I don't know why some devs think rewriting everything will make everything better, the resultant lost business knowledge be damned. From what I've seen the ones who advocate that usually just don't know what they're doing in general.
For what it’s worth sometimes it’s worth abandoning old legacy context in a rewrite if the new solution drastically reduces operational complexity. For example, it’s cheaper/better to retrain everyone on the new solution than it is to continue to onboard new folks to the legacy one. This is not always the case, but sometime it is as the parent comment mentions.
Can you point to an example of a software project that was re-written without the rewrite causing more problems than it solved? Joel Spolsky of Stack Overflow wrote an article on this very topic, titled "Things you should never do - part 1“