I'm not sure what exactly were you trying to argue against. Pretty much all real world distributed systems are perfectly asynchronous, don't have perfect clocks

I'm not sure what exactly were you trying to argue against. Pretty much all real world distributed systems are perfectly asynchronous, don't have perfect clocks perfectly synchronized to rely on them, don't have bug-free real-time algorithms, OS and hardware, etc. So real world is exactly where the paper applies, i.e. it is impossible to guarantee consensus in bounded time. Sure, with large enough bound and large enough quorums you can get very high probability of achieving it in bounded time, but never 100%.