The purpose of Candyland is to play a game, without playing a game. It's for toddlers to learn to take turns, to accept negative or positive outcomes without stomping off in a huff, and to play a game with a level playing field between the adults and children... and at the level we are talking about, if there is a meaningful choice at all, then the child, or perhaps rather toddler, will simply lose. It's more than just "following directions", it's the whole set of skills necessary to play a game.

This is why family games have a gradient to them, gradually trading chances for decisions, and gradually expanding the state space, until the game player is ready to join the adults fully with something like Scrabble in the early teens or so. The mentioned Connect 4, for instance, is relatively simple and can be effectively "solved" by an 8 or 9 year old, for instance, so a 8yo and an adult are still not separated by such a large gulf that the game is a joke. Children are actually pretty good at figuring out that they are being "let" win.

If you replace Candyland with something that has any choices at all, you've replaced something other than Candyland.

That said, I'm not saying the progression is optimal as it stands, but if you don't understand the reasons for the enduring popularity of the "standard game loadout", which hasn't seem to have hardly changed in 50 years, you're not going to improve on it properly. And I'd say most of the flaws are on the higher end and solved with things like Settlers of Catan and such, not the children's part of the progression.