Partial vs Complete Catchup: what are the differences in terms of security?

Question

Lets assume I power-up a new core and join (as validator) to a network of pre-existing other nodes. In addition, to speed things up, I chose to do a partial catchup from one history archive. Now, having synced up, is it possible that my new core has a state that's been tampered with?

In other words, when I perform partial catchup, what are the risks in terms of security?

UPDATE: I'm slightly changing the question to make it more interesting: Suppose the network I join has conspired against me and they've all tampered with the state (the bucketlist) stored on their history to reflect a state that's different than the one I would arrive to by replaying the txs from the genesis block. Is this possible? If it is, then a full catchup does have benefits over a partial one.

Answer 1

Each node synchronizes the state with the quorum set, so archives tampering won't work. The node will be unable to catch up with SCP quorum if the local state differs from the quorum-approved state. Nobody can modify a single archive file without messing up the whole archive because each ledger contains hash of the previous ledger and they are validated during the catchup. The node will discard the ledger with invalid hash.

Therefore, there is no difference whether you are doing full or partial catchup.

EDIT:

1. Attackers control both the quorum set and archives destination → Newly-connected node will be 100% compromised.
2. Attackers control only the quorum set or the archives data → New node will connect and follow the SCP majority, but won't be able to catchup with the history. It will be unable to apply the state from buckets if hashes don't match. I don't think that the node will be able to act as validator in this case.

If the archives are tempered, you won't be able to ingest the history regardless whether you are starting from the genesis block or not.