[everyone]

Your sort of reiterating my point.

I know UI stuff breaks a lot, but my point is that when it does break, its obvious. Bugs should be easily caught in human acceptance testing.

Also an automated test UI is never going to be the same as a human test, a lot of UI bugs can result from the vagaries of various physical devices touchscreens / gamepads / motion controls etc.

Also anything that tests such high level functionality is going to be very complicated and touch many parts of your code, it will probably be more complicated and less maintainable than the code its testing, which defeats the purpose imo, do you want to write tests for your tests?

UI stuff tends to change radically over development, so you'd be more than doubling your work with all the testcode youd have to write.

and it would still never replicate an actual human running the actual game, you absolutely are still going to be doing human acceptance tests + the closer you try and get automated tests to that the more complex and less maintainable they get.