Hello,

IRDC 2013 came and went, and as usual it was a great time, excellent
source of ideas and awesome motivation booster. I wanted to thank
everyone for coming.

Now, because I couldn't speak during the conference, I attempted to
perform my talk by drawing and writing on the board and frantically
waving my hands. I think I managed to get my main point across, but
I'm not entirely happy with it. That's why I sat down and wrote it
all down in a form that can be, hopefully, read at a leisure:


Games That Don't Teach To Obey
==============================

When you are designing a computer game, there is a number of constraints
you need to take into account. Some of those constraints come from the
technical possibilities at the time, some come from limited time and
budget, some come from the desired theme and inspirations. But there is
also a whole category of constraints that come from the fact that you are
designing a game to be played by humans. Our brains have their limits. We
orient in 2.5 dimensions, we can only hold 5±2 items at a time in our
working memory, we get tired by calculations and step-by-step analysis, we
try to see immediate and binary cause-and-effect relations everywhere and
we use very bad heuristics for probabilities. To be enjoyable, the games
have to take those limitations into account, at the same time being
challenging just enough to keep us in the flow. There are several
well-established techniques that game designers follow since ages to make
their games easy to pick up and enjoy. But there are some consequences.

Because those techniques appear so often, and not just in video games, we
have learned to expect them and take advantage of them. If they are
missing, we tend to dislike the game. There are many such techniques, but
let us look at just one of them as an example: instant feedback.

When we do something in a game, we need to see a reaction within 0.1s,
otherwise it interferes with our amazing ability to "get into the skin" of
the game's character. If we get no feedback for more than 1s, then we
assume that clearly something is broken. After 10s we start doing random
things to see if we get any response at all. Then we lose interest and go
to do something else. That's how we are wired. There can be a lot of
variation in this behavior, depending on how import the task at hand is,
how prepared we are for the lag and whether we have recently eaten some
sweets or not (I'm not kidding!). But the lesson is clear: a game needs to
provide immediate feedback to our actions. It gets worse. If the effect is
too far in time from its cause, we are never going to make the connection.
At least not intuitively – we might be able to reason about it and even do
some in-depth research if we have the incentive – but normally we wouldn't
bother and just assume the effect was a random event, or caused by some
higher power other than RNG. (Humans didn't figure out what causes
pregnancy for hundreds of years!)

This forces the game designers to use certain established patters.
Consider two possible designs of a skill system. In one approach, you
select a skill that you would like to train, and then you go around the
game and gain experience points, until you gather enough of them to get
a level up in the selected skill. In a second approach, you earn
experience points that you can then spend to buy a skill level. The second
system will feel much better to the majority of players, because the
effect (new skill being available) is not removed in time from the cause
(the selection of the skill to train). And indeed, if you look at the
games out there, the second system is much more popular. It just feels so
much better for us to make a choice and immediately be rewarded for it,
than to work hard on something without seeing results.

There are many other such patterns. Indeed, there are so many, that when
we see the initial screen of the Pacman game, we immediately know what to
do: we have to eat all the dots! How do we know it? The game has been
designed with an intent, and because it was designed by humans, and we are
humans too, we are able to pick up hundreds of small hints and guess what
the designer had in mind. And once you guess it, all you have to do is to
perform it – make it happen. Does that feel familiar? When you were at
school, did you notice that they don't really teach you to, say, analyze
poetry? They don't want you to do any analysis. You are just supposed to
guess what the teacher thinks, and then write that. Get the "correct"
answer. Obey. That is what the school trains you to do, and that is also
what the video games train you to do. Obey.

As far as I know, this effect is inherent in some extent in practically
all man-made objects. We can't help ourselves but to put messages in every
little thing we make. Messages that tell about our expectations of the
objects, their affordances. You can't design a video game without
including that "obey" message in it. Or can you? It turns out that you
actually can! But you have to let go of the design, at least in part. Make
it procedurally generated. I think that this is at least in part why
Minecraft has gotten so popular – it's a game that doesn't say "obey",
it's a game that lets you do whatever you want. I think that roguelike
games also have the opportunity to be such games. Of course, this depends
on the game, but the feeling that you are fighting against a random world,
and not against a particularly crafty level designer is worth a lot to me.

But is a "sandbox" game the best that we can make? We can make a game that
doesn't tell you to obey, but would it be possible to make a game that
actively teaches you to not obey? I'm not sure, but here is one idea for
such a game:

* Generate a level procedurally, make sure that it has at least one
  solution.
* Make that solution obvious by marking it up on the level (arrows, signs,
  foot marks, hints).
* Make that marked solution very hard, ineffective and dangerous, by
  generating traps, monsters, locks, etc. and making it expensive in terms
  of time and resources.
* Give the player tools for finding a better solution: damage and erode
  the level randomly, provide wands of digging, bombs, grappling hooks,
  freezing spells, potions of charm monster, teleports, etc.

The designed, easy to read, obvious messages are all against you. The
useful, helping things are all random. You are forced to come up with your
own solutions, instead of relying on guessing the intentions. Sure,
guessing the intentions may still be useful, but only to subvert them.
Ideally, the game would have a similar setting to the puzzle game Portal
– you are a subject of an experiment forced to follow a path. Except in
Portal, even the subvertive parts of the game were designed – no matter
where you got, you knew that there is a designed solution to the puzzle.
Here all bets are off. There might be an insanely simple shortcut that
lets you finish in 3 moves, or there might be none and you will be forced
to follow the marked path after all. I think that this kind of approach
could be good for promoting independent thinking and problem solving.
Unfortunately, although I attempted to implement it during this year's
Seven Day Roguelike Challenge, I saw pretty early that I wouldn't be able
to pull that off and switched to a backup plan.

Whether you like this idea or not, and whether you want to use it or not,
one important lesson that I want you to learn from this is to be careful,
as a game designer, what your games are teaching. Instead of blindly
copying the design patterns from other games, take a step back and think
what kind of behaviors you are promoting. I'm sure that it will lead to
more interesting, less grindy games.

--
Radomir Dopieralski, sheep.art.pl

On Sunday, 16 June 2013 20:47:35 UTC+1, Radomir Dopieralski  wrote:

> Whether you like this idea or not, and whether you want to use it or not,
> one important lesson that I want you to learn from this is to be careful,
> as a game designer, what your games are teaching. Instead of blindly
> copying the design patterns from other games, take a step back and think
> what kind of behaviors you are promoting. I'm sure that it will lead to
> more interesting, less grindy games.

Great write-up, and I think you did communicate this very well on the day.
Plus it was rather fun playing "guess the drawing" as you scribbled about
:)

One thing that comes to mind for me is how in Broken Bottle I gave the
player two choices - to drink or not to drink. But in this context the
more difficult path is the one I want players to take, as I want players
to understand the fun and satisfaction of challenging yourself.

Anyway, overall it's a very neat idea. Portal does it quite nicely but
there's certainly a lot of scope for roguelikes to go even further. I
love the idea of a Dungeon Master watching you progress and commenting,
trying to push you this way and that as you disobey to survive. Could
combine with a Left4Dead AI director gameplay element.

I've put the slides for my talk online with some notes:
http://www.gamesofgrey.com/blog/?p=413
Hopefully others will do the same! There were a lot of good talks.

I'll also hopefully have the podcast edited and uploaded tonight...

--
Darren Grey

Thanks for both of those, coincidentally I've been digitising my conference notes today, very nice to have the actual talks to add to them.

>
> Hopefully others will do the same! There were a lot of good talks.
>

I particularly hope to get my eyes on flend's metamap-making talk materials, I want to use that algorithm sometime.

Meanwhile, my slides currently are at http://bitbucket.org/zasvid/cosmiccommando/downloads/zasvidIRDC2013.pdfzasvidIRDC2013.pdf

However, I don't think they're very telling without commentary (and my OCD is annoyed by the weirdness of double file name in the link), so I'll probably redo them into an article at some point.

--
zasvid

Here's the Roguelike Radio episode from the conference:

http://www.roguelikeradio.com/2013/06/episode-72-irdc-2013.html

Once upon a time, Radomir Dopieralski wrote thus:

> Hello,
>
> IRDC 2013 came and went, and as usual it was a great time, excellent
> source of ideas and awesome motivation booster. I wanted to thank
> everyone for coming.

That turned out most inspiring.  As for motivation I have been burying
myself in code almost constantly since my return.

> Now, because I couldn't speak during the conference, I attempted to
> perform my talk by drawing and writing on the board and frantically
> waving my hands. I think I managed to get my main point across, but
> I'm not entirely happy with it. That's why I sat down and wrote it
> all down in a form that can be, hopefully, read at a leisure:

I enjoyed your presentation very much.  Your drawing skills were up
to the task.  However you are right it is much better to read it whole.

I was thinking about BOSS and PRIME.  Here is part that struck me.

> The designed, easy to read, obvious messages are all against you.

Final enemy holding the Bizarro Orgasmatron is the Shodan.  Lots and lots
of hit points plus it hits like a tank.  Not to mention special attacks
that can slow you, bug (equivalent of curse) your items, sicken you or
burn your stuff.

Shodan teleports around and will do so to ambush you at inopportune moment.
When you manage to whittle boss' health down expect to make a chase all
over the place as the monster teleports away immediately after seeing you.

> The useful, helping things are all random. You are forced to come up
> with your own solutions, instead of relying on guessing the intentions.

(...)

> There might be an insanely simple shortcut that lets you finish in
> 3 moves, or there might be none and you will be forced to follow the
> marked path after all.

If certain item generates (somewhat below average chance) you can score
an instakill against Shodan.  One player profession actually starts with
it in basic inventory.  It is very much possible to come close to fast
kill by other means too but many rely on stuff that needs to be found.

At the same time PRIME fails in another aspect.  The Gamma Caves have to
be crossed in more than 80% of successful games.  If not one probably
would do so anyway for fun or loot.  The threat of the place is hard
radiation.  There are many ways to survive it but in the event that no
item granting some kind of protection is generated there is a radiation
suit deep at the Sewer Plant.  From the sparse reports I receive from
players it seems that everybody goes for the suit and ignores other,
often better ways to insulate oneself from gamma particles.  For now
I have not figured what went wrong here.  Yet.

> Whether you like this idea or not, and whether you want to use it or
> not, one important lesson that I want you to learn from this is to be
> careful, as a game designer, what your games are teaching.

I really need to reflect on this but I am afraid some revelations may
not be to my liking.  Oh well, DCSS Dev Team was bold enough to cut
most favorite player race, redo whole magic schools and put certain
mechanics upside down.  If they could I can too.

--
Michał Bieliński

On 2013-06-18, Michał Bieliński <dungeon...@tlen.pl> wrote:
> Once upon a time, Radomir Dopieralski wrote thus:

From what you are describing, this is exactly the opposite of what I tried
to achieve. In your case, your opponent is random, and you have to rely on
the static, designed, hand-crafted game elements to somehow alleviate that
randomness and defeat it. You need to play enough times to learn about the
game and guess how the author wanted you to behave and what he envisioned.
On the other hand, I want the helpful things in the game to be unlearnable
by repetition -- to change every time you play, so that the players invent
new ways of defeating their obstacles every time. No spoilers are going to
help you (although surely there will be some inevitable patters).

--
Radomir Dopieralski, sheep.art.pl

On 2013-06-16, Radomir Dopieralski <ne...@sheep.art.pl> wrote:
[snip]

> As far as I know, this effect is inherent in some extent in practically
> all man-made objects. We can't help ourselves but to put messages in every
> little thing we make. Messages that tell about our expectations of the
> objects, their affordances. You can't design a video game without
> including that "obey" message in it. Or can you? It turns out that you
> actually can! But you have to let go of the design, at least in part. Make
> it procedurally generated.

Hmm, I think almost any game has to contain some fixed action-reward
associations. At least if the reward can be 'changing the game state
in a desired way'.

a) At the lowest level, there is the user interface and the basic
'physics' of the game world, The player should be able to customize
the UI, but the general structure and the set of things that can be
done is fixed. I don't think there is much potential for variety at
this level _within any given game_. Even a sandbox needs the box.

b) Tactical level: how to defeat the giant rat, get across the river,
regain lost HP, get into the vault? In roguelikes, there are usually
many options, depending on the inventory, spells, abilities, etc.

In DCSS for example, my characters usually have an 'engine' - a set
of ways of dealing with tactical problems. This engine undergoes
major changes several times during the game. Finding the engine
that can work well at the particular stage, with the given equipment
and abilities, is part of the fun of the game. OTOH, I would not like
to find a new way to deal with every monster, or regain the HP / mana
after each battle in a novel way.

c) Strategic level: how to get to the end of the branch, achieve
experience level 14, obtain the Antcrusher Sceptre?

d) Game level: how do I win the game? The game may offer various
endings which the player may chose to achieve - and some he tries
to avoid. The initial class / race selection may also fall into
this level.

The 'levels' as described above may be fuzzy, and quite possibly do
not apply too well to all roguelikes, but I think the pattern is
common enough.

Now, I think some things must be kept constant and relied upon so they
can be used/combined in variable, possibly novel way. The player may
find it hard to notice the unexpected fact that mountain trolls can
make great spellcasters at night in the Chaos Domain if he can't rely
on some knowledge of mountain trolls, spellcasting, nights, and the
Chaos Domain. If the player has to constantly reinvent tactics, it is
difficult for him to be creative at any higher level.

The constant things need not all be hardwired for eternity, they just
need to be learnable by the player. For example, an area spell might
have the affected area procedurally generated - constant for the given
character.

Also, if the player is to become more skilled playing the game again
and again, _some_ things must be fixed - and thus the player has to
learn to obey some rules.

> I think that this is at least in part why
> Minecraft has gotten so popular – it's a game that doesn't say "obey",
> it's a game that lets you do whatever you want. I think that roguelike
> games also have the opportunity to be such games. Of course, this depends
> on the game, but the feeling that you are fighting against a random world,
> and not against a particularly crafty level designer is worth a lot to me.
>
> But is a "sandbox" game the best that we can make? We can make a game that
> doesn't tell you to obey, but would it be possible to make a game that
> actively teaches you to not obey? I'm not sure, but here is one idea for
> such a game:
>
> * Generate a level procedurally, make sure that it has at least one
>   solution.
> * Make that solution obvious by marking it up on the level (arrows, signs,
>   foot marks, hints).
> * Make that marked solution very hard, ineffective and dangerous, by
>   generating traps, monsters, locks, etc. and making it expensive in terms
>   of time and resources.
> * Give the player tools for finding a better solution: damage and erode
>   the level randomly, provide wands of digging, bombs, grappling hooks,
>   freezing spells, potions of charm monster, teleports, etc.
>
> The designed, easy to read, obvious messages are all against you. The
> useful, helping things are all random. You are forced to come up with your
> own solutions, instead of relying on guessing the intentions. Sure,
> guessing the intentions may still be useful, but only to subvert them.

But won't that just lead the player to try to read the 'real'
intentions? "The arrow points down == Going down hurts, avoid if
possible."

> Ideally, the game would have a similar setting to the puzzle game Portal

There is also Antichamber, where misdirection is the rule.

> – you are a subject of an experiment forced to follow a path. Except in
> Portal, even the subvertive parts of the game were designed – no matter
> where you got, you knew that there is a designed solution to the puzzle.
> Here all bets are off. There might be an insanely simple shortcut that
> lets you finish in 3 moves, or there might be none and you will be forced
> to follow the marked path after all. I think that this kind of approach
> could be good for promoting independent thinking and problem solving.
> Unfortunately, although I attempted to implement it during this year's
> Seven Day Roguelike Challenge, I saw pretty early that I wouldn't be able
> to pull that off and switched to a backup plan.
>
> Whether you like this idea or not, and whether you want to use it or not,
> one important lesson that I want you to learn from this is to be careful,
> as a game designer, what your games are teaching.

Whatever the designer intends to teach, some ingenious players will
learn something completely different ;-)

--
To send mail, remove 'AU' from the address
You see here a scroll labeled "MDPifwNN4S8"

On 2013-06-19, Kusigrosz <Kusi...@AUtorun.itvk.pl> wrote:
> On 2013-06-16, Radomir Dopieralski <ne...@sheep.art.pl> wrote:
[snip]

> The player should be able to customize
> the UI, but the general structure and the set of things that can be
> done is fixed.

I strongly disagree here. If you do that, you will never get a good UI.
The customization should be limited to what is required by differences
in hardware, and not a thing more, otherwise you risk that no possible
set of settings is any good.

> b) Tactical level [snip]
> c) Strategic level: [snip]
> d) Game level: [snip]

I'm mostly interested in c), as that has the largest learning opportunity.
Doing it at strategic level gives you a game of chess, and doing ti at
game level gives you minecraft -- both very good games in their own right,
but not that interesting from my point of view.

[snip]

>> The designed, easy to read, obvious messages are all against you. The
>> useful, helping things are all random. You are forced to come up with your
>> own solutions, instead of relying on guessing the intentions. Sure,
>> guessing the intentions may still be useful, but only to subvert them.
>
> But won't that just lead the player to try to read the 'real'
> intentions? "The arrow points down == Going down hurts, avoid if
> possible."

Yes. The whole trick is that "go this way" has this "here, I've made all
the decisions for you, do this, this and this" taste, while "do anything
but don't go this way" doesn't tell you what to do. All you know is that
you are expected to disobey (so there is a message at the game level),
but you are on your own on how to "not go this way".

As for learnability -- sure, that's what I started my talk with. This
"obey" message doesn't come from the malice of the game designers, it's
part of design that is *convenient* for humans. Having all the answers
served on a silver platter sure is convenient! But I believe there are
ways around those limitations. One way to avoid teaching ready solutions
and patterns is to provide an exponential explosion of possibilities
(through interactions of items, monsters and terrain). If you look at
games like Go (Baduk/Weiqui) you will see how that can work to provide
thousands of years worth of learning material, with new strategies and
tactics being discovered after almost 3000 years of playing!

--
Radomir Dopieralski, sheep.art.pl

On Sunday, 16 June 2013 22:47:35 UTC+3, Radomir Dopieralski  wrote:
> But is a "sandbox" game the best that we can make? We can make a game that
> doesn't tell you to obey, but would it be possible to make a game that
> actively teaches you to not obey?

Typically it is that everything can be done with several ways but
one is optimal (cheapest). If to make optimal less obvious and
to push suboptimal then the game teaches to not obey.

Designer has to do it knowingly and carefully but  at the end the
players will anyway find a way how to cheaply spam the scissors
so any rock will fall.

Funny that you mentioned minecraft, it seems that cheapest way
to gain xp and riches there is to feed chicken and to sell wheat and
chicken meat to NPC-s for emeralds. It must be by design or
otherwise why else there is even machine that throws away
chicken eggs. :)

Quoting  Radomir Dopieralski  <ne...@sheep.art.pl>:
>Ideally, the game would have a similar setting to the puzzle game Portal
>– you are a subject of an experiment forced to follow a path. Except in
>Portal, even the subvertive parts of the game were designed – no matter
>where you got, you knew that there is a designed solution to the puzzle.
>Here all bets are off. There might be an insanely simple shortcut that
>lets you finish in 3 moves, or there might be none and you will be forced
>to follow the marked path after all.

Well, you say that, but that was more like the situation in Portal 1;
there's an intended solution, but sometimes there is an unintended
shortcut. In Portal 2 they had more of a playtesting budget, and there's
typically exactly one solution; in particular (disappointingly) it's very
vulnerable to metagaming analysis where you start from the assumption that
every gameplay element and portalable surface was placed there to be used.
--
David Damerell <dame...@chiark.greenend.org.uk>
Clown shoes. I hope that doesn't bother you.
Today is Teleute, Presuary.
Tomorrow will be Oneiros, Presuary.

Isn't that generally how games work? At least in games which give the
player enough latitude to even attempt alternate solutions, usually some
non-intended form of gameplay is optimal. At worst, exploiting glitches
becomes your basic mode of gameplay, like in Dark Souls, where your
enemies' navigational failings are important to defeating them
(e.g. getting giants stuck on stairs).

I don't think this will effectively teach players to think for
themselves. Some players may follow the intended route and miss the
message entirely (less likely). More likely, players will be oblivious
to the intended route. They won't see themselves as disobeying
authority, but rather as ignoring the "noise" that are the solution
marks, or as ignoring lies from their enemies.

To be effective, I think the player would need to be punished for
disobedience, by the same authority that they disobeyed. Otherwise,
authority isn't authority, it's just some kind of bad plan for the
player to ignore. Without mimicing the real-world context in which
people choose to obey or disobey authority, I don't think you could
effectively teach anything about it.

But at this point, I think you then have to decide what kind of
interaction with authority you're trying to teach. For example, which of
these scripts are you trying to get them to identify with?

* When someone says to do something, I don't need to know what they
  said.

* When someone says to do something, it will be better if I don't do it
  and do something else.

* When someone says to do something, they are trying to make something
  bad happen to me.

* When someone says to do something, I should think about what they
  said, and if it will make good things happen I should do that thing,
  but if it will make bad things happen I should do some other
  thing. That person will not do anything bad to me if I don't do it.

* When someone says to do something, I should think about what they
  said, and if it will make good things happen I should do that thing,
  but if it will make bad things happen I should do some other
  thing. Sometimes that person will do something bad to me if I do
  something else. Sometimes I should still do something else.

* When people say to do some things, I have to do one of these things or
  do something else. When someone says to do something and I don't do
  it, that person will make something bad happen to me. I can't do
  everything everyone tells me to do. I can make good things happen to
  me if I do certain things. Those things might not be things that
  someone told me to do.

When should a person disobey authority? What happens if they disobey
authority? Should they disobey authority even if they will be
punished? Will disobeying authority end their relationship with that
authority? Will that authority sometimes be glad they were disobeyed,
depending on the outcome? You've got to have answers to these questions
that match with the real-world situations you're trying to teach about.

Taking all that into account, one way to model authority/player
relations in a roguelike could be through gods, since they have become a
common feature, and because a player won't lose respect for a god the
way they will for a general. Gods are expected to demand extreme things
of you, but a general is expected to be shrewd.

In DCSS, which provides the bulk of my experience with gods, they are
mostly hands-off: they like certain things, they dislike others, and you
know what they are before you join their religion. But what if they
often gave specific instructions that you couldn't predict ahead of
time? Trog, who hates magic users, might command you to kill a powerful
wizard. This might be more suicidal than enduring a period of
wrath. Gods could even prescribe certain paths for you through the
level, "We will convert the unbelievers by marching, unafraid, straight
through their castle." Do you do it? Do you kinda do it so you don't get
in much trouble? Do you just go do something else?

Just a few thoughts on the topic.

It's the preconceptions talking through you. The whole definition of a
"glitch" is based on the assumption that there is an intended way of
playing the game, that someone designed it, but then didn't implement
it properly.

> I don't think this will effectively teach players to think for
> themselves. Some players may follow the intended route and miss the
> message entirely (less likely). More likely, players will be oblivious
> to the intended route. They won't see themselves as disobeying
> authority, but rather as ignoring the "noise" that are the solution
> marks, or as ignoring lies from their enemies.

That's perfectly fine, except that in this case the solution marks serve
more as danger marks, so the good players will probably still have some
use for them. Of course, this requires some careful planning of the
dangers -- making them in such a way that they are dangerous when you
follow the route, but relatively harmless otherwise. Portal's turrets are
an excellent example -- they are only dangerous when you approach them
head-on.
That's going back to the Portal's blunt, hardwired story. It's the "wink,
wink, nudge, nudge, you have to pretend to disobey us now because that's
how we want it" cliche of escaping the incinerator. Also the "we made sure
that you have an escape route, you just have to look carefully for the
subtle hints". I don't want that.

And the whole idea of a game punishing the player is pretty disgusting to
me. I want my games to be fun.

> Taking all that into account, one way to model authority/player
> relations in a roguelike could be through gods, since they have become a
> common feature, and because a player won't lose respect for a god the
> way they will for a general. Gods are expected to demand extreme things
> of you, but a general is expected to be shrewd.
>
> In DCSS, which provides the bulk of my experience with gods, they are
> mostly hands-off: they like certain things, they dislike others, and you
> know what they are before you join their religion. But what if they
> often gave specific instructions that you couldn't predict ahead of
> time? Trog, who hates magic users, might command you to kill a powerful
> wizard. This might be more suicidal than enduring a period of
> wrath. Gods could even prescribe certain paths for you through the
> level, "We will convert the unbelievers by marching, unafraid, straight
> through their castle." Do you do it? Do you kinda do it so you don't get
> in much trouble? Do you just go do something else?

But then you are not disobeying the level designer -- you are disobeying
a sock puppet that the designer introduced. That's hardly teaching
independent problem solving.

--
Radomir Dopieralski, sheep.art.pl

Once upon a time, Radomir Dopieralski wrote thus:
> On 2013-06-18, Michał Bieliński <dungeon...@tlen.pl> wrote:
>> Once upon a time, Radomir Dopieralski wrote thus:
>>> The designed, easy to read, obvious messages are all against you.
>>
>> Final enemy holding the Bizarro Orgasmatron is the Shodan.  Lots and lots
>> of hit points plus it hits like a tank.  Not to mention special attacks
>> that can slow you, bug (equivalent of curse) your items, sicken you or
>> burn your stuff.
>>
>> Shodan teleports around and will do so to ambush you at inopportune moment.
>> When you manage to whittle boss' health down expect to make a chase all
>> over the place as the monster teleports away immediately after seeing you.
>>
>>> The useful, helping things are all random. You are forced to come up
>>> with your own solutions, instead of relying on guessing the intentions.
>> (...)
>>> There might be an insanely simple shortcut that lets you finish in
>>> 3 moves, or there might be none and you will be forced to follow the
>>> marked path after all.
>>
>> If certain item generates (somewhat below average chance) you can score
>> an instakill against Shodan.  One player profession actually starts with
>> it in basic inventory.  It is very much possible to come close to fast
>> kill by other means too but many rely on stuff that needs to be found.
>

> From what you are describing, this is exactly the opposite of what I tried
> to achieve. In your case, your opponent is random, and you have to rely on
> the static, designed, hand-crafted game elements to somehow alleviate that
> randomness and defeat it. You need to play enough times to learn about the
> game and guess how the author wanted you to behave and what he envisioned.
> On the other hand, I want the helpful things in the game to be unlearnable
> by repetition -- to change every time you play, so that the players invent
> new ways of defeating their obstacles every time. No spoilers are going to
> help you (although surely there will be some inevitable patters).

Yes, a stun grenade is only going to stun creatures in its blast range every
game.  It is static behavior hardcoded right there.  Did you expect grenades
to stun one game and blind in another?  Prime achieves randomness by letting
the RNG decide how many flashbangs, stun grenades or other stuff is given to
player each run.  I think this is enough unlearnable stuff in the game.  For
what your stun grenades will be useful is extremely dependent on the rest of
your equipment.  What use is disorienting your enemies if you cannot use the
free turns to waste them?  If you meet too many stun resistant monsters your
grenades are as good as random trash.

What the author wanted the player to do with Shodan is really easy to guess.
Kill it of course!  While it was not me who put it in game I did think about
possible ways to defeat the final boss.  I remembered there is a way to kill
almost any monster by pouring canister of antimatter in melee range.  I just
decided not to block it.  No intention on my part to force player to go this
way or even make it the best or preferred solution.  Very often the canister
is required to save your life earlier and is best spent then.  Absolutely no
point in dying with the item in your inventory.

So you may think the profession starting with the killer item is bad because
it is predictable and learnable.  Perhaps but most often people fill ray gun
with it the first time they find one.  Some chuck it at a bunch of aliens or
other such dangerous groups.  A few save it for Shodan or similar very tough
foe like Dalek, Unix Daemon or whatever proves to threaten character's life.

Prime tries to make spoilers useless by giving all information about monster
when you do a look command on it.  All attacks are listed with damage ranges
explicitly stated.

--
Michał Bieliński

On 2013-06-20, Radomir Dopieralski <ne...@sheep.art.pl> wrote:
> On 2013-06-19, Kusigrosz <Kusi...@AUtorun.itvk.pl> wrote:
>> On 2013-06-16, Radomir Dopieralski <ne...@sheep.art.pl> wrote:
> [snip]
>
>> The player should be able to customize
>> the UI, but the general structure and the set of things that can be
>> done is fixed.
>
> I strongly disagree here. If you do that, you will never get a good UI.
> The customization should be limited to what is required by differences
> in hardware, and not a thing more, otherwise you risk that no possible
> set of settings is any good.

By customization I mean things like redefining key controls, etc. It is
desirable that the default settings are good, but different people have
different concepts of good - even if they have similar hardware.
I rather like vi keys, but I've read somewhere some people don't...

>> b) Tactical level [snip]
>> c) Strategic level: [snip]
>> d) Game level: [snip]
>
> I'm mostly interested in c), as that has the largest learning opportunity.
> Doing it at strategic level gives you a game of chess, and doing ti at
> game level gives you minecraft -- both very good games in their own right,
> but not that interesting from my point of view.
>
> [snip]
>>> The designed, easy to read, obvious messages are all against you. The
>>> useful, helping things are all random. You are forced to come up with your
>>> own solutions, instead of relying on guessing the intentions. Sure,
>>> guessing the intentions may still be useful, but only to subvert them.
>>
>> But won't that just lead the player to try to read the 'real'
>> intentions? "The arrow points down == Going down hurts, avoid if
>> possible."
>
> Yes. The whole trick is that "go this way" has this "here, I've made all
> the decisions for you, do this, this and this" taste, while "do anything
> but don't go this way" doesn't tell you what to do. All you know is that
> you are expected to disobey (so there is a message at the game level),
> but you are on your own on how to "not go this way".

Still, you communicate your expectations to the player, and the player
is rewarded for following them. That the set of options you expect the
player to chose from is larger than the (non-empty) forbidden one makes
little difference as far as obedience is concerned. "You are free to eat
from any tree in the garden, but you must not eat from the tree of the
knowledge of good and evil" also leaves a lot of choice.

That makes sense. But is that way they're intended to be relevant?
Aren't symbols that are there to be disobeyed a "wink wink, nudge nudge"
pretend disobedience? Anyway, I meant to describe the opposite of that!

> And the whole idea of a game punishing the player is pretty disgusting to
> me. I want my games to be fun.

Think of it more like a consequence. And I didn't say the game should
punish the player, I said the authority they are disobeying should
punish them.

>> Taking all that into account, one way to model authority/player
>> relations in a roguelike could be through gods, since they have become a
>> common feature, and because a player won't lose respect for a god the
>> way they will for a general. Gods are expected to demand extreme things
>> of you, but a general is expected to be shrewd.
>>
>> In DCSS, which provides the bulk of my experience with gods, they are
>> mostly hands-off: they like certain things, they dislike others, and you
>> know what they are before you join their religion. But what if they
>> often gave specific instructions that you couldn't predict ahead of
>> time? Trog, who hates magic users, might command you to kill a powerful
>> wizard. This might be more suicidal than enduring a period of
>> wrath. Gods could even prescribe certain paths for you through the
>> level, "We will convert the unbelievers by marching, unafraid, straight
>> through their castle." Do you do it? Do you kinda do it so you don't get
>> in much trouble? Do you just go do something else?
>
> But then you are not disobeying the level designer -- you are disobeying
> a sock puppet that the designer introduced. That's hardly teaching
> independent problem solving.

You're not disobeying the level designer in your scenario either--you're
pursuing alternate routes, just like they wanted you to. I think to
teach anything about disobedience, you need to have some kind of
authority that *actually* wants to be obeyed, and consequences if you
don't.