Most places that I've been at that have worked this way have had periodic "checkpoints" in the tracker (usually when the code involved is sent to QA), where the current state is summarized at a 10,000 ft level.

In practice, these records are only read from beginning to end when someone new to the issue at hand needs to come up to speed. Otherwise, the value is in being able to search through the record to find specific things.

Jira has the ticket description section which gets updated to the latest goals / spec as things change. The history is for context of "who said do this" / "why did we do x"