Game designer Greg Costikyan has written a detailed analysis of why Candy Land succeeds as a game in spite of the fact that winning the game is completely random and requires no strategy of any kind:
There are those who criticize Candy Land as being jejune and ultimately futile, since the nature of its rules construct and the (non-existent) emergent complexity it supports is utterly unsusceptible to any sort of rational analysis, or indeed, choice of player strategy.
... let us view Candy Land as a mathematical entity. It is very nearly a Markov chain, a stochastic process in which, given the current state, future states are independent of past states. (It would be a pure Markov chain if the deck were shuffled after each play; instead, it is a crippled Markov chain coupled to a push-pop stack.) As such, it is a metaphorical representation of the fundamental ideology of the United States; the past is no constraint on the future, and each individual should strive resolutely for personal advance despite whatever the past may hold.
That's my five-year-old's main complaint - that Candyland is nearly a Markov chain and not a perfect Markov chain. But, you know - kids.
How does Mr. Mongo fit into this equation?