Randy Smith of Tiger Style Games discusses developing an iOS game called "Octobot Descent" that involved cave exploration and ecosystem gameplay. After an initial playable, feedback indicated the gameplay involved too many mechanics without meaning or progress. The focus shifted to emergent gameplay from interactions between lifeforms. A third playable using this approach led to more engaging interactions and a mid-level that allowed building your own level. The key lessons were to prioritize interaction density and ensure nuanced player input leads to meaningful and unpredictable but clear results.
16. • Continuously make your way
down, down, deeper and deeper.
• You can rock climb, squeeze
through small openings, limited
rope resource for lowering
yourself safely, etc.
• Avoid death by falling, dangerous
spikes, rushing water, etc..
• Use finger to orient head lamp so
you can see upcoming dangers
better.
• Crumbling rock surfaces, ways to
roll out of long falls to reduce
damage, etc..
• Collect bonus items on the way.
• Like Doodle Jump gameplay in
reverse. Move fast! Go deep!
Keep moving! Quick!
• Or not? Timed mode, Survival
mode, Collection/Points mode?
• Randomly generated cave terrain
for max replayability?
• Different phases of cave terrain
and style of exploration /
features? Eg – get deep enough
and it’s darker, or more crumbly
surfaces, or more water, etc..
22. Still a personal, rock climbing
/ cave diving game.
Survival. Isolation. Human vulnerability. Etc.
Just: on Mars now.
NOT: shooting, RPG, treasure collecting, etc.
23. REFLECTIONS OF EARTH
I like alien creatures as a way to explore
the startling diversity on earth.
Would like to portray a strongly-plausible
ecosystem, inspired by facts of the real
universe.
hugely
inspirational
childhood
book
24. ECOSYSTEM
In addition to moving,
exploring, and finding
food, water, and air for
survival, a big part of the
player’s actions involve
bringing life and activity
back to the cave under
their own design.
It’s a thin “SimLife” type
system on top of a
platformer.
25. JETPACK
Triple Jump – Player can’t fly
but can fire jets repeatedly to
jump higher and further.
Low Gravity – Allows the
player to jump further and fall
slower.
Reduced Falling Damage –
Jetpack automatically fires to
slow impact. This encourages
bolder player exploration.
Fuel – Player can learn how to
harvest fuel from elements in
the cave environment.
(info)
26. HEADLAMP
Objects in the environment
respond to the headlamp.
Eg:
• Crab-like creatures scuttle
away.
• Plant-like objects grow.
• Crystal growths collect glow
from it, then emit slowly over
time.
• Bat-like creatures are
disturbed from sleep by it, then
repelled by it.
50. CUP AND BALL PLANT
Repeatedly spits a seed
up and catches it.
Reason = This is an
easy trajectory for the
player to avoid flying
into, or to hit with a
projectile on purpose,
etc..
Player might knock
seed away
deliberately or
accidentally, or it
might happen
emergently (eg –
jumping crab hits it).
When it loses seed, plant closes up
to produce new seed. During this it
has an important mid-level impact,
eg: consumes lots of Nitrogen, or
stops coloring terrain in its radius.
Player may or may not want this,
depending on circumstances.
You’ll see in some of the following slides
that I use this idea a lot, of a plant that
changes state to fix something about itself,
or grow a new thing, etc.. It feels
“ecosystem-y” and it works well to connect
mid and low level play.
Variations:
- Plant freaks out and trashes
around when it loses seed.
- Seed goes into player’s inventory
51. CAVE FISHER Cave fishers are naturally-occurring lifeforms, but what if the player
can collect their seeds/eggs and shoot them to position new ones?
Then cave fisher ammo = sets an environmental trap.
Cave fisher on ground = eats creatures that walk into it?
Or floats its stick thread upward?
Reason = one tool is both land mine AND ceiling trap
52. PESTS These live in an indestructible hive until something that triggers
them to come out and launch an attack on plants you want to
protect.
You can prepare traps for their attack, shoot them out of mid-air,
etc..
Player’s plants that
need protection
Mid-level: Maybe the player’s plants are an “oxygen farm,”
meaning that once you plant 5 oxygen plants, you’re producing tons
of oxygen into the environment or something similarly helpful to
your mid-level goals.
You learn that once you hit 5 plants, though, the hive opens, so you
better be prepared to defend your oxygen farm.
53. Why Combat Works
• High stakes drama (life or death)
• Clarity on
– Start, finish
– Win / lose
– Intermediate progress
54. Why Combat Works
• Nuanced input is MEANINGFUL
– Which actions
– Specific timing
– Continuous analog micro-adjustments
Depth and mastery
“Meaningful” = Contributes toward a
result player cares about.
e.g. – success / failure
55. Combat, Racing, Platforming
• High stakes drama
• Clarity on
– Start, finish
– Win / lose
– Intermediate progress
• Nuanced input is meaningful
– Which actions
– Specific timing
– Continuous analog micro-adjustments
• Depth and mastery
56. Thief Stealth
• High stakes drama
• Clarity on
– Start, finish
– Win / lose
– Intermediate progress
• Nuanced input is meaningful
– Which actions
– Specific timing
– Continuous analog micro-adjustments
• Depth and mastery
57. • High stakes drama
• Clarity on
– Start, finish
– Win / lose
– Intermediate progress
• Nuanced input is meaningful
– Which actions
– Specific timing
– Continuous analog micro-adjustments
• Depth and mastery
High
Level
Mid Level
59. Action Gameplay
• Unpredictable / Uncertain
• Results are Acceptable / Clear
Player makes choice to take action
Skill at execution / Current context
Unpredictable but Acceptable result
60. Grow Plant With Headlamp
Player choice
Skill doesn’t matter
Predictable result
61. Throw Seeds
Skill and context matter
Unpredictable but Acceptable result
71. Add More Meaning
To Collisions
• Why it worked
– Player has nuanced input into physics
– Important game objects were physical
– Physics is unpredictable but acceptable
– Collisions are a clear event
– Player cares about meaning (win/lose)
• Motivates mastery
– Dodge acid
– Hit bat
– Catch seed
– Herd crab into fisher
72. Player
input
Events Results
Nuanced
Analog
Time-based
Simulated
Unpredictable
Acceptable
Success
Failure
Meaning
81. Red plants only in
this terrain
Green plants only
in this Wterrahin y does it matter
WHERE you plant?
82. Switch to One Focus
• NOT Caving – darkness, movement
• NOT Gardening – tending to plants
• YES Ecosystem – interactions between
lifeforms
• Innovation:
– Can’t support 3 games
– Tomb Raider, Lunar Lander not Innovative
– Made it easier to play Ecosystem
83. Be Wary Of
• “Doing Your Chores”
• Complex Design
• Focus on Mid-Level
84. Useful Lenses
• Interaction Density
• Nuanced Input Meaningful Results
• Mastery More Success
• Unpredictable but Clear & Acceptable
– Foundation of a Real Simulation
• Noodling vs. Anchoring
More recently, formed this studio where I’m the creative director
We released this game in 2009, it did pretty well
Here today to talk about this game, released March for iOS and recently PC / Android.
Who has played WM?
Because not everyone knows it, watch trailer.
Action adventure game set in a cave on Mars interact with alien ecosystem.
Cast of alien plants and creatures you interact with.
Player collects and throws seeds to interact.
This is a post-mortem on how we designed a new type of gameplay.
Young indie studio - console/AAA background.
First game was critical/commercial success. Won awards.
2nd game = wanted to level up, “even better” STILL MEETING STUDIO GOALS (soph anx = nervous about not living up to our first game)
Bolt = a) realized Spider was (relatively) bolt of lighting,
& b) that normally innovation = false starts, experiments, backtracking, effort.
But ultimately success! Invent something new, proud of it. Learned a lot about doing it faster, better.
“without Guns” – push ourselves into frontiers by not relying on violent conflict resolution
Artistic meaning – not just disposable, some lasting value.
native – both the device hardware and appropriate to the audience
Bring concepts like emergent gameplay and environmental storytelling to a casual audience
Chronological by phase
First project phase = concept
In concept for 4 months, considering different ideas
Here are 3 docs that factored into WM
Tiger head = a badge to indicate you’re looking at our internal documentation.
Often lots of words, you don’t need to read them all.
Concept called World Creator.
Draw a line.
Fills in to become a mountain.
Rain comes and creates lakes and waterfalls.
(we didn’t build this)
Key point = creation instead of destruction
creating things that would grow and have a life of their own
navigate an IK-driven robot around a cave environment.
(only has 6 arms)
First time we thought about caves
This character later appeared in the game as OCTO, the robot who explored the cave first but got lost
This is a total aside – this article came out very recently, and it’s about using
A spider-like robot to explore lava tubes on mars just IS octo
It had been months - then DESCENT treatment
This is the one the team latched onto
Present day earth
Simple game of climbing ever deeper into a cave.
No story. Just action gameplay and essence of character and adventure
As with most videos, this is early proto, not what we shipped
Mechanics = Movement, Headlamp, Rope
Caving game!
Were struggling, so: ran thought experiment:
what if we took our cave game and set it in space?
Took about 3 minutes to decide to pursue it.
So we had ideas about creating things. And now ideas about lifeforms.
Spawned a couple phrases that would drive our design process.
HOWEVER we didn’t know what these meant yet.
It was a concept in isolation from specific examples of gameplay.
This is where we had to innovate.
And often where things go wrong, eg – don’t know how to capture the essence of Barbie, so just reskin Super Mario Bros.
Next phase – working toward first playable demo of Mars Descent
Big inspiration = Our Universe
Mostly factual about solar system
Alien lifeforms had diets, reproduction, grounded in their unearthly environments
You can see ecosystem increasingly being a focus.
Not that we knew what the gameplay would be yet.
Still have Headlamp
Now used to grow plants, repel alien bats
Learning movement basics while being exposed to the story via environmental gameplay
Kind of a more raw version of an ecosystem game, very survival / cavey
Sent playtest builds to external people
Start by analyzing player mechanics
Some more problematic than others
Why did we have oxygen resource?
Time pressure = motivate risk
Collection = reward exploration
But players didn’t feel safe exploring unless they knew where next meal was coming from
Realistic = difficult
Gave players something to think about, but was it the right something?
Why we liked headlamp
Made a tiny screen even more difficult because filled it with black pixels
Inventory seemed more hardcore
But counter-intuitively, it was easier and less fussy, not more so
gave you more chances to hit with a seed, less consequence for screwing up, less backtracking.
All of these things together: complex, fussy, lots of player attention for even simple actions, lots of telling the player NO.
You missed with that seed, you’re running out of o2, etc.
Developed this design lens
Meaningful choice in the Sid Meir sense
Interaction density = how frequently these happen (especially ones that really matter.)
So it’s a ratio, let’s see some examples
It might be interesting to watch, but if you only touched the screen once, that’s low density.
few points where it might become varied or dynamic.
One decision, minutes of rote execution.
Plants grow to maturity automatically when they land in the right terrain.
Maximum density = meaningful actions can happen as quickly as you can think of them
Line of thought raised red flag about design dogmas.
Obvious thing to say when designing a shooter, for example – “make everything go faster” but doing what all shooters do <> how you innovate.
Expansive, deliberately paced caving game – not sure it’s a bad idea
But we decided it was a bad match for the iOS, might work on the xbox.
So that was the feedback from 1st playable. Moved on..
2nd playable addressed some of these concerns, as we’ll see in a video pretty soon.
AND we Needed to focus more on ecosystem gameplay
had been dodging it
Initial focus on mid-level, ie – low level = grow plants. mid level = goal for low level actions. Why do you grow plants?
There’s some kind of barrier and to open it you have to take the actions it tells you.
How is this better? This is like being told which chores to do.
This idea = there are red and green plants. When you plant one, the nearby terrain transforms and you can only plant similar plants there.
Makes use of terrain, some spots have more neighbors than others.
Answered the question - why does the shape of the cave matter? Why is it better to plant one place than another?
Wanted planting decisions to matter, to be less rote.
This idea = plants have relationships to other plants.
Eg – light plants should be planted to shine on your o2 plants.
Water plants are good but must be planted near explosive plants which are dangerous.
Why not remove player entirely? What was the player contributing other than UI complexity?
Remove player = very touch native, very iOS friendly
But, no, we wanted personal.
Point = were really thinking broadly about the mid-level
So we had all these possible mid levels, Went with just O2/CO2
Must raise biomass high enough, then balance the parameters
Dropped anchor on a specific decision
innovation tends to feel very floaty, lots of possibilities
Noodle = tweak one parameter at a time, try different stuff.
Anchor = make decisions, stick to them, work toward a single holistic design
Seeing some promising results, but game is still not what we want it to be, not fun yet
Mid-level structure didn’t solve our problems.
Had to shift our focus from mid-level to low-level gameplay
Here’s an idea that you can plant your own cave fishers as traps
Throw them on the floors and ceilings
And that there’s a hive of pest creatures that eventually opens up and comes to eat your oxygen plants
So you better have prepared well and fend them off
Interesting, but very complex – and not very flexible, mostly does this one thing
Instead, we took a step back..
Formulate theories about low level action gameplay
not one grand unified theory, but a set of related ideas
What makes combat such an appealing interaction?
Lot of reasons, starting with these…
Another reason = how input is used. Nuanced / analog controls whether you take a corner correctly in a racing game and how correct.
Which specific attack you select in fighting game, when you launch it.
This kind of nuance creates depth and opportunity for mastery.
Meaningful = you can draw a straight line from the player’s input through your game systems, in code, to the outcome.
Instead of success/failure could be, say, a particular garden of their own design
Or an experiment to see what happens when you hit an alien creature with a particular seed type
Great action genres tend to have these types of properties – combat, racing, platforming
And thief stealth systems were designed with some of this stuff in mind - Thief stealth was innovative,
Eg – the reasons guards talk all the time
“I know you’re out there” is to give you feedback about how you’re doing
“Damn I lost him” is clarity that the encounter is over
When we applied this lens to WM.
Eg – drama came from story
- clarity on encounters came from levels, eg- win = open the airlock
But we could use the bottom half as a lens to analyze our gameplay
Another lens to analyze action gameplay – it has these properties
Outcome is unpredictable, but acceptable.
So were looking for a sequence like on the bottom.
Eg – throw seed toward FT, context = falling stalactite, skill = miss stalactite but hit FT requires nuanced timing and control,
Didn’t see a lot of this in the videos
Throwing seeds doesn’t have these problems – what’s the difference?
Difference = built on the foundation of an actual simulation, physics system
Real simulations are:
Super complicated, when 2 objects collide they transfer all this data about momentum and speed and direction, etc..
But also very intuitive because players have spent their whole lives learning the rule.
All of these simulations: physics, fire propagation, fluid dynamics, traffic
very complicated interaction that players already understand
Therefore unpredictable, but acceptable.
ASIDE: this is why successful gameplay is always about things, not people – we don’t have equally strong simulations for human interactions, they are subjective and thin, whereas physical sims are objective and deep
Because these actions are arbitrated by the physics sim:
Player throws a seed at FT
Stalactite falls, might hit a crab scuttling over to eat a seed
Acquire desirable properties of unpredictable but acceptable
Brings us back to this concept.
counter intuitive that removing player actions would make a game better
the real issue is the qualities those actions have.
Prune actions without these qualities, or give them these qualities
For Waking Mars specifically -
Key for making our gameplay better = add more meaning when objects collide
Scrutinize any time a collision happens: can we produce a result the player might care about?
There are bat creatures who fly around carrying seeds.
Previously if they were hit by a seed, nothing happened which seemed vacant.
Now they would drop them.
Previously nothing happened when the player got close to a crab who was eating a seed,
Now that was hooked up to the crab dropping the seed and running away
Drop the seed = back into physic sim, can roll downhill, land in FT
We introduced new objects just to create collision meaning.
So for us, focusing on collisions turned out to be the way to bring all these lessons together.
Seems like kind of a simple solution.
But: the player has the right kind of input and control
Also the seeds, lifeforms, fertile terrrain, etc., were all physical
The physics is the right kind of arbitrator
Collisions in particular are a very clear physical events (we also did some proximity)
And we could create meaning the player cared about.
As a result, it gave players tasks to attempt to master – dodge acid drips, hit a bat in mid-air, catch a seed before it plants itself, chase a crab into a tentacle
Here’s another way to look at the same thing.
If this is a generic game.
Give the left two the qualities that action games have
If you’re getting this vibe, it’s something we were thinking about from the beginning.
Emergent gameplay could be and has been its own whole talk.
So let’s just look at an example of emergence in our game.
Projectile is in mid-air. Maybe by player, maybe by plant. Doesn’t matter.
Hits a bat carrying a seed. Bat drops seed, thus keeping the causal chain going.
Lands in FT and start to plant.
Crab is nearby, maybe because of something player did recently (same as the projectile, bat)
Starts to eat the seed.
Is this good or bad? Depends on context, player’s goals. If wanted to plant seed, bad. If grow crabs, good.
:20 - player flying between acid drips and catching seeds, mastery of jetpack
2:18 - various emergence -
With tuning, pretty much the core gameplay we shipped.
Mostly after this was production and story, etc, other parts of the direction
So: focus on low-level was definitely the right call.
In fact, remember this?
Cut any notion of o2/co2 balance, went with super simple.
Creatures produce biomass, which you need to raise to open airlocks.
One of the reasons the simple mid-level worked is because it fed the low level:
Player created more units more interactions
More things to survive: their own playground.
Would have required players to solve a system of linear equations
Not where you want the focus in an action game
Remember we asked this question?
We found lots of answers in the low level, eg – don’t plant dangerous plants in narrow corridors where they are hard to dodge, do plant in places where seeds will cascade down the terrain and land in FT (assuming that’s your goal, assuming that’s the type of seed you are planting)
Fate of caving and gardening = “lite” versions only.
Not focused on the experiences those are concerned with
Instead concerned with ecosystem stuff
When innovating, can’t support 3 games. TR and LL were not innovative.
Streamlines other games, so easier to do ecosystem stuff
Final summary / conclusions. These are things to be wary of when designing:
The phrase “Doing your chores” or when your game feels like that
Focusing on adding more complex interactions - instead create simple interactions that connect well
or focus on solving the mid-level – make sure your low level Is strong first
Interaction density – monitor the time and actions required to act on meaningful player choice
Nuanced input (which actions, specific timing, analog adjustments) should lead to success, failure, or other meaningful results – this is an indication of depth and mastery
Players with more mastery should be seeing more success, if not why not?
Gameplay should be unpredictable but lead to outcomes that are acceptable to the player, clear why they happened
Noodling vs. anchoring – if you’ve been wandering for a while, take a stand and try something specific and holistic, learn from it then go back into noodling