The second fundamental question crucial to designing enemies for an action game is:
2. How will this enemy counter the player’s first-tier tactics?
Action games give the player a hammer. It looks like a broadsword or an assault rifle, but it’s still a hammer. And when you have a hammer, every enemy looks like a nail. A nail doesn’t have to do anything complicated to look smart; all it has to do is avoid getting hammered right away! If an enemy forces the player to reach into their toolbox for a second option, they’ll show that they are clever, and the player will feel clever, too.
Good AI
The obvious way to counter the player’s basic tactics is through simple invulnerability. “That armor is too strong for blasters.” It’s crucial that this invulnerability is communicated clearly, through the look of the enemy, how they respond to the player’s attack, and any supporting effects or dialog. If the player doesn’t realize they are not being effective — or is misled into thinking they have done damage when they haven’t — the enemy will just seem poorly tuned.
- Carries a shield that blocks bullets, but any other kind of attack causes them to drop their shield and expose themselves to fire
- Cannot be killed with melee attacks and must be thrown into an environmental hazard
- Bullets bounce off the vehicle, but you can snipe the driver out
A more subtle way to foil the player’s main method of attacking is to preemptively take action to prevent it or require another action be taken before the attack to make it a two-step process. These enemies appear intelligent, but they still allow the player to user their (hopefully) satisfying primary attack.
- Takes cover behind an object, requiring the player to flank or drive them out of cover
- Wears a full-body energy shield that must be taken down before they can be damaged by normal bullets
- Blocks every sword swing, but is stunned if their own swing is blocked
It is also possible for an enemy to behave in such a way that it is more difficult for the player to execute on their first-tier attack. This can be difficult to tune for multiple difficulty levels, but if the counter-strategy only works on the their main attack, then players of any skill level should be able to cope by changing their tactical choices instead of having to execute beyond their skill level.
- Flees out of melee range faster than the player can pursue, but has no defense against ranged attacks
- Keeps up a constant barrage of fire, forcing the player to hide, but unable to get away from a grenade thrown from behind cover
Again, the idea is not to make an enemy that is difficult or frustrating, or even especially sophisticated, but that has a clear counter for the player’s primary attack. This will require the player to stretch, tactically, and figure out their alternatives, but any amount of experimentation should be rewarded by a quick victory.
Rewarding Play 
A good point to be sure, but I find that a lot of times this kind of thinking leads to a more frustrating experience – especially in action-adventure games leaning a bit more on the adventure part. An enemy is introduced that require you to use a new skill other than the good old sword (assault rifle, plasma gun … you’re right, let’s call it hammer), and while this is interesting the first time it is no different from a “regular” enemy once the puzzle is solved. Instead, you force the player to stop and change their equipment every time this enemy is introduced, and opt to use techniques that were not as well tuned for enjoyment as the hammer in the first place – be it that less design time is given to the secondary tools, that the new strategies clash with what the player already learned about the game or whatnot. So while a good game certainly needs to challenge the player intellectually I would issue a caution against forcing the player to discard his old strategies without thinking it through.
Well, if you have to “stop and change equipment” then you have much bigger game design problems. All I am saying is that every enemy character ought to have an intelligent strategy for dealing with the player’s primary attack. If an enemy shows up and says “I never expected the hero to aim his gun at me and pull the trigger” defeating them isn’t going to feel like much of an accomplishment — but many game enemies do exactly that and are surprised that their combat feels repetitive.
I feel what Jaime was trying to get at was the context that came with making enemies in a sandbox style game where the puzzles are defeating the different enemies on the battlefield. This sort of game design allows for different puzzles to be made through the use of varied enemy types. Tools are not swapped out to defeat these enemies in this sort of game, only assigned to different buttons, which is something many adventure games can and have implemented successfully too.
On the other hand, a different tool may also be regarded as the same piece of equipment with a secondary use. Why try and kill an Elite using the same single fire shots that work on Grunts when charging the Plasma Pistol would work so much better? It seems if the developer has a problem with not putting time into their secondary tool, it’s not the design of the enemy that’s broken, it’s the tool itself.
Oh, definitely. Maybe it was just a reaction out of habit on my part, the “Design in Detail” posts had a lot more breakdowns of what not to do and why not to do it while this one had a more positive tone. I agree that the problem is not in the idea of making the enemies smart, but it would make the existing problem worse – hence the post. That’s a bit off-topic though. In fact, starting a design discussion about details with the premise that the core design is flawed is probably not very rewarding to begin with (unless trying to salvage a doomed project in a short timeframe).
Very well put about enemies not expecting to be attacked by the way!
Not at all. I find many designers fall into a “prevent defense” strategy where they just to try keep a game from getting worse as it gets more complex. It’s hard not to over-focus on what might go wrong when each decision has unpredictable ripple effects. It makes us suspicious of suggestions that are too similar to something we’ve seen in another game that didn’t work out. We just have to take a deep breath and proceed cautiously to avoid making the exact same mistakes. In the end, we have control over the implementation and can hopefully steer the design away from those common pitfalls.