1
00:00:00,340 --> 00:00:01,193
Balatro!

2
00:00:01,240 --> 00:00:03,573
It’s the hot new indie darling.

3
00:00:03,660 --> 00:00:06,233
It shifted
a million copies in a month,

4
00:00:06,234 --> 00:00:09,292
it’s been streamed
by pretty much everyone on Twitch,

5
00:00:09,422 --> 00:00:13,564
and it’s one of the top-rated games of 2024 so far.

6
00:00:13,824 --> 00:00:14,885
I’ve also played it...

7
00:00:15,133 --> 00:00:16,167
quite a bit.

8
00:00:16,273 --> 00:00:19,856
But, this game has,
according to its own designer,

9
00:00:19,968 --> 00:00:22,749
a “fundamental design flaw”.

10
00:00:23,346 --> 00:00:27,534
A “cursed problem” that the designer has been
unable to solve.

11
00:00:28,459 --> 00:00:29,386
Let me explain.

12
00:00:30,277 --> 00:00:33,181
First, if you somehow haven’t
played Balatro,

13
00:00:33,335 --> 00:00:36,275
it’s a card game about trying
to find poker hands

14
00:00:36,361 --> 00:00:37,806
in order to score points.

15
00:00:38,005 --> 00:00:40,005
Better hands score bigger points.

16
00:00:40,114 --> 00:00:41,511
But on top of that,

17
00:00:41,585 --> 00:00:45,235
you can do all sorts of wily tricks
to boost your score.

18
00:00:45,381 --> 00:00:47,634
Special cards rack up more points.

19
00:00:47,805 --> 00:00:51,131
Crazy joker cards change
the rules of the game.

20
00:00:51,374 --> 00:00:54,391
And you can stack your deck,
or toss away cards,

21
00:00:54,571 --> 00:00:57,076
to make certain hands more happenable.

22
00:00:57,644 --> 00:00:59,401
It’s a really fun game.

23
00:00:59,553 --> 00:01:04,179
A sublime, synergistic slot machine
that feels fun to play,

24
00:01:04,293 --> 00:01:06,101
and is even more fun to break.

25
00:01:06,214 --> 00:01:08,658
It’s also really elegantly designed

26
00:01:08,788 --> 00:01:11,366
with its clean UI,
straightforward concept,

27
00:01:11,393 --> 00:01:12,452
and, well,

28
00:01:12,600 --> 00:01:17,084
just how much depth has been squeezed
out of a few key systems.

29
00:01:17,365 --> 00:01:20,549
But there’s one…
interesting design choice in there.

30
00:01:20,779 --> 00:01:21,653
And it’s this:

31
00:01:21,794 --> 00:01:25,413
the game doesn’t tell you
how many points you’re going to score

32
00:01:25,541 --> 00:01:27,017
before you play your hand.

33
00:01:27,404 --> 00:01:31,575
You simply pick your cards,
cross your fingers, and hit go.

34
00:01:32,181 --> 00:01:35,556
Now, this would be really
quite helpful information.

35
00:01:35,863 --> 00:01:39,231
It could tell you to play one hand
over another.

36
00:01:39,416 --> 00:01:42,871
It could tell you if you’re about to
scrape past the ante

37
00:01:43,091 --> 00:01:45,032
or miss it by a few points.

38
00:01:45,194 --> 00:01:48,218
It could tell you if you’re about
to win the entire game

39
00:01:48,528 --> 00:01:51,726
or lose the whole thing
and have to start from scratch.

40
00:01:51,931 --> 00:01:52,591
And yet…

41
00:01:52,636 --> 00:01:57,814
Balatro doesn’t give you a,
let’s call it a "score preview".

42
00:01:58,303 --> 00:02:02,471
Of course, this was
an entirely intentional design choice.

43
00:02:02,698 --> 00:02:06,183
LocalThunk, the game’s anonymous
Canadian designer,

44
00:02:06,357 --> 00:02:07,896
has explained that, for him,

45
00:02:08,063 --> 00:02:13,410
the joy of Balatro lives in
that precise moment I just described.

46
00:02:13,505 --> 00:02:15,725
When you cross your fingers
and hit play.

47
00:02:15,976 --> 00:02:18,644
When you’ve set up
your point-scoring engine

48
00:02:18,851 --> 00:02:21,302
and hope that it will
bring home the bacon.

49
00:02:21,519 --> 00:02:26,201
LocalThunk says: “My personal belief is
that the game is more fun when you set up

50
00:02:26,202 --> 00:02:29,010
your Rube Goldberg machine
and watch it go

51
00:02:29,230 --> 00:02:32,925
before knowing whether or not
the hand will win the round.”

52
00:02:33,345 --> 00:02:35,892
And Balatro totally plays into this!

53
00:02:36,025 --> 00:02:39,752
There’s so much hype and pageantry
after playing your hand.

54
00:02:40,056 --> 00:02:43,134
The numbers tick up,
with escalating sound effects.

55
00:02:43,283 --> 00:02:46,329
Each card and joker steps forward
in turn

56
00:02:46,469 --> 00:02:48,231
to add their points to the total.

57
00:02:48,484 --> 00:02:51,911
If you’re lucky,
the score multiplier will set on fire

58
00:02:52,051 --> 00:02:54,310
and start to burn hotter and hotter

59
00:02:54,450 --> 00:02:56,195
with each multiplication.

60
00:02:56,544 --> 00:02:59,846
And so, if you already knew
how many points you were going to get.

61
00:02:59,886 --> 00:03:03,350
If a bit of UI had
pre-calculated the score

62
00:03:03,390 --> 00:03:06,534
and told you that you were going to
win the ante with this hand…

63
00:03:06,746 --> 00:03:08,060
well none of that would matter.

64
00:03:08,113 --> 00:03:10,114
In fact, it would just get in the way.

65
00:03:10,806 --> 00:03:13,931
This is not the only reason
to forgo a score preview, mind.

66
00:03:14,078 --> 00:03:16,145
It would also add cruft to the UI,

67
00:03:16,365 --> 00:03:20,156
especially when you need to account
for cards that have random properties.

68
00:03:20,396 --> 00:03:25,037
How do you elegantly show
a range of possible scores?

69
00:03:25,353 --> 00:03:26,923
It would slow the game down,

70
00:03:27,117 --> 00:03:30,195
incentivising players to check
every possible hand

71
00:03:30,355 --> 00:03:32,551
to find
the highest-scoring combination.

72
00:03:32,671 --> 00:03:34,682
And it would change the entire feel,

73
00:03:34,771 --> 00:03:37,692
from a chill game about
vibing with cards

74
00:03:37,818 --> 00:03:41,349
to a stern spreadsheet-style
strategy game.

75
00:03:41,679 --> 00:03:45,934
And so this is a totally legit
game design decision, right?

76
00:03:46,154 --> 00:03:47,832
Every game designer has to choose

77
00:03:47,952 --> 00:03:50,238
how much information
to give to the player.

78
00:03:50,399 --> 00:03:53,408
Like, should you show
a boss’s health bar,

79
00:03:53,561 --> 00:03:55,060
or keep it hidden?

80
00:03:55,714 --> 00:03:58,260
Should enemies come
up with their strategies in secret,

81
00:03:58,373 --> 00:04:01,241
or should their intent be explained
to the player?

82
00:04:01,541 --> 00:04:04,108
As I’ve explored in
various other videos,

83
00:04:04,289 --> 00:04:06,120
how much information
a player has

84
00:04:06,300 --> 00:04:10,253
will change their behaviour,
and change the way the game feels.

85
00:04:10,776 --> 00:04:15,568
And so, Balatro hides its score preview
to make players act more quickly

86
00:04:15,772 --> 00:04:18,595
and to create
a feeling of suspense and drama

87
00:04:18,776 --> 00:04:20,107
whenever you play a hand.

88
00:04:20,359 --> 00:04:22,814
LocalThunk had an experience in mind

89
00:04:22,948 --> 00:04:24,894
and picked mechanics
that would nudge players

90
00:04:25,089 --> 00:04:26,450
towards that feeling.

91
00:04:26,669 --> 00:04:28,296
A smart design choice.

92
00:04:28,890 --> 00:04:29,658
However!

93
00:04:30,024 --> 00:04:33,548
Balatro is not like those other games
I just showed.

94
00:04:33,674 --> 00:04:37,472
Because while the score preview is
hidden from the player…

95
00:04:37,705 --> 00:04:41,219
the information is still
technically available!

96
00:04:41,339 --> 00:04:44,056
Because you can just…
you can just calculate it yourself.

97
00:04:44,442 --> 00:04:48,992
So, like…. a straight is worth
30 chips and 4 mult.

98
00:04:49,258 --> 00:04:51,351
These cards are going to add an extra

99
00:04:51,545 --> 00:04:55,129
10, 20, 30, 39, 47 chips,

100
00:04:55,214 --> 00:04:58,678
and then the two face cards will add
another 30 each

101
00:04:58,852 --> 00:04:59,783
thanks to this joker.

102
00:04:59,863 --> 00:05:02,988
So that’s 137 chips times 4…

103
00:05:03,468 --> 00:05:04,994
548.

104
00:05:05,118 --> 00:05:07,031
Not quite enough to beat the ante,

105
00:05:07,164 --> 00:05:07,924
but close.

106
00:05:08,236 --> 00:05:10,951
And so,
if information in a game can be

107
00:05:11,018 --> 00:05:12,737
hidden, or visible…

108
00:05:13,003 --> 00:05:16,963
Balatro’s score preview falls
into a weird half category:

109
00:05:17,270 --> 00:05:21,032
hidden, but attainable
if you really want it.

110
00:05:21,379 --> 00:05:25,318
And that’s the fundamental design flaw
at the heart of Balatro.

111
00:05:25,525 --> 00:05:28,805
The designer wants the excitement
of a slot machine,

112
00:05:28,978 --> 00:05:31,500
but also the numerical predicability

113
00:05:31,500 --> 00:05:32,793
of an Excel spreadsheet.

114
00:05:33,060 --> 00:05:35,801
And so the only way
to square that circle

115
00:05:35,967 --> 00:05:40,630
is to hope that players won’t bother
to calculate the final score.

116
00:05:40,963 --> 00:05:44,912
But, if we go back to
that timeless Soren Johnson quote:

117
00:05:45,052 --> 00:05:48,858
“Given the opportunity,
players will optimize the fun out of a game.”

118
00:05:49,171 --> 00:05:52,217
It shouldn’t be surprising that
a number of Balatro players

119
00:05:52,544 --> 00:05:55,769
are playing the game with the calculator app
open on their phone,

120
00:05:55,956 --> 00:05:59,027
or with a spreadsheet
set up on a second monitor,

121
00:05:59,174 --> 00:06:01,265
or with Steam’s in-game overlay

122
00:06:01,392 --> 00:06:05,388
showing a bespoke website
that calculates Balatro hands.

123
00:06:05,682 --> 00:06:10,617
And, actually this is exactly why
Balatro has a deck view.

124
00:06:10,840 --> 00:06:13,510
During playtesting,
the game did not show you

125
00:06:13,657 --> 00:06:15,669
which cards were left in your deck.

126
00:06:15,902 --> 00:06:20,148
But, playtesters could technically
get that information

127
00:06:20,261 --> 00:06:22,601
by tracking which cards
had already been played.

128
00:06:22,961 --> 00:06:24,461
And after polling users,

129
00:06:24,641 --> 00:06:27,587
LocalThunk found
that many were doing just that

130
00:06:27,787 --> 00:06:30,021
even though it really wasn’t much fun.

131
00:06:30,114 --> 00:06:33,124
So he added
a powerful deck peek feature.

132
00:06:33,578 --> 00:06:36,258
But a score preview felt different.

133
00:06:36,325 --> 00:06:39,886
It felt like it encroached upon
the DNA of the game.

134
00:06:39,986 --> 00:06:42,048
It stepped on the stuff
that made Balatro…

135
00:06:42,550 --> 00:06:43,355
Balatro.

136
00:06:43,582 --> 00:06:45,592
And so,
while the designer is empathetic

137
00:06:45,593 --> 00:06:48,177
to people who wish to
play more strategically.

138
00:06:48,417 --> 00:06:52,987
And is bummed out that
the optimum way to play involves busywork,

139
00:06:53,067 --> 00:06:56,119
and doing calculations
outside of the game,

140
00:06:56,319 --> 00:06:59,816
he worries that adding
a score preview would spoil the fun

141
00:07:00,003 --> 00:07:02,029
for those who wish to play
more casually.

142
00:07:02,335 --> 00:07:03,914
And that’s totally true!

143
00:07:04,108 --> 00:07:08,425
Making a game better for one group
can make it worse for another.

144
00:07:08,638 --> 00:07:12,078
As a designer, you need to be certain
who the game is for

145
00:07:12,238 --> 00:07:16,224
and then protect that player base
from certain design decisions.

146
00:07:16,524 --> 00:07:20,512
Even if that design decision is provided
merely as an option.

147
00:07:20,705 --> 00:07:24,147
Speaking on the Eggplant podcast,
LocalThunk says:

148
00:07:24,315 --> 00:07:26,752
“If I add an option
to have this score preview,

149
00:07:26,859 --> 00:07:28,630
people are just going to click on it,

150
00:07:28,770 --> 00:07:32,171
and they're not going to experience
the game that I wanted to create.”

151
00:07:32,337 --> 00:07:36,716
And besides, should a designer even have
to endorse an option

152
00:07:36,717 --> 00:07:39,731
that directly goes against
their intentions for the game?

153
00:07:39,944 --> 00:07:43,287
LocalThunk has been clear
that he made the game for himself

154
00:07:43,494 --> 00:07:46,675
and isn’t interested in changing the game for other people.

155
00:07:46,929 --> 00:07:49,919
Even if there are a million of them.

156
00:07:50,593 --> 00:07:52,024
But, here’s the rub.

157
00:07:52,271 --> 00:07:54,591
It’s one thing to make
a bold design choice

158
00:07:54,638 --> 00:07:58,261
and then stand by it, for
the betterment of the game.

159
00:07:58,395 --> 00:08:01,290
To shun the haters
and stick by your design.

160
00:08:01,479 --> 00:08:03,556
But that doesn’t really work

161
00:08:03,929 --> 00:08:07,739
if there’s a way for players
to find a way around your choice,

162
00:08:07,792 --> 00:08:10,747
no matter how tedious
that loophole might be.

163
00:08:11,020 --> 00:08:12,066
And we know this!

164
00:08:12,153 --> 00:08:15,897
Because this is not the first time
this has happened to a game.

165
00:08:16,024 --> 00:08:18,605
In fact, it’s not even the first time
it’s happened

166
00:08:18,752 --> 00:08:21,132
to an extremely popular roguelike.

167
00:08:21,270 --> 00:08:23,405
Enter The Binding of Isaac.

168
00:08:23,775 --> 00:08:26,174
So this basement-dwelling
dungeon crawler

169
00:08:26,247 --> 00:08:28,822
is packed with powerful
items and upgrades…

170
00:08:29,049 --> 00:08:31,814
but the game doesn’t tell you
what they do.

171
00:08:32,060 --> 00:08:33,892
They just have a name,

172
00:08:34,106 --> 00:08:35,892
or a cryptic tagline,

173
00:08:35,998 --> 00:08:38,563
or maybe just three question marks.

174
00:08:38,943 --> 00:08:41,791
The game’s designer, Edmund McMillen,
did this on purpose

175
00:08:41,844 --> 00:08:43,952
to create
a feeling of mystery,

176
00:08:44,038 --> 00:08:47,053
similar to the sensation he got
when playing games as a kid

177
00:08:47,132 --> 00:08:49,343
like the original Legend of Zelda.

178
00:08:49,494 --> 00:08:51,065
He described that game by saying:

179
00:08:51,278 --> 00:08:54,658
“You weren't sure what things did
until you experimented with them,

180
00:08:54,715 --> 00:08:56,471
and you had to brainstorm
with your friends

181
00:08:56,505 --> 00:08:59,887
and put all your findings together
in order to progress.”

182
00:09:00,153 --> 00:09:03,308
And so to mimic
that mysterious sensation in Isaac,

183
00:09:03,413 --> 00:09:06,113
the items are deliberately left
unexplained.

184
00:09:06,269 --> 00:09:08,784
You’ll need to pick things up,
try them,

185
00:09:08,851 --> 00:09:10,824
and puzzle out their properties.

186
00:09:10,944 --> 00:09:14,002
Finding a new item
should lead to curiosity,

187
00:09:14,275 --> 00:09:16,720
experimentation, and surprise.

188
00:09:17,220 --> 00:09:18,109
And that worked…

189
00:09:18,682 --> 00:09:20,459
for about five seconds.

190
00:09:20,599 --> 00:09:23,003
And then people figured out
what all the items did

191
00:09:23,150 --> 00:09:26,274
and put that information up on wikis
and other websites.

192
00:09:26,601 --> 00:09:29,219
Want to know what
this weird little thing will do?

193
00:09:29,412 --> 00:09:31,362
Just find it on Platinum God

194
00:09:31,469 --> 00:09:33,752
and mouse over it
for a full description.

195
00:09:34,146 --> 00:09:35,675
So, like Balatro,

196
00:09:35,769 --> 00:09:39,361
McMillen chose to hide information
to create a certain feeling.

197
00:09:39,522 --> 00:09:42,452
But because that information is
technically attainable

198
00:09:42,559 --> 00:09:45,249
this time with a Google search
rather than a spreadsheet,

199
00:09:45,343 --> 00:09:47,501
a number of players ended up
playing the game

200
00:09:47,502 --> 00:09:50,356
in a completely different way
than the designer intended.

201
00:09:50,749 --> 00:09:52,501
Arguably, a worse way.

202
00:09:52,815 --> 00:09:56,891
And so after multiple DLC packs
which added hundreds of new pick-ups,

203
00:09:57,125 --> 00:09:58,953
this has become, basically,

204
00:09:59,167 --> 00:10:02,218
the defacto way to play
The Binding of Isaac.

205
00:10:02,578 --> 00:10:04,462
McMillen says:
“People would always say,

206
00:10:04,549 --> 00:10:07,506
'You can't play Isaac
without a browser open on your phone.'

207
00:10:07,613 --> 00:10:10,569
I hated that that's how everyone played
for so long...

208
00:10:10,703 --> 00:10:11,927
and still play."

209
00:10:12,307 --> 00:10:15,194
In fact,
he’s described the lack of item descriptions

210
00:10:15,274 --> 00:10:17,800
as the biggest flaw
with Binding of Isaac.

211
00:10:18,147 --> 00:10:20,384
This design choice has
basically haunted him

212
00:10:20,524 --> 00:10:22,475
in the years since Isaac’s release.

213
00:10:22,629 --> 00:10:25,510
And in a post in 2023,
McMillen has said

214
00:10:25,511 --> 00:10:28,320
that he’s considering finally adding
item descriptions

215
00:10:28,320 --> 00:10:30,542
into the game as an optional feature.

216
00:10:30,682 --> 00:10:33,275
Perhaps deciding that it’s better
to support them officially

217
00:10:33,462 --> 00:10:35,589
than players having a worse time
with your game

218
00:10:35,769 --> 00:10:37,930
because of the way you designed it.

219
00:10:38,356 --> 00:10:42,155
And I wonder if something similar
might happen with Balatro.

220
00:10:42,503 --> 00:10:45,480
Now, I don’t think the two examples are
exactly the same.

221
00:10:45,742 --> 00:10:49,507
I agree that Balatro is more fun to play
without score previews

222
00:10:49,613 --> 00:10:52,853
and I’ve never once thought
to pre-calculate a score

223
00:10:53,074 --> 00:10:55,244
in the 30-odd hours
I’ve played the game.

224
00:10:55,371 --> 00:10:58,159
This issue only really affects
a small portion

225
00:10:58,166 --> 00:11:02,085
of the game’s most hardcore,
strategy-minded audience.

226
00:11:02,312 --> 00:11:06,035
But over time,
as the game’s long tail stretches out,

227
00:11:06,228 --> 00:11:09,271
I think this decision might come
to haunt the developer,

228
00:11:09,417 --> 00:11:11,392
just like Isaac’s item descriptions.

229
00:11:11,742 --> 00:11:13,708
But, if you’re watching LocalThunk,

230
00:11:13,968 --> 00:11:17,407
I think there are ways to provide this
as an option to these players…

231
00:11:17,593 --> 00:11:20,220
without spoiling the game
for everyone else.

232
00:11:20,474 --> 00:11:23,434
For one, a score preview is only needed

233
00:11:23,487 --> 00:11:26,575
by players who are incredibly invested
in the game,

234
00:11:26,762 --> 00:11:29,810
so the option could be granted
as a late-game unlock,

235
00:11:30,097 --> 00:11:32,687
and not as something
you can switch on from the word go.

236
00:11:33,040 --> 00:11:37,433
Kinda like how Chrono Cross has
a fast-forward button,

237
00:11:37,666 --> 00:11:40,022
but it only unlocks
after you’ve beaten the game.

238
00:11:40,522 --> 00:11:43,800
The option could also be
clearly communicated to the player,

239
00:11:43,954 --> 00:11:47,085
like how Celeste prefaces
its powerful assist mode

240
00:11:47,392 --> 00:11:50,266
with a message that explains
who this option is for.

241
00:11:50,553 --> 00:11:54,254
Or how Heat Signature politely asks you
to not turn off Permadeath,

242
00:11:54,494 --> 00:11:56,539
please, it’s there for a reason.

243
00:11:56,950 --> 00:11:59,915
Or Balatro could open itself up
to mods,

244
00:12:00,022 --> 00:12:03,398
so users can hack
their own score preview into the game,

245
00:12:03,478 --> 00:12:06,250
without the developer needing to
officially support it.

246
00:12:06,483 --> 00:12:08,962
This is actually what
happened to Isaac.

247
00:12:09,188 --> 00:12:13,359
The ‘External Item Descriptions’ mod is
the most popular Isaac add-on

248
00:12:13,479 --> 00:12:17,398
in the game’s Steam workshop,
with almost 2 million subscribers.

249
00:12:17,531 --> 00:12:19,517
That’s not great for console players,
though.

250
00:12:19,518 --> 00:12:22,258
So it could instead be provided
as a cheat code,

251
00:12:22,319 --> 00:12:25,211
so players have to actively seek
this thing out,

252
00:12:25,451 --> 00:12:29,130
rather than stumble onto it
as an innocuous option in the menu.

253
00:12:29,360 --> 00:12:32,601
As I’ve discussed in my
videos about accessibility,

254
00:12:32,688 --> 00:12:36,150
there are plenty of ways
to open a game up to a wider audience,

255
00:12:36,503 --> 00:12:40,511
without necessarily spoiling it
for the target group of players.

256
00:12:40,983 --> 00:12:43,021
Whatever LocalThunk decides to do,

257
00:12:43,088 --> 00:12:46,730
this has proven to be
a fascinating game design case study.

258
00:12:46,964 --> 00:12:49,216
About how you can change
how a game feels,

259
00:12:49,423 --> 00:12:52,155
by changing how much information
you give to the player.

260
00:12:52,288 --> 00:12:55,568
About how players won’t always act
in the way you want them to,

261
00:12:55,787 --> 00:12:58,420
especially if you
leave open a loophole.

262
00:12:58,620 --> 00:13:01,173
And about how the best intentions
in game design

263
00:13:01,280 --> 00:13:02,743
sometimes have to change,

264
00:13:03,050 --> 00:13:06,431
when you see how players
actually interact with your game.

265
00:13:06,988 --> 00:13:10,069
I’ll be curious to see
what happens with Balatro.

266
00:13:10,402 --> 00:13:13,372
For now, check out this video
on heads-up displays,

267
00:13:13,559 --> 00:13:15,707
where I talk more
about how information can change

268
00:13:15,708 --> 00:13:18,540
the way a player acts and feels.

269
00:13:18,540 --> 00:13:20,921
Thanks for watching.