Indeed the limit comes from the QA_MAX_LIMIT_START constant. I can think of two reasons why it is there. The first one, as the comment in its definition explains, it is performance. The deeper the database has to dive in the index the longer the query will take.
The second one, is related to getting your data copied. It is relatively simple to build a web scrapper that would go through all questions and copy their content. The more accessible questions there are, the more questions will be cloned. Also note that it is hard to believe a person has actually reviewed 50000 questions. Requests to those pages are likely to be bots.
Conclusion: If you don't care about getting your site fully webscrapped, feel free to increase it. Just try reaching the first question in your site and time it.