Before I get into the question I wanted to point out that I am referring to a specific part of a specific article. You don't actually have to read the article to help with the question but it's here if you're interested.
In that article, two flavors of economic finality are described.
The first one seems to describe PoW, where validators lose money if the block they added does not end up in the final chain:
A block can be economically finalized if a sufficient number of validators have signed cryptoeconomic claims of the form "I agree to lose X in all histories where block B is not included". This gives clients assurance that either (i) B is part of the canonical chain, or (ii) validators lost a large amount of money in order to trick them into thinking that this is the case.
I don't quite follow the second item though:
A block can be economically finalized if a sufficient number of validators have signed messages expressing support for block B, and there is a mathematical proof that if some B' != B is also finalized under the same definition then validators lose a large amount of money. If clients see this, and also validate the chain, and validity plus finality is a sufficient condition for precedence in the canonical fork choice rule, then they get an assurance that either (i) B is part of the canonical chain, or (ii) validators lost a large amount of money in making a conflicting chain that was also finalized.
This second description presumably refers to PoS, but this explanation feels very abstract to me. Can somebody explain this to me in different words?