Difficulty in Stealth Games

Iuliu-Cosmin Oniscu
6 min readNov 28, 2020


As you probably know when designing a video game, you need to determine how the player interacts with the environment and how the environment pushes back.

Depending on the type of game, the player can either push or pull on the systems threads leading to interesting results:

  • Actions games ask the player to push the system and in return the system pushes back. This leads to something not completely different then a boxing match. Last man standing takes the prize.
  • Stealth games ask the player to pull on the system for the system to react in interesting ways. You pull on the strings to facilitate avoiding the confrontation. You are inciting the game to play itself. You are a disruptor.

So, the question is how do you determine difficulty factors in stealth games?

One simple answer is that that you can play around with the parameters and simply make them harder to deal it:

  • AI can detect you faster (Sight, Hearing, Smarter search patterns)
  • Make soft cover and stealth opportunities scarcer or harder to deal with
  • Shrink Window of opportunity times
  • Etc.

However, looking at all of these and playing games that have been doing these for years I cannot stop thinking that this approach does nothing to improve the actual experience.

In fact, the more you increase difficulty → and therefor gimp the stealth player the harder it becomes for the player to play the stealth game.

This leads to him either being severally pushed (as he should be in a stealth video game) but for the wrong reasons:

  • Player does not fail at playing safe, he fails at: Timing, Reflexes, etc.
  • Player failure should come from: Poor Scouting, Poor planning, Poor Execution.

In traditional action games difficulty of encounters is usually determined by associating a degree of success to amount of health he still has at the end of the encounter. Harder encounters cause more stress on the players health, leading to situations where the player needs to:

  • Either improve his Physical skills: Aiming, Dodging, Shooting.
  • Improvise his way through the encounter to win the fight.

The balance of power is kept in equilibrium by physical player skills.

In stealth games the there is no balance of power, or more precisely it is scaled in favor of the challenge. The difference is that the system starts in a passive state giving the player a large degree of initial power.

Success is determined by:

  • A binary check → Has the player completed the encounter without being detected? Yes or No!
  • How much did the player engage with the systems to complete the encounter? A little, Moderate, A lot, Not at all?

So, we can extrapolate that the difficulty of an encounter can be raised or lowered based the degree of systemic interactions that the player perceives as available in the encounter.

Player success rate — determined by his degree of interaction with the systems in the game.

But does it really? Or does it just lead to a variety factor that can dictate how the encounter will change based on the conditions that are available to you at run time?

Having a city patrol in Assassins Creed 3 is fine.

Giving you an objective to go past the patrol is fine.

Giving you several ways to go around the patrol that you can identify is great (you get more options)

Not giving you any and pushing you into direct confrontation is not great, and leads letting the combat system to dictate the encounter, and when thinking about the stealth scenario what this leads to is tipping the difficulty in the favor of the encounter, thus the player fails stealth and needs to be punished.

I guess the paradigm is all about how you tailor the encounter in such a way that the player has:

  • A large variety of valid options
  • The Options are designed to work in tandem with AI behavior, so having a dynamic/manual way of modifying the game difficulty can lead to some sort of dissonance between LD intent and encounter behavioral patterns. (attempting to balance player options to encounter difficulty factors)

I am by no means dismissing the importance of skills such as Timing and Reflexes in stealth games, however it feels to my like they are not absolutely relevant when considering the : Scout, Plan and Execute paradigm.

  • Scouting should be reliable
  • Planning should lead to good solutions

· The execution should happen swiftly, unless the previous 2 steps have been borked. However, if the execution fails the player should be able to retreat, re-asses and re-iterate on the plan. The chaos caused by the systems should be interesting to experience.

This leads me to Assassins Creed Valhalla.

I started playing the game the game on the highest difficulty levels possible and this led to the following experience to me:

  • It is super hard to figure out what is going on
  • Keeping track of enemy forces is super tedious
  • Stealth fails super-fast and you spend most of the game in open combat. I know what the theme of the game is, that is not the point.
  • It also leads to a lot more looking/running around

And this is a lot different that how AC Odyssey played out for me:

  • Always Scouting
  • Always Marking and Planning
  • Looking for thought encounters (mercenaries) to overwhelm using my parkour and assassination skills.
  • Always having a clear idea where things are when camps are cleared.
  • Moving fast and hitting hard.

My point I guess is that upscaling the difficulty of the stealth/combat/exploration mechanics isn’t necessarily going to improve your game experience.

On example of good game difficulties in stealth games are probably from the Thief Games:

  • The higher the difficulty, the more you need to steal
  • The AI behavior does not change, or at least it does not change in a perceivable way
  • You get more stuff to do if you decide to bump up the difficulty.

So, you do not get a higher systemic difficulty per se, instead you get more explored to the systems themselves → Since the most difficult things to get are in the areas that are:

  • Harder to Find
  • Better defended.

In these kinds of situation I could even argue that variety/quantity of options are a lot more important here then difficulty incrementation.



Iuliu-Cosmin Oniscu

As a level designer I am not a creator, I am a facilitator. Senior Open World Designer. #Leveldesign #Open World #Design @notimetoulose.