If you suggest a number like 50, I would (and I would imagine the people who actually have to implement it) like to see proof that a substantial number of bad responses comes from people with 10 < rep < 50 and that it would make a dent at all in the signal-to-noise ratio.
There are approximately 2800 users with between 10 and 50 rep who have been seen in the past 180 days that would be impacted by this change. This is out of roughly 30000 total registered users who have also been seen during that time. So your proposed change to the barrier would affect somewhere around 9% of the "active" user-base. For comparison, there are approximately 20000 users in the past 180 days with a reputation of less than 10 or exactly 101 (the association bonus). So the current barrier limits posting such that only 33% of the active participants can post. Your proposed barrier would limit the participation to 25%.
(Disclaimer: All of these numbers were found by poking around in the Data Explorer today -- the last data dump was Oct 25, 2015 so it should be close to current numbers. If anybody in the future decides to run the numbers again, they may not match exactly!)
Therefore, you must somehow show that a non-negligible number of the bad posts to which you are referring are posted by this 9% of the site and that changing the barrier would significantly reduce the amount of bad posts. These numbers are based on your proposed barrier of 50, if you pick a different barrier, the same analysis would apply and the same requirement to justify the number exists.