Confused by this; I assume you mean the opposite? IE in 90% cases you shouldn’t change the config? Given the above what would be a good use case for Split Chunks; if not to split in the way the article suggests.

Say I have a number of core packages that I update a lot; would it not make sense to split these out as single chunk modules so as to not blow away the entire cache on change?

Keen to hear your thoughts.