> Agricola and Dominion fixed the problems Caylus had by adding significant randomization. Not only does that add strategic variety to the games [2], it also means that anyone can win any given game.
While Agricola and Dominion do a pretty darn good job of it, it's often a matter of individual opinion whether adding randomization is a 'fix' or not. I hate playing a game where I feel like I could lose by pure chance no matter what my level of skill is. There's a balance to be had between having guaranteed good strategies, too much of which makes a game feel stale, and real randomization, too much of which makes a game feel like glorified gambling to no effect.
What you really want is just some way of ensuring that you don't play the same situation twice, so that some mental exertion is always required, while keeping the playing field level. One way to go about it is to just devise a game with sufficiently complex positions that players are likely to create new ones on their own- that's what Chess, Go, and Hnefatafl do, despite being very lightweight. Another way is to find some means of randomizing positions that doesn't too seriously impact player standings- that's what randomized Nim does and what Backgammon tries to do. A third is to make game progression depend on player judgments/voting, which magnifies the social aspects of a game; it's a form of randomization, but fundamentally different from, say, rolling dice or drawing shuffled cards, and there are very few popular games I know of that make use of it (Dixit is the only board game I know of, though there are card games like Taboo and Apples to Apples). It seems to me that modern board games tend more and more towards the randomization approach (sometimes with good reasons to do so given the simulation conceits of the game- i.e., agricultural productivity in Settlers). This makes sense because its easy to do, especially in the context of trying to make games that play faster- randomizing starting conditions is a great way to save the first ten or twenty opening moves prior to interestingness that you'd go through in a fully deterministic Chess-style game, even if there's no further randomness- and especially if you're willing to be lax with the requirement to not obviously upset player standings by random chance. But I wonder if we're missing something by not doing more exploration of the space of "interesting" deterministic games.
Incidentally, Tsuro is an interesting edge case- as the board is built during game play, I'm not sure if that counts as randomized initial conditions, or continuous randomization. While Tsuro is the simplest example, there are several games similar to this, like Infinite City or Wasabi. I was recently introduced to a Tsuro expansion that added dice rolls that alter the previously-played board tiles, and it was pretty uniformly considered by the players to be inferior due to over-randomization destroying any real reliance on individual skill.
P.S. Most of my recent exposure to modern games comes from periodic sampling of board game events run by my university SF&F club, so it's entirely possible that there are filtering and selection biases in the kinds of games that I have seen. I would certainly not mind having my entire argument torn to shreds by being introduced to yet more new games that I was not previously familiar with.