Assuming a game has a large number of objects and actions an avatar can perform, which do you think is clearer to the player, a design built around choosing objects and performing actions granted by them or choosing actions and selecting objects to realize the action?

For example, consider planting a listening device: If there were a range of devices to choose from, it might make more sense to have a "Bug" or "Surveillance" action with the ability to choose the right bug to plant. But if there was only one type of bug but several ways to plant it, it might make more sense to select the listening device and then choose which action you wanted to perform. But if there is both?

Another (more out there) example: Let's say you have a city and can do things at an abstract level in different locations within the city. Maybe you can burgle or vandalize, or help the needy or guard locations. Which makes more sense, to have a range of actions you choose and then locations to perform them (with each block sort of acting like an object), or to maybe make each location context sensitive, showing the available actions maybe when you click on it?

which do you think is clearer to the player, a design built around choosing objects and performing actions granted by them or choosing actions and selecting objects to realize the action?

Why wouldn't you offer both, mix things up?

object - action vs action - object is a classic design choice required in user interfaces.

object - action is usually deemed more intuitive, and seems to be followed more often.

but action - object can have advantages.

some examples:

in the world of desktop applications, the object - action interface is usually used: file - open being the all time classic.

but note that right next to that you'll have edit - select all, which is action - object.

The SIMs is a classic example of object - action. click on an object - select an action.

i once did a person simulator (like The SIMs) with action - object interface. this allowed you to cue up actions that used objects that weren't visible on the screen. so you could cue up move to lake, drink, bathe, fill water skins, and move back to cave actions all at once.

i'm now working on a new version of the same game, with a fps type interface. it is largely object-action. select something in the environment, and get a menu of actions for that type of object. if no object is selected, you get a menu of general actions that can be performed anytime, anywhere. this menu is action - object in style (example: make - stone knife).

a lot seems to boil down to user preference. some (like me) tent to think in an action - object manner, but most seem to prefer object -

action.

note that for some action - object combos, one way will definitely seem more intuitive than the other. such as select - all vs all - select.

in that case though, it may just be the way we've all learned to think because all software is written that way. so it may only be intuitive because its a commonly used convention.

Thanks, appreciate the feedback. Although it does make sense to offer a mixed approach I have the feeling that in order to teach players how to play the game it's wise to go with uniform convention and stick with it. I think if the world is less concrete and more abstract this becomes more true.

i once did a person simulator (like The SIMs) with action - object interface. this allowed you to cue up actions that used objects that weren't visible on the screen. so you could cue up move to lake, drink, bathe, fill water skins, and move back to cave actions all at once.

I like the idea with this that there'd be a bit of lateral thinking involved, in that you'd maybe see the world and think about how to manipulate it without maybe worrying about the exact specifics in advance. A more actualized, traditional depiction of the game world would ask the player to gather objects in advance and move through each node of interaction linearly, step by step. While you might say this demands more foresight and planning, it can also be more fiddly and annoying (as in, "oh crap, got to the lake but can't fill the water skin because I forgot it back at the cave!")

i'm now working on a new version of the same game, with a fps type interface. it is largely object-action. select something in the environment, and get a menu of actions for that type of object. if no object is selected, you get a menu of general actions that can be performed anytime, anywhere. this menu is action - object in style (example: make - stone knife).

This approach sounds very promising, especially the way lack of an object triggers the general actions. I notice that as the world becomes more concrete (as with an FPS) it makes more sense to go with object-action in many cases because the world is filled with objects that you can walk right up to. I think this makes the conceptual map of how the game world works in the player's head clearer, but I also imagine technology more stringently limits what can be depicted as you have to script out each action.

I notice that as the world becomes more concrete (as with an FPS) it makes more sense to go with object-action in many cases because the world is filled with objects that you can walk right up to.

actually its all because of the change of point of view of the player interface.

the original version was a "god game" interface like The SIMs, so it was all about issuing "orders" to your cavemen (sims).

a fps interface meant no mouse pointer or onscreen buttons. so another method for initiating actions was needed. enter the "select" key: the spacebar. since the time of doom, the spacebar has been used in fps games to interact with the environment. so it was a natural choice for a fps means of initiating actions. thus the object-action approach used for everything except general actions.

probably a good thing too. object-action splits things up into context sensitive menus. action-object results in one massively huge menu system of possible actions. and it gets even more complex if you try to grey out actions on the menu they can't perform due to location, lack of equipment, etc, because you have to do all those checks while drawing the menus. ik! caveman has over 100 types of action handlers, each of which can handle anywhere from one to over 300 actions. i've never calculated the exact number of actions in the game, but its estimated at over 2000, probably much more. needles to say, wading through a menu system of 2000+ actions would be difficult at best.

I also imagine technology more stringently limits what can be depicted as you have to script out each action

not really. in the original version of the game, i was able to implement over 200 actions with perhaps 12 to 15 generic action animations, such as swim, make, gather, etc. the animations were setup so you could change the objects in the players hands, so you could use the make animation for any make action, just by switching the tools used, and the parts and items laying nearby.