The problem is that people have different definitions of "not working." As a proxy provider, our definition of a "working" proxy is one that you can connect to and is fast. In that sense, all our proxies are "working" almost all the time. We do have constant monitoring and we do fix issues when they crop up. When a proxy is unconnectable, we fix it immediately. The issue arises because people use proxies for a variety of other tasks: Instagram, Pintrest, Google, etc. So even though a proxy might be completely working for someone using it for Instagram, it might "not work" for someone on Pintrest. That doesn't mean the proxy is "dead" or "non-working." It means Instagram/Pintrest/whatever has blocked that particular IP.
When this happens, we gladly refresh your whole set of proxies because obviously you need the proxies to work for your particular task. We don't have a system in place to replace specific proxies. The reason for this is two-fold. One, the amount of customers who want it does not justify the financial implications of implementing it (these customers tend to be the most abusive anyway). Two, by replacing proxies that don't work for a particular purpose with ones that do, we are accelerating the rate at which the working proxies get banned. It's not in our best interest to have customers burning through proxies.