Difficult. Not so much because of any interest in hoarding code or what not, but simply because said code is already deployed on some live service or another and you really don't want that stuff being broken or potentially broken just to 'tidy it up a bit'.
Delete old code? Not hard. Getting the ability to delete old code? Harder. Essentially, going back and refactoring is not possible unless it is linked to a task in Jira. So, we can't just go and fix code that badly needs some help unless it is directly or tangentially related to an issue. It's understandable but frustrating as there is a lot of code and not enough time.
Very difficult:
Technical debt and systems that are older than today's first-year students in computer science relying on broken implementations. It is work-arounds all the way down.
Very difficult. It's due to what I consider a hoarding mentality ("what if we need it later?") and the misconception that there is no cost to keeping it around.
Can't you solve "what if we need it later?" mentality with git (or some other append-only repo) and static analysis?
Yes, but if the code isn't in the current version of production or master, there is a high probability that nobody will know or remember it even exists.