Sky Team
I've been playing a lot of Sky Team recently. It's a two-player cooperative dice-placement game about landing a plane. I'm not a big fan of cooperative games, partly due to the quarterbacking problem, but mostly because I've come to dislike the zombie attack waves mechanic that was popularized by Pandemic. You are presented a problem, there is only one solution, and you better do it now. And that repeats for n more time until 120 minutes is up.
In Sky Team, nothing is really urgent on its own. A new player will try to do the rational thing at each step of the way; keep the plane level in every round, lower the landing gear as soon as possible. These are conditions that are only required on the last round, so you need to be consistent eventually, but have room to maneuver earlier on. The fun thing is you will come to realize that every irreversible action collapses the space of safe future states because most rounds you're balancing 5–7 constraints, each with a different acceptable band of die values, some of which are modified by previous choices.
Sometimes the right action is to let the plane go a bit off-kilter because doing so preserves good options elsewhere. A new player might nudge the plane level "just to be safe", not realizing they're burning the exact die they needed for something far more important, and suddenly we've entered a state where the expected value of survival drops by ~40%.
And no explicit communication is allowed, so there's also a small layer of exchange similar to bridge or hanabi. After a few plays, you develop these micro-signals: an early placement that looks stupid usually means "my roll is garbage please cover for me."
My favorite part is that the game tricks you into thinking it’s tactical, just react to the dice. When it's really strategic in a second-order, probabilistic way. You're not optimizing actions; you're optimizing the robustness of future decision trees.