I was trying to hint that someone in practice writing an App using a GoL as a platform running a turing machine running an app is about as unlikely as Apple actually rejecting a GoL implementation for this reason.
Or in more pithy terms, "The gap between theory and practice is narrower in theory than it is in practice."
p.s. Speaking of theory... One of my favourite books is "The Recursive Universe: Cosmic Complexity and the Limits of Scientific Knowledge," wherein the Game of Life is used to discuss complexity, self-replication, and other fascinating topics:
If you want to see an example of where this approach might have helped, look at the children convicted of making kiddie porn for taking cell phone pics of themselves. You could have brushed off reductio ad absurdum arguments by saying, "Oh, that precise situation is so unlikely," but the flaws are real just the same.
AFAIK, the game of life (as a mathematical concept) is in itself a turing complete automaton, you don't need to explicitly simulate a turing machine with it, in the same way that you don't need to implement a turing machine on top of the lambda calculus to be able to express algorithms with it.
In practice, since all computers have a limited memory, they are in fact deterministic finite-state machines. So the size of the board/RAM/whatever constrains what you can do with it, but it is as true for Life as for any language or computer system.
From a practical standpoint, watching a board of Life is indeed about as interresting as watching a random memory block rendered on screen while your program is running... It's sometimes entertainging, but usually boring.
I'll admit I actually have Life on my iPhone :D