3. Assumptions
Hard West: missing features
Phantom Doctrine:
● “complete” turn-based tactical (~XCOM)
● Lower budget, production value & pricing bar
The main kind of feedback we got for Hard West was basically “not XCOM enough”.
No character development, no long campaign, no base building, no economy, no tech tree.
With Phantom Doctrine we decided to fill in all the blanks left when compared to XCOM.
It seems this assumption was correct - there’s almost no feedback regarding “missing features”.
4. Assumptions
Budget ✓
Time limit ✓
Team size ✓
The budget was enough.
And the time limit was enough.
Our full-size production team could do it.
Alas, these correct assumptions led to a wrong outcome.
5. Assumptions
Budget ✓
Time limit ✓
Team size ✓
Growing the team size within time limit X
There just weren’t enough people in the employee market.
Even though we eventually got to the point where we collected a team that could manage this scope with
satisfying quality, it was too late.
6. Problem
The variables and circumstances changed, but the scope did not.
Due to studio fragmentation, we didn't reestimate and rescope the game when we should have.
All the problems I'm about to talk take root in this key issue.
7. Lesson learned
Re-evaluate the scope frequently.
Even if we did adjust the scope later on, scoping down would break the game structure.
We should have updated the scope, every couple of months at least.
9. What are sidesteps, actually?
Waaay back in the day, in turn-based tacticals, it'd go like this:
there’s a target (T) just around the corner, but your shooter (S) wouldn’t be aware of the danger until you’d
run into them
10. No sidesteps back in the day
In classic XCOM you could unwittingly run straight into the face of an enemy.
And whoever did this first - you, or the enemy - won the duel.
Which isn't very tactical.
11. Modern approach - cover side-step
Hence nowadays, a frequently used solution is a cover-sidestep mechanic
12. When adjacent to cover, visually shooter “sticks” to the wall, peeking out; the target would be revealed - and
shot.
Technically speaking, characters glued to cover edge are treated as if they also occupied the tile to the side
hence: side-stepping.
13. Your typical sidesteps
However that doesn't solve everything - like this.
in this case, you'd see the first RED target, but not the second one - for no good reason.
14. Free sidesteps
So we added free-sidesteps - characters would be considered visible if you could see any adjacent tile.
Given that Phantom Doctrine is predominantly interior based, this solved a lot of cat and mouse and
peekaboo situations.
15. Problems with typical sidesteps
There were a couple of simple cases that convinced us this makes sense: in here, you'd only see the red
target
16. Problems with typical sidesteps
Consider a single tile of hedge.
Hedge can be shot through and isn't cover, so characters don't stick to it, so no side-step.
If this tile was solid cover, sidesteps would kick in, characters would see each other.
17. Problems with typical sidesteps
And this one is especially absurd.
And as far as I know, all these situations can happen in XCOM. Certainly could happen in Hard West and
bothered us.
18. Problem
The problem with this, however, is twofold.
One, if obstacles stack, line of sight is hard to predict.
19. Second, simplified graphs don't accurately depict what happened in the game.
When the layouts, animations and so on were finished: characters don't step out as far, walls are thicker,
camera angle makes it confusing.
20. Here’s how it might end up if all the issues kick in: occasionally characters just shoot through walls.
21. Problem
We tried to use the sidesteps to solve two problems that seemed similar.
The problem was that we had two similar issues - regarding line of sight - and tried to use a system related to
both - sidesteps - to fix them both.
22. Lesson learned
Separate your problems and tackle them
with perfectly fitted solutions.
Elegant design is fine and all, but don’t force it.
Separate your problems and tackle them with perfectly fitted solutions.
Free sidesteps were to fix the problem of seeking out enemies in dense environment, which was good,
but heavily affected visuals and firefights - which was bad.
23. Example fix
1) free sidesteps make enemy SENSED
2) free sidesteps don’t allow for Line of Fire
3) SENSED characters are visible (grayed out) but non-interactive directly
This gives you tactical information, and doesn't make the Line of Sight so confusing.
25. How it works in XCom
Chance to hit = AIM - DEFENSE - COVER
Roll below chance to hit = shot connects
26. How it works in Hard West
Chance to hit = AIM - DEFENSE - COVER
Characters have LUCK
If Chance to hit < target LUCK: reduce LUCK
Else: shot connects, LUCK regenerates
Instead of having full random, there's a Mana-like resource: LUCK
You keep damaging LUCK, until you hit, deal damage, LUCK goes back up.
27. How it works in Phantom Doctrine
Characters have AWARENESS
If target has AWARENESS, graze (deal less damage), reduce AWARENESS
Else: shot connects
Awareness regenerates passively
main changes compared to hard west: LUCK renamed to AWARENESS
dealing some damage on miss with most weapons
AWARENESS regenerates on it's own (don't have to get hit)
28. Feedback
Players who really got it, appreciated it
Players who didn’t were quite mad
Appreciated by power users
Hated by beginners
I actually had an argument with a gamer who stated Hard West DID have RNG because it had percentage
Chance to Hit.
29. The problem:
The impending doom of no RNG
is no good for the faint of heart.
Players who didn't grasp the system felt cheated. They felt doomed.
The most frequent comment was ""enemies hit you with every shot"", neglecting the fact that you hit enemies
with every shot just as well.
The system is fully predictable and controllable, which makes the game fun when you're in control.
At the same time the game will CRUSH you inevitably when you're not.
Inevitable failure destroys hope.
And however bad you are, with random numbers, you always have some measure of hope.
30. The big fat lie in Hard West UI
Hard West straight out LIED about this and it was better received.
This was a simplification good enough for beginner players.
Expert players saw right through and appreciate the clockwork behind it.
31. Lesson learned
Conceal hardcore details
for hardcore gamers’ eyes only.
If your feature will be appreciated by hardcore players, and is aimed at them - hide it for them under the hood.
They'll find it.
32. Lesson learned
Weird mechanics are OK -
in a weird game.
Should be also noted that Hard West is a weird west game. Cowboys fight alongside demons, player strikes
deals with the devil and death himself is the narrator.
Hard West was very ballsy in many aspects and there were all sorts of weird things that frequently happened,
so the luck system didn’t stand out.
Phantom Doctrine is far more realistic.
34. Varied mission types
● Enemy scout
● Ambush
● Conspiracy plot
● Informer rescue
● Conspiracy cell
● Story missions - each unique
We have a variety of mission types.
And we put in extra effort to make sure they are different, and all comparative playtests proved that correct.
For example: in PD you don't get to choose how many agents you can take on a mission,
because that influences how mission feels the most.
35. Feedback
Same mission over and over
Backlash for repetition
Note: Almost no backlash regarding how many agents can be taken on a mission.
36. World map gameplay
Here's where this comes from: there's this World map we have, where all missions start.
Enemy agents travel around the world, do nasty shit, you go there to stop them.
These are tied to tactical missions:
enemy scout - neutralize enemy scout mission
informer assassination - VIP rescue mission etc.
37. Problem
Balance in world map imbalanced mission type balance.
World map event Chance Conditions
Balance for
tactical variety
Ambush 11% lone agent ~20%
Conspiracy plot 35% active time limit ~20%
Informer rescue 24% active informer ~20%
Conspiracy cell - lots of conditions ~20%
Enemy scout 30% ~20%
Balance in world map imbalanced mission type balance. When I balanced the frequency of occurence of
world map missions - which is basically a time & resources balance gameplay loop - I broke the balance of
tactical variety.
Ideally the tactical mission types would similarly frequent; because of how world map balance operates, and
what conditions are required, this has been severely skewed the most common mission played is ""enemy
scout""
38. Balance key aspects of game
in separation from each other.
Lesson learned
Tactical missions and world map missions should be unrelated.
Each of those should have been balanced in it's own regard and then tied.
The overall atmosphere would suffer, but gameplay would have been better.
40. Assumption
Possibly endless at any point:
● Endless number of world map missions
● Endless number of tactical missions
● No pressure to proceed
We wanted the game to be possibly endless at any point.
You don't have to proceed unless you want to.
You can collect loot, gather secrets, level up your agents; and if you feel ready, you go on.
No pressure.
41. Feedback
Not enough levels
Not enough everything
100h campaign runs
Boring grind in repetitive context.
100h campaign runs, while our estimate was 40h (!).
42. Big gameplay loop
Story mission N+1
Story mission N
Generated side-mission
Get random loot
Get random
unlocks
Campaign stage N
Here's how the loop works:
Game is split into campaign stages; campaign stage is the open part between two story missions, where
players take on generated missions.
For each stage there's a finite pool of stuff they can find at random, but players delay going towards a story
missions for very long periods of time.
43. ● Finite loot to collect
● Finite secrets to gather
● No feedback when you got it all and
it’s time to proceed
Problem
Players can play infinite number of missions, with finite number of random loot drops and random unlocks;
but they don't know when they got it all and can get off the ferris wheel and onto the next challenge.
44. Clearly communicate the threshold
between build-up and grind.
Lesson learned
Communicate player progress clearly.
Clear buildup: clear message (popup?) that you collected it all and should move on.
Game doesn't tell them when it's OK to proceed to story mission and push the campaign forward.
46. What are
these?
We have this cool features that give meaningful context to character customization.
If you meet some requirements, you can assign agents to wear disguises.
It’s automatic - you just turn it on, and agent switches clothes to blend in.
48. Problem
Out of fear of repetition, we did not fully deliver the experience.
We did not create obstacles which players can overcome.
Out of fear of this being too mundane, we did not force customization.
49. Reference
Like in eg. Sherlock Holmes, we could just grant disguise status after player changed a couple of things and
give them a GREEN LIGHT when they did.
50. Plan how many times a feature
should be used and design accordingly.
Lesson learned
Should have calculated how many times during the campaign it will happen and balance with this
assumption.
Mundane repetition happens only if you allow it, and you can design and balance to avoid it.
52. What is it
You collect bits of information, files, photos, etc.
You make the connections to figure out the key information.
53. Feedback
1. It’s just like the movies!
2. Becomes repetitive and boring.
(after 16-20h)
On average, about 20 hours into game, players had too much of it and could see right through the illusion -
because it is all smoke and mirrors. (like all gameplay systems)
54. We put in a lot of effort to turn this prototype...
58. We feared that players will grow tired of it, so we
made it possible to automate and skip the whole thing.
And when you look at achievement stats - at least 91% did use the skip mechanic.
We relied on the capability to skip and left it boring by default.
What went wrong
1. Assumed players will get tired of it
2. Created a way to automate/skip
3. Relied on the automation
at least 91% used
automation
59. A way for gamers to avoid feature is fine.
Relying on it is not.
Lesson learned
Do not rely on opting-out of your features.
It's a trap!
I relied on the safety net and didn’t balance properly.
60. Plan how many times a feature
should be used and design accordingly.
Lesson learned (again)
Should have designed it for X uses.
Should have planned it out: how many different actions of various kinds will the player carry out ideally
encourage and reward them to stick to your balance script.
Shouldn’t have relied on the fact that it could be generated ad infinitum.
Shouldn’t have bothered with the second playthrough either.
62. Feedback
Systems don't plateau with significant moments.
Systems don't connect. Exciting things that were expected to happen - happen rarely.
Systems don't come together to create tension and obstacles and risk.
63. Reason
Systems in PD can create climactic moments.
But:
1) they don't always do
2) even when they do - players miss it
Systems in PD connect and collide and CAN create climactic moments but
1) they don't always do
2) if your systems don't build to climax points as clear and as significant as mediocre scripted content, the
structure falls short
64. Example
Tactical recon: various bonuses at the same time;
● Reveal Fog of War
● Allow for disguises
● Allow for support powers
● Mark CCTV
● Mark loot containers
Should give 1-2 at random at communicate that.
Players grow accustomed to these bonuses and the only change is when they don't get to do recon.
Should have picked 2 at random and notify the player about the results in a very clear manner.
This would give the player 6 clearly different variations of game states and more clarity.
65. A feature will not be appreciated
if THE GAME doesn’t appreciate it.
Lesson learned
A lot of systems that aren't communicated well enough: aren't foreshadowed, aren't underlined and aren't
appreciated clearly and openly BY THE GAME so that the player knows it.
67. What are these
Before the mission, you assign agents to support powers (sniper, etc), you place them outside of mission
area. Gameplay-wise - these are just directional global cast abilities with a nice cutscene.
68. Went well
● Just enough
● Simple directional
cast abilities
● Cheap cutscenes
69. Feature is bound to be good if:
final version = prototype + juice
Lesson learned
The prototype was very effective and perfectly accurate: the final effect was exactly the same as that first
trivial prototype plus all the bells and whistles.
There was practically no iteration in gameplay.
Too often prototypes we all make have very little with final result, and the more they differ, the more messy
the production is bound to be.
71. Simultaneous
orders
Assumption: provide a change in turn-based monotony.
Idea: give player an option to act out orders simultaneously.
Initial design: give orders - any orders - to your agents, and then have them perform them all at once.
This allowed virtually anything to be carried out.
72. Problem
Feature designed to do everything was:
1) Not particularly useful for anything
2) Pretty much impossible to do
No one ever used it, because it was a swiss army knife that wasn’t particularly great at anything.
Also the idea to just wreck all turn-based assumptions on which all systems relied was a suicide mission.
73. Solution
Feature got FOCUSED:
1) agents enter a room and shoot all at once
2) one click
We narrowed down the usage of this and called it Breach.
As in “all agents Breach this given room all at once”.
74. Ideally your USP should be experienced
at the push of a single button.
Lesson learned
Ask yourself this: can your USP be shown and more importantly - experienced - at the push of a button?
And it worked so much better, was understood at a glance and above all COULD BE USED with a click.
76. Problem
Easy difficulty too complex.
Hard difficulty too straightforward.
The game is overly complex on easy and grows quite simple on hard,
because all features are assumed to serve all groups of players.
77. a video well worth your time
https://www.youtube.com/watch?v=YISKcRDcDJg
long story short: you should organize your UI into clear levels of information and prioritize size and order on
screen based on that
78. For each feature, define if it’s required on
easy, medium or hard.
Organize GUI and complexity
accordingly.
Lesson learned
I'd divide features into easy, medium, hard.
Organize information on the screen to easy features on first read, medium features on second etc.
I’d balance the game that easy only requires you to effectively use the easy features, and so on.
79. Example
Easy - first glance GUI Medium - smaller Hard - hidden
Flanking Abilities Weapon customization
For example: flanking would be the simpliest and most visible mechanic, abilities would be medium and less
visible, and gun tweaks would be hard and quite hidden.
And I would balance difficulty level obstacles according to what I expect the player to use efficiently.
80. Make rules available.
Litmus test: I forgot everything.
Can I still play?
Lesson learned
You don't have to tutorial everything, but every rule governing the outcome of player action should be
available (eg. an ingame manual).
Assume the player forgot EVERYTHING. Can they effectively play? Can they catch on?
Can they get to the level of understanding they were at before?"
81. 1. Re-evaluate the scope frequently.
2. Separate your problems and tackle them
with perfectly fitted solutions.
Don't force elegant design.
1. Conceal hardcore details for hardcore
gamers’ eyes only.
2. Weird mechanics are OK - in a weird
game.
3. Balance key aspects of game in separation
from each other.
4. Clearly communicate the threshold
End. 7. Plan how many times a feature should be
used and design accordingly.
8. A way for gamers to avoid feature is fine.
Relying on it is not.
9. A feature will not be appreciated if THE
GAME doesn’t appreciate it.
10. Feature is bound to be good if: final version
= prototype + juice
11. Ideally your USP should be experienced at
the push of a single button.
12. For each feature, define if it’s required on
easy, medium or hard.
Organize GUI and complexity
accordingly.
7. Make rules available.
Litmus test: I forgot everything. Can I
still play?
@illusionGD
szymczak.kacper@gmail.com