Image: G.Garitan/Wikimedia Commons - CC: BY SA
I'm going to propose one anyway: there are only 3 possible actions in any given turn of any given turn based game.
That's right. Every roundel, every worker type, every act, can be boiled down to three basic parameters:
* Gain
* Redistribute
* Risk
Gain means getting something for nothing. The Village card in dominion is a gain action: it costs 1 action and 1 card to play but gets you 1 card and 2 actions. Net effect: gain 1 action.
The "take wood" action in Agricola is not a Gain action. It gains you X wood, but it costs you a worker. Thus, it's a redistribute action: you're redistributing your resources, from workers to wood.
Note that a gain action can be negative: lose 3 food is the same as gain -3 food (there's an analytical point to this, see below).
Risk is anything that relies on wagering and chance, something that can, but doesn't have to redistribute your resources. Note that a risk action is not a gain action. If you roll a die or draw a card to get something, even though there's a chance element involved it's a gain action: you're not risking anything but getting the stuff for free.
Gain actions are usually dependent on chance, or passed to the cleanup phase. The reason for this is that gain actions aren't very interesting in themselves. Gain X wood, Y cloth and Z gold is nice but there aren't any decisions involved - you stick your paw in the bank and withdraw your brand new resources. There's no trade-offs. Gain actions are no-brainers: if you can do them, do them.
But what if it's a gain action that affects something else? Say that I get 2 wood, you get 1 wood. Then it's a redistribute: you redistribute the resources balance around the table. Thus you might be increasing your power, in terms of resources, but increasing an opponents power as well. There's a trade-off in that, and an amount of thought. Of course, if you have only two players and the resources are worth exactly the same to both of you, there's no freedoms or balances of power involved, then it's a no-brainer again, and once again a gain action: you gain 1 wood (even though you took 2 wood, as in terms of power you've gained 1 over your opponent). This is usually the case in a zero sum game when you get VP if the VP can't be used for anything in the game: if you get X and your opponent gets Y and X>Y then you've in effect gained X-Y. Gain action.
Redistribute and risk actions are what most games are about. You redistribute the resources you've got. You redistribute your forces on the map. You buy, sell, trade, build, anything that converts something into something else is a redistribute action. Anything that risks something you've got in order to gain something you haven't got, like attacking with your armies, wagering at the casino or opening up your position in Chess in order to draw your opponent in, is a risk action (Ok, I haven't got the foggiest if you do that in Chess - it looks like it, but I'm not a skilled enough Chess player to know).
Note that we're looking at a single turn here. If you've got a real-time game then there are other factors involved, and if you're looking at multiple turns then quite a lot of gain actions become drawn out redistribute actions (spend something this turn in order to get something in a later turn). And a lot of redistribute actions become risk actions (taking the clay now and risks not getting the wood next turn).
Ok, let's assume that I'm right (isn't it nice when the writer's the one to make such assumptions?). What does it mean for game design?
First off, gain actions are boring. Chalk them up as bookkeeping and minimize them if you can't get rid of them completely (imagine a game where you get 10 000 cubes each turn - but everything costs 1 000 cubes - cut out the zeroes and you've minimized the bookkeeping).
Gain actions as a result of previous turns' action's are rewards and should be encouraged (this is what all engine building games are about) but they carry the danger of snowballing and giving you runaway leader/fallaway loser syndrome.
Gain actions as a result of chance run the risk of deeming the game random, especially if they're both positive and negative. If you've got such actions it's often better, for the players' emotional enjoyment of the game, to make all gain actions either positive or negative. Then you've got something that players either long for or dread but not the position where players curse the fates they can't control (note that it's completely different if the player has done a risk action: that's their choice - a gain action forces the effects on them).
Redistribute actions are the interesting parts in most games. They tend to obfuscate the effects of a player's agency: with enough redistribute actions you have a hard time predicting what will happen, sometimes to the point of the game feeling random. The danger with redistribute actions are chains: if you can chain an action into others in such a way as to end up with what you started out with + something else then it's a gain action and should be cut. But don't forget that there's an action cost as well: if you use 3 workers to convert 2 wood into 2 wood and 1 clay then you haven't gained anything, you've redistributed 3 workers into 1 clay. So if you find yourself with a gain action where you want a redistribute, take a look at what time costs or limitations you can add to it. Or convert it into a risk action by breaking it up and letting other players act. Thus you could start out on your 2 wood into 2 wood and 1 clay action sequence only to find that another player's action has made it impossible and you end up with 1 wood.
Risk actions are interesting as well - provided that they're predictable or controllable in some way. And provided that they have some form of decision or trade-off involved in them even if it's imaginary: Rock-paper-scissors is (slightly) more interesting than two players rolling a die and comparing results even though the RPS choice is pretty much an imaginary one as you can't predict the outcome or chances in any reasonable way (at least if you play against someone who's got some idea of how to play RPS).
There's a pretty fine line between Redistribution and Risk. Take a look at the first Civilization: moving your troops is redistribution. It's a trade-off between time (to get there) and opportunity (what you can do and where). But once you build railroads you can move for free. Movement is no longer a question of redistribution (ok, a bit since you're still limited to land movement). Since you can move anywhere you want for free, movement becomes a gain of position. Instead the placement of your troops becomes a question of risk: do I put them here to defend or there to attack. What if I leave this city without defenses and go all in on that one (OK, the AI is pretty bad in Civ I so in reality most of the late game is sliding into gain, but assume the AI would have been competent)?
One could argue that there's a "best action" risk in every redistribution: if I do this, am I risking that I could have done something better instead? This would make every redistribution into a risk. In my opinion it's a matter of where you put the cut-off point: do you consider the decision a part of the game itself? If so then yes, every decision is a risk and there are only two types of actions: gain and risk. But my view is that you can never know (unless you're a hype-optimizing gamer with near limitless mental facilities, or are playing a small board game of Go) what the final effects of your actions will be and how the other players will challenge the outcomes. That, after all, is why we play games: to pit our minds against others'. Thus I don't place the decision point within the scope of the game but rather within the scope of the meta-game. And as I've said, this model works when looking at one turn, it becomes much more complex if you look at multiple turns and that most engine building gains become redistributions and the whole model falls apart.
