Smart contracts are blind by nature: they can’t see the off-chain world, and they can’t read their own past. The industry’s first answer was the oracle—bring an external number on-chain: prices, exchange rates, interest rates. It solved the question of "what is it right now," but not "how did this number come to be": which sources were used, how anomalies were filtered out, what weighting rules applied—those were still computed off-chain by the provider and copied onto the chain.

What’s more troublesome is history. A contract may want to know "how many times this address has been liquidated in the past eighteen months, and how long it took to recover each time". An oracle can’t help—that is not a point-in-time value, it is a query.

@SpaceandTime changes the layer at which the answer is provided: instead of transporting numbers, it transports "queries that can be verified". A contract sends a SQL-like request, receives the result plus a proof, and uses it directly after verification passes. The difference between the two is not speed, but where trust sits: an oracle says "I trust this data source," while verifiable computation says "I verify this computation"—the former concentrates trust in a few providers, the latter reduces it to cryptographic assumptions.

This brings three capabilities that were previously unavailable. First is historical state: a contract can read markets that existed before it was deployed, not just the current block. Second is cross-protocol behavior: the activity of an address in other lending markets, or on other chains, can be incorporated into the same judgment. Third is custom criteria: a protocol can define its own risk formula instead of accepting a score provided by a third party—a score cannot be explained, but a query can.

The division of labor should also be made clear. For high-frequency, low-value, ultra-low-latency scenarios like price feeds, oracles remain more suitable; generating a proof for every price read is not cost-effective. The main cost of verifiable computation lies in proof generation, so it is best suited to low-frequency, high-value judgments that need to be auditable: liquidation thresholds, collateral ratio adjustments, reward settlements, audit criteria, cross-protocol limits. It is not a replacement, but a layering.

The first half of on-chain finance solved "seeing the outside world"; the second half needs to solve "explaining where it came from." #sxt $SXT @SpaceandTime