Simplified State Verification and the blockchain compression are the most promising parts IMO (since they solve problems we have today, not hypothetical ones).

However, after reading the SSV description, the lighweight SSV client seems like a simple frontend to the SSV-providing node, which puts the question how will the lighweight client "know" that the SSV-providing node is trustful?