The real problem is not that people don't test on all browsers, but that people have to test, still, on multiple browsers. The standardisation does not work well, new stuff is out constantly, vendors not independent experts control the process.