction is reduced in some models by only charging/bonding the nodes that want to join the network, instead of consumers) ie $LINK, $BAND, $TRB, $DIA, and the derivatives provider would be tokenized as well ie. $SNX, alongside all the tokens you want to create and trade, or just trade

[hiq]

I guess it's not using just one oracle, otherwise you still have a single point of failure, right? So N oracles are used? What happens if 1 out of N is malicious? k out of N?

How do you check that the N oracles are not actually run by the same person behind the scenes? How do you make sure that it's not economically profitable for an entity to run these N oracles and send bogus data, even if only once?