Nick Szabo  🔑

One's own full node is far more trust-minimized than an SPV node, but an SPV node that depends on the consensus of 100s of jurisdictionally diverse full nodes of others still involves much less trust than PayPal. I favor former but "might as well run PayPal" ill describes latter.

Running a full node really only allows you to know if the chain is being 51% attacked. It doesn't actually make you any less vulnerable to such attacks (FNs are still vulnerable to double spend attacks/re-orgs). The security of both spv and full nodes relies on >50% being honest.

Knowing makes you less vulnerable if you have the capability to respond, e.g. by finding other correct nodes and maintaining your own un-attacked fork.

But there is nothing invalid about a chain that has been re-orged, miners will be economically incentivised to mine on the longest (re-orged) chain.

Right, knowing about a 51% attack doesn’t change anything. Miners are not required by consensus rules to build on the blocks of other miners.

I am not sure what you are trying to say here but miners are incentivised to mine on the longest (valid chain) to decrease the likelyhood of their blocks being orphaned

Assuming that incentives will make people do the right thing is still an assumption. Without full nodes to hold minors accountable that incentive goes away.

Agreed. I am talking about what security diffs a full node vs an SPV node has for an individual actor, not under the assumption that everyone runs SPV. The valid chain is the chain with the most difficulty that still complies with protocol rules.