It's the kind of thing that's very easy to accidentally write, it's not that shameful. What's shameful is not investigating the load times at all, since the problem is so easy to see when any measurement is done.
It's easy to write because it doesn't run noticably slowly on smaller data, and it's easy to accidently introduce quadratic behaviour in some systems. Obviously if someone tested this on 10MB of JSON and saw it took minutes and thought that was reasonable, that's a bit ridiculous, but I doubt at the time the code was written anyone expected it to be fed such a large JSON object.
