But like you said "Attacker should wait for 1 PoS block and delay announcing this block." Why couldn't this process be simply reiterated to form a string? You wouldn't need to "find" >2 blocks in a row, you would just need to announce them in a row. Or am I missing something?
My edit of what I said above, must have been editing while you responded:
"So, in that case, what is preventing an attacker from waiting for multiple PoS blocks and delaying announcing the multiple PoS blocks to form a string of PoS blocks similar to a TX attack chain like the one anonymousg64 was talking about? If it is not possible to wait/delay more than one block per wallet, then one could easy use multiple wallets. If the timing of generating/announcing a single PoS block can be controlled, what is preventing reiteration of the process to control a series of single blocks?
Attacker should build a chain longer than main. The more he waits the less chance to success.
Also, what would prevent another block from being announced at that same time or right before you? Say you control when you can announce your PoS block, but does that mean you can control when other people generate/announce theirs?"
Similar to mercSuey's point that PoS is a stochastic process,
What would prevent another block from being announced at that same time or right before you? Assuming you can control when you can announce your PoS block, does that mean you can control when other people generate/announce theirs?
Yes, this attack has not 100% chance to success. The point is that average block time is known.
One honest PoS block will not stop attack. Two will.