WEBVTT

00:00:00.680 --> 00:00:06.120
So in the 1970s, there was this experiment
where a bunch of kids were told to

00:00:06.120 --> 00:00:08.039
draw some pictures.

00:00:08.039 --> 00:00:11.040
But before that, the children were split into
groups.

00:00:11.040 --> 00:00:15.000
One group was told that that they would receive
a reward at the end, while the second group

00:00:15.000 --> 00:00:17.080
had no reward.

00:00:17.080 --> 00:00:21.410
After the drawings were finished, the researchers
continued to watch the kids in their classroom

00:00:21.410 --> 00:00:24.890
for a couple weeks - and the results were
pretty interesting.

00:00:24.890 --> 00:00:28.260
And not just "the unanticipated arrival of
a goat in the classroom".

00:00:28.260 --> 00:00:29.410
*Goat bleat*

00:00:29.410 --> 00:00:35.600
But I'll come back to that - because I should
explain what this has got to do with game design.

00:00:35.610 --> 00:00:38.500
Often in design, we want to motivate players.

00:00:38.500 --> 00:00:43.950
Perhaps motivate them to learn a new mechanic,
or encourage them to use a specific feature,

00:00:43.950 --> 00:00:46.420
or just get them to play the game for longer.

00:00:46.420 --> 00:00:50.680
And a popular solution for this is the goal…
and reward.

00:00:50.680 --> 00:00:52.750
Do this, get that.

00:00:52.750 --> 00:00:55.690
Like, quests that lead to experience points.

00:00:55.690 --> 00:00:58.370
Challenges that unlock cosmetics.

00:00:58.370 --> 00:01:04.159
And those cheeky Xbox achievements which are
both a goal and a reward in one tidy package.

00:01:04.159 --> 00:01:08.840
But I'm here to tell you that goals and rewards
don't always work how you want them to.

00:01:08.840 --> 00:01:15.460
And, in fact, in this video I'm going to explain
how they can even have the complete, opposite effect…

00:01:19.140 --> 00:01:24.080
When Klei was making the initial prototype
for its sandbox survival game Don't Starve,

00:01:24.080 --> 00:01:28.250
they quickly realised that testers had no
idea how to play the game - and they instantly

00:01:28.250 --> 00:01:29.470
became stuck.

00:01:29.470 --> 00:01:33.930
So the testers were given a few hints - and
once they got over the hump, they were able

00:01:33.930 --> 00:01:37.670
to experiment, explore, and started to have
a lot of fun.

00:01:37.670 --> 00:01:43.390
In response, Klei decided to create a series
of small, tutorial-like quests to help players

00:01:43.390 --> 00:01:44.390
get started.

00:01:44.390 --> 00:01:46.070
Survive this many nights.

00:01:46.070 --> 00:01:47.180
Find this many items.

00:01:47.180 --> 00:01:48.510
That sort of thing.

00:01:48.510 --> 00:01:49.750
And it worked!

00:01:49.750 --> 00:01:53.340
But only so much as players learned how to
play the game.

00:01:53.340 --> 00:01:57.890
Because beyond that, the quests were a complete
and utter disaster.

00:01:57.890 --> 00:02:02.740
Klei discovered that players focused exclusively
on those quests, and thought of everything

00:02:02.740 --> 00:02:05.159
else as a noisy distraction.

00:02:05.159 --> 00:02:09.969
They optimised their play in really boring
ways in order to finish the quest at hand.

00:02:09.969 --> 00:02:13.200
They avoided doing anything risky, because
it meant they might fail.

00:02:13.200 --> 00:02:17.150
And then they became completely demotivated
the second the quests ran out.

00:02:17.150 --> 00:02:22.299
Klei says "In structuring the game as a series
of explicit tasks to be completed, we taught

00:02:22.299 --> 00:02:26.730
the player to depend upon those tasks to create
meaning in the game".

00:02:26.730 --> 00:02:32.120
In the end, Klei solved its onboarding problem
by tweaking the UI to give players subtle

00:02:32.120 --> 00:02:37.540
hints about how to get started - such as highlighting
the most important items you can craft.

00:02:37.540 --> 00:02:42.989
But the quests were left on the cutting room
floor - leaving players to learn for themselves.

00:02:42.989 --> 00:02:48.409
Because if a game is about experimentation,
exploration, or player-guided discovery - explicit

00:02:48.409 --> 00:02:52.459
goals can limit a player's creativity and
imagination.

00:02:52.459 --> 00:02:54.760
Even after the goals run out.

00:02:54.760 --> 00:02:59.859
This is exactly what drove the development
of the cosmic archeology game Outer Wilds.

00:02:59.859 --> 00:03:04.309
The developers deliberately avoided giving
players explicit goals about where to go,

00:03:04.309 --> 00:03:08.249
or even what you're trying to achieve - so
that players are driven to explore this miniature

00:03:08.249 --> 00:03:11.699
solar system through a sense of curiosity
alone.

00:03:11.699 --> 00:03:14.180
Okay, here's another story.

00:03:14.180 --> 00:03:19.670
Zach Barth makes problem-solving puzzle games
about designing your own automated machines,

00:03:19.670 --> 00:03:22.360
like Exapunks and Shenzen I/O.

00:03:22.360 --> 00:03:27.430
In these games, you can make the machines
however you like - if it works, it works.

00:03:27.430 --> 00:03:31.909
But it's actually really fun to go back in,
and see if you can refine your creation to

00:03:31.909 --> 00:03:34.749
make it, say, smaller or faster.

00:03:34.749 --> 00:03:41.059
So in Zach's first two commercial games, Spacechem
and Infinifactory, he added a few Steam achievements

00:03:41.059 --> 00:03:45.919
that encourage this sort of optimisation - like
the Spacechem achievement "Beat the assignment

00:03:45.919 --> 00:03:49.260
"No Thanks Necessary" in under 2200 cycles."

00:03:49.260 --> 00:03:54.680
But, in all the games released after that
- those achievements are completely gone.

00:03:54.680 --> 00:03:55.869
What's up with that?

00:03:55.869 --> 00:04:00.010
ZACH: "We wanted to add achievements because
that was back when achievements were cool.

00:04:00.010 --> 00:04:01.409
That was back before I thought achievements
were awful."

00:04:01.409 --> 00:04:04.389
ZACH: "The thing i don't like about them is
that the game already has a reward system.

00:04:04.389 --> 00:04:08.609
We have something that's far more meaningful
and far less arbitrary than a random threshold."

00:04:08.609 --> 00:04:14.480
What Zach's talking about is a bounty of metrics
that you can use to gauge how well you've done.

00:04:14.480 --> 00:04:16.400
There's your own personal score.

00:04:16.400 --> 00:04:18.609
There's leaderboards that compare you to your
Steam friends.

00:04:18.609 --> 00:04:24.000
And there are these brilliant histograms that
show you how your solution stacks up in comparison

00:04:24.000 --> 00:04:25.789
to every other player.

00:04:25.789 --> 00:04:30.199
All of these - the strive to beat your personal
best, or the drive to do better than other

00:04:30.199 --> 00:04:33.360
players - are extremely strong motivators
to do better.

00:04:33.360 --> 00:04:38.460
As Zach says: "a goal that you set yourself
is way more powerful than a goal someone else

00:04:38.460 --> 00:04:40.090
sets for you".

00:04:40.090 --> 00:04:45.320
So if a game is about improving yourself,
a personal or social goal can be a stronger

00:04:45.320 --> 00:04:48.509
motivator than a set threshold.

00:04:48.509 --> 00:04:53.090
My final story comes from the adorable track-laying
puzzle game Mini Metro.

00:04:53.090 --> 00:04:57.000
The game's developers wanted to focus on personal
growth and high scores.

00:04:57.000 --> 00:05:02.180
And so - according to UI designer Jamie Churchman
- the team specifically tried to avoid these

00:05:02.180 --> 00:05:06.720
goal and reward meta structures as they can
become a "means to an end".

00:05:06.720 --> 00:05:10.860
For example, the game does have unlockable
cities - which is just to limit player choice

00:05:10.860 --> 00:05:12.520
at the beginning of the game.

00:05:12.520 --> 00:05:17.341
But Jamie acknowledges that some people will
play each city until the threshold, unlock

00:05:17.341 --> 00:05:21.430
the next one, and when they've unlocked all
the cities they feel like they've finished

00:05:21.430 --> 00:05:23.530
the game and can stop playing.

00:05:23.530 --> 00:05:27.539
We should remember that goals are a checklist
that can be completed.

00:05:27.539 --> 00:05:32.010
And like with Don't Starve, some players will
exclusively rely on the game to give them

00:05:32.010 --> 00:05:33.919
purpose and direction.

00:05:33.919 --> 00:05:39.400
But measurements of your skill - such as leaderboards
and scoring systems, have no finish: you can

00:05:39.400 --> 00:05:43.960
continue to improve on your personal best
forever - which partly explains why we can

00:05:43.960 --> 00:05:48.280
still play Tetris after three decades.

00:05:48.280 --> 00:05:53.080
To truly understand what's happening here,
we need to take a quick detour into the world

00:05:53.080 --> 00:05:55.110
of behavioural psychology.

00:05:55.110 --> 00:06:00.229
When thinking about motivation, one of the
most popular models is the idea of extrinsic

00:06:00.229 --> 00:06:02.259
and intrinsic motivation.

00:06:02.259 --> 00:06:06.759
To make it simple, extrinsic motivation is
when we are doing a task for reasons beyond

00:06:06.759 --> 00:06:09.960
the task itself - usually in order to receive
a reward.

00:06:09.960 --> 00:06:12.930
Or, as that's better known: a job.

00:06:12.930 --> 00:06:17.360
On the other hand, intrinsic motivation is
when we do a task for its own sake, simply

00:06:17.360 --> 00:06:19.800
because we find it enjoyable or meaningful.

00:06:19.800 --> 00:06:22.970
Or, as that's better known: a hobby.

00:06:22.970 --> 00:06:27.009
Intrinsic motivation is shown to be far stronger
- and it lasts longer too.

00:06:27.009 --> 00:06:29.780
People can enjoy a hobby for a lifetime.

00:06:29.780 --> 00:06:33.300
Extrinsic motivation will only last as long
as the rewards are there.

00:06:33.300 --> 00:06:37.340
Just see if someone will still work in your
factory after you stop paying them.

00:06:37.340 --> 00:06:40.159
And this bring us back to that classroom from
earlier.

00:06:40.159 --> 00:06:44.430
Okay, so the point of the study was that the
kids had already shown interest in drawing

00:06:44.430 --> 00:06:45.600
before the study began.

00:06:45.600 --> 00:06:47.980
They were intrinsically motivated.

00:06:47.980 --> 00:06:52.020
Then, they were asked to make a picture - and,
like I said, one group was promised a reward,

00:06:52.020 --> 00:06:53.690
and the second group wasn't.

00:06:53.690 --> 00:06:58.319
Afterwards, the researchers continued to watch
the kids in their classroom for a couple weeks

00:06:58.319 --> 00:07:01.279
and found that the children who received a
reward for their drawing?

00:07:01.279 --> 00:07:05.120
Well, they showed much less interest in drawing
afterwards.

00:07:05.120 --> 00:07:07.449
And their pictures were of a lower quality
too.

00:07:07.449 --> 00:07:11.210
Which is - wow, way to burn a bunch of kids,
science.

00:07:11.210 --> 00:07:13.599
This is called the overjustification effect.

00:07:13.599 --> 00:07:18.099
And there's a huge body of evidence that says
when extrinsic motivation is attached to a

00:07:18.099 --> 00:07:23.490
task that we already find intrinsically motivating,
we suddenly become way less interested in

00:07:23.490 --> 00:07:24.490
the task.

00:07:24.490 --> 00:07:29.669
And other studies show rewards also make people
less creative, worse at problem solving, more

00:07:29.669 --> 00:07:35.710
prone to cheating, and may lose motivation
entirely once the rewards stop - even though

00:07:35.710 --> 00:07:37.879
previously they were happy to do it for its
own sake!

00:07:37.879 --> 00:07:38.879
Whoops!

00:07:38.879 --> 00:07:42.349
And I think we can apply this idea to game
design.

00:07:42.349 --> 00:07:46.729
Because there are certainly games that lean
more towards intrinsic motivation.

00:07:46.729 --> 00:07:52.039
Like games that focus on exploration, creativity,
expression, and growth.

00:07:52.039 --> 00:07:55.999
There are games where you set your own goals
and expect no rewards in return.

00:07:55.999 --> 00:08:00.680
And so when more extrinsically motivating
systems - like explicit goals, progression

00:08:00.680 --> 00:08:05.610
meters, and achievements - are added to these
games, our motivation can take a hit.

00:08:05.610 --> 00:08:07.689
We become blinkered to creative solutions.

00:08:07.689 --> 00:08:10.069
We're less motivated to improve ourselves.

00:08:10.069 --> 00:08:13.180
We put an arbitrary threshold on how much
we attain.

00:08:13.180 --> 00:08:17.789
And developers now need to create a constant
drip feed of new goals and rewards, or risk

00:08:17.789 --> 00:08:20.120
losing us entirely.

00:08:20.120 --> 00:08:24.289
Of course, that's not to say that developers
should never add goals and rewards to these

00:08:24.289 --> 00:08:26.550
more intrinsically motivating games.

00:08:26.550 --> 00:08:31.509
Because, I think it's clear that some people
just aren't very good or interested in motivating

00:08:31.509 --> 00:08:32.509
themselves.

00:08:32.509 --> 00:08:36.320
For every Minecraft super fan who generates
their own fun, there's someone else who is

00:08:36.320 --> 00:08:38.620
simply lost and without direction.

00:08:38.620 --> 00:08:41.870
It reminds me of my all-time favourite Steam
forum post.

00:08:41.870 --> 00:08:46.579
In a thread about the open-ended whodunnit
Her Story, one user said "It's up to you to

00:08:46.579 --> 00:08:49.880
decide when you are satisfied with the information
you have found".

00:08:49.880 --> 00:08:54.949
To which the thread's author replied, "how
do I decide when I am satisfied?".

00:08:54.949 --> 00:08:57.130
That post keeps me up at night.

00:08:57.130 --> 00:09:02.540
Anyway - the nice thing about goals and rewards
is that they can provide structure and progression

00:09:02.540 --> 00:09:03.540
to play.

00:09:03.540 --> 00:09:07.220
So they can still be used, they just have
to be applied carefully.

00:09:07.220 --> 00:09:12.420
For example, with goals - it's better to use
large, overarching goals that players can

00:09:12.420 --> 00:09:17.510
complete however they want, rather than restrictive
step-by-step instructions.

00:09:17.510 --> 00:09:22.950
You can focus on comparative metrics, like
leaderboards, histograms, and personal bests,

00:09:22.950 --> 00:09:25.220
rather than absolute thresholds.

00:09:25.220 --> 00:09:30.949
Make goals optional, like Hitman's challenges,
or hidden, like Outer Wilds' achievements.

00:09:30.949 --> 00:09:36.139
And in terms of rewards - well, actually there
is one type of reward that has been shown

00:09:36.139 --> 00:09:39.110
to not trigger the overjustification effect.

00:09:39.110 --> 00:09:43.660
Because, in that study with the children - there
was actually a third group: children that

00:09:43.660 --> 00:09:49.140
were simply told to go off and draw - but
then were given a reward at the end, as a surprise.

00:09:49.140 --> 00:09:53.079
In the following weeks, these children spent
the largest amount of time drawing… of all

00:09:53.079 --> 00:09:56.980
- if only by a small margin beyond the kids
without rewards.

00:09:56.980 --> 00:10:01.360
This, and plenty of other studies, show that
rewards can have a motivational effect in

00:10:01.360 --> 00:10:06.509
intrinsic situations - provided that they're
unexpected, reasonably low value, and feel

00:10:06.509 --> 00:10:09.350
tied to the actual performance of the action.

00:10:09.350 --> 00:10:13.900
An example of this in games might be Overwatch's
Play of the Game, which is a short highlight

00:10:13.900 --> 00:10:16.690
reel showcasing the best moment in the match.

00:10:16.690 --> 00:10:21.250
It doesn't really do anything, but it's a
huge boost to the ego of the player who gets

00:10:21.250 --> 00:10:22.740
the starring role.

00:10:22.740 --> 00:10:26.440
And this is all over Nintendo's latest blockbusters.

00:10:26.440 --> 00:10:31.160
In Odyssey, there's nothing telling you to
clamber up here with Mario's advanced move-set:

00:10:31.160 --> 00:10:34.380
but here's a cheeky cache of coins as a pat
on the back.

00:10:34.380 --> 00:10:40.259
And in Breath of the Wild, every suspicious
nook could be a reward, like a Korok seed.

00:10:40.259 --> 00:10:45.050
As Nintendo's Bill Trinen says: "When they
create their games, [Nintendo's designers]

00:10:45.050 --> 00:10:49.420
don't tell you how to play their game in order
to achieve some kind of mythical reward.

00:10:49.420 --> 00:10:53.600
There are things you can do in the game that
will result in some sort of reward or unexpected

00:10:53.600 --> 00:10:54.600
surprise.

00:10:54.600 --> 00:10:59.550
In my mind, that really encourages the sense
of exploration rather than the sense of 'If

00:10:59.550 --> 00:11:06.230
I do that, I'm going to get some sort of artificial
point or score'."

00:11:06.230 --> 00:11:09.000
Hey, thanks for watching!

00:11:09.000 --> 00:11:14.529
Just wanted to let you know that GMTK videos
will now be ad free - so a big thanks to all

00:11:14.529 --> 00:11:18.720
of my Patrons for supporting this work and
making these videos possible.

00:11:18.720 --> 00:11:20.839
You are amazing.