The action card system in 404 has some hidden properties that I'd like to talk a little more about. Generally these are things that I don't imagine my players will notice on a conscious level, but I assume that anyone reading here is interested in the things that underlie game design so it might be interesting. The system is that each player draws five cards, each of which have two actions, from these they select three cards and perform one action from each of them. The trick comes in the comparison to the "Each player has all of the actions and can do whichever one they want." system.
Drawing the action cards is only relevant if it restricts options on at least some occasions. If a player always drew one of each action that they needed the act of drawing becomes pointless. However preventing a player from doing things that are fundamentally necessary to the game will end the player's ability to interact with the game and kill their enjoyment. So for the system to work well it needs to be relatively frequent for players to not draw exactly what they want, while ensuring that players get a minimum of useful material.
Originally I approached this problem by having players draw a hand of cards with one action on each and mulligan their hands if their options were poor. The option to discard a hand of five unwanted cards to draw four new ones, repeating the process until they had something that they liked practically guaranteed each player at least one useful move. However it was cumbersome, adding time to the game that was spent on an activity that wasn't much fun (discarding and drawing - or for most players, watching someone else discard and draw.)
Assigning two actions to each card proved a more elegant solution. Each player's five card hand effectively provided ten options, making it very likely that they'd have important options. However each action selected disabled another choice (the other action on that card) which meant that there's still a reasonable chance that players wouldn't get exactly what they wanted.
That still leaves the question as to why restricted choices are desirable. A common answer is that it prevents analysis paralysis, but while that might be a handy side effect it's not my main interest. At its core 404 is a game about solving problems in novel ways, the best moments in the game aren't when someone finishes a directive in the most action-efficient possible way. The moments that players remember are the times that they pull off something without the right cards, using other player's movements or elaborate workarounds to get things done. You can see the feeling I'm trying to capture at 1:28 on this video.
It also adds a layer of strategy to the game. I've often argued that meaningful decisions are important and this mechanic adds "What position does this leave me in for next turn? How likely am I to be able to follow up in the right way?" to the list of things that players need to consider in making their moves.
Finally it assists the game's deception layer. Players have (partially) hidden objectives and will sometimes wish to perform actions in service to their objectives without giving away what they're trying to do. The restricted cards offer players a plausible reason to perform actions other than that they lead to a particular directive being completed.
The approach is not without its drawbacks, the biggest of which is the danger of a player having a turn in which they're not able to do anything in service to their objectives. Such turns are poison to the game; killing engagement for that player and making them feel like the outcome of the game has been based more on luck of the draw than on their actions. It's impossible to entirely eliminate these turns, there's always an outside chance of drawing a combination of cards that ruins everything forever, but I've done what I can to mitigate it.
Early steps included trying to make the turns quite short so that a wasted turn wasn't annoying for too long and to make sure that every action did something to reduce the frequency of useless actions. Previous versions of the game had action cards that moved players in a particular direction and if they had reached the end of the ship then they did nothing, changing this so that each movement card features a symbol and every room has each symbol attached to a door so that a movement card will always let the robot go somewhere helped a great deal.
There was also room for more subtle assistance in the distribution of actions throughout the deck. Some actions can be simulated using a greater number of other actions, for instance the organise action allows a player to alter the order of items in a stack, where the pickup/drop action allows players to pick things up and put them down. Enough pickup/drop actions could simulate anything that'd be possible with the organise action, though the method might be a bit arcane. As such I could increase the number of them in the deck and take some organise actions out, so that a robot is more likely to be able to progress towards its goal while being less likely to get the perfect solution.
Combining these features makes it increasingly unlikely for players to experience entirely ineffective turns, while still rewarding a flexibility of thought and action in how a problem is approached. It does have the unfortunate effect that a lucky or unlucky hand can impact on a player's odds, but over the course of the game these random factors tend to balance out and player skill still seems to be a greater predictor of victory than luck of the draw. A lot of players like a bit of randomness anyway, given the madcap nature of the box art I should probably be more worried that people will over-estimate the role of randomness in the game.
Hopefully that gives a little bit of a deeper insight into why I'm using these mechanics in this way during my game. While I've arrived at a lot of these decisions through playtesting and development, there's nothing new under the sun and I'm sure you recognise some of them. Thinking about it has given me a deeper understanding of other games, as a player I'd thought the double sided action cards in Space Alert were just to give players room to make an extra type of mistake. Now it seems likely that the combination of restricted choice, time limits and creative solutions was on the designer's mind. I'd also figured that players shared a deck in Robo Rally because there was no viable alternative, but because the restrictions helped to make the game work and could be fed into other mechanics (damage). The nature of the movement dice in Runebound plays into the notion of restricted play, but with options always available as well.
These sorts of consideration are all over the gaming world, but I had to start creating my own game and seeing what worked and what didn't to really get to the meat of "why" on a deep enough level.