There's also a principle that I think can be attributed to Michael Brough (it's there in ZAGA-33 and 868-HACK), that dungeon randomization should already be enough to guarantee variety. You can have completely deterministic combat and still have every battle turn out differently, because they all start from different initial conditions. (I think these two games still use randomness to determine enemy movement, but even this could be made deterministic.) Hoplite is like this too, plus no random enemy movement if I recall. So it's all just dungeon randomization + the player's input that determines the result of the game.

(I have a roguelike idea in my "idea folder" where it's *only* the player's input that matters. The player starts in a large, mostly empty room, and the dungeon is progressively generated offscreen using the player's first N movements as input.)

But that's still "random" - I mean yes, strictly it's deterministic, but that's true of all the random number generators in the games being discussed, they're pseudo-random rather than random. Even if it's seeded from player movement, I would still call it random if I can't reasonably predict the outcome from the inputs.

Unless I'm misunderstanding what you're describing?

To some degree, however, I still feel that, for me at least, one of the main reasons that your suggestion would help is that it reduces the effect of the dice. It means that I'm going to miss every so often (which is fine), rather than on every third or fourth attack (which I find frustrating). To me, the former feels as though I'm largely in control, with a small extraneous force spicing things up; the latter feels as though I'm not in control, the dice are.

That's obviously a balancing issue, but though, oftentimes, people find it easier to reduce the difficulty of the roll (making it possibly less likely to happen), I find that a normal distribution also reduces the risk of a bad streak. To me, a bad streak is something that really turns me off, and makes me hate a game I otherwise might have loved.

Worst yet, even if people test their game's RNG to death, they might not experience as bad a streak as a first-time user might because of sheer odds of that happening. It is still a game breaker though.

A normal distribution can keep things hard, but not out of reach, whereas trying to reduce the difficulty of a single die might make it mundane and have the player wonder why there are dice at all.

Are you arguing that knowing the model used (such as the 3d6 rolls that you described) makes it easier for the player to accept the results of the die-rolls, or are you saying that this technique, which reduces the appearance of extreme values, produces results that are more palatable?

I would keep that information hidden (how the actual rolling works). I think the results speak for themselves for an average user and limit the risk of very bad outcomes.

there's little apparent space for player skill in the game

Heh, play ADOM or Crawl and tell me there is a little skill needed :D

I think you simply tried the wrong games. The system definitely is skill based and it definitely works. Try most ASCII roguelikes, you will see it there. But I agree that the modern "casual" graphical "roguelikes" (actually, these are more like RPGs, not roguelikes) have it watered down too much.

Also don't forget savescumming, it completely changes the gameplay and risk assessment. The "I have a 10% chance I will instantly die if I do this" is an important part of roguelikes, it's not about luck, it's about skill (risk assessment and decision when to take it).

Note, if you "fix it" the game will be not casual anymore, and that's probably the core of the issue (devs don't find it attractive to fix it).

That's obviously a balancing issue ...

I do think that I agree with you here.

(For the sake of clarity, I'll note here that with regards to this next quote I'm somewhat arguing Devil's Advocate.)

... I find that a normal distribution also reduces the risk of a bad streak. To me, a bad streak is something that really turns me off, and makes me hate a game I otherwise might have loved.

But if bad streaks reduce the fun of a game, would it not be better to remove that extraneous source of bad streaks altogether? After all, player mistakes and level randomisation can still cause runs of untoward outcomes.

(/Devil's Advocate.)

A normal distribution can keep things hard, but not out of reach, whereas trying to reduce the difficulty of a single die might make it mundane and have the player wonder why there are dice at all.

Hmm... I take it that a "hard" encounter would be one in which the player's probability of landing a hit is low; in that case, surely even a normal distribution will yield runs of misses?

That said, I do agree that a good distribution is likely to rather improve the results "rolled".

I think the results speak for themselves for an average user and limit the risk of very bad outcomes.

I infer then that you're saying that the reduction of bad streaks produces a more palatable result--is that correct? (I don't want to jump to that conclusion without checking, in order to reduce the risk that we're miscommunicating. I'm also somewhat under-slept, and so am not confident of my judgement.) If so, then I do think that I agree.

Heh, play ADOM or Crawl and tell me there is a little skill needed

I haven't played ADOM, I don't think. As to Crawl, do you mean this game? If so, then it looks like something different--if not (as I'm guessing is the case), could you clarify which game you mean, please?

Try most ASCII roguelikes, you will see it there.

Like, say, Rogue itself? That, as I recall, was where I first noted this issue. :P

But I agree that the modern "casual" graphical "roguelikes" (actually, these are more like RPGs, not roguelikes) have it watered down too much.

Hum... I'd like to be clear, to avoid misunderstanding: when you refer to "watering down", are you referring specifically to the combat-resolution mechanics? The mechanics in Pixel Dungeon seem pretty much the same as those in Rogue: you bump into an enemy (or press an on-screen button) to attack; the result of that attack is based on random numbers and character statistics. There's space for some tactical decision-making, such as retreating beyond a door, quaffing a potion or when and how to use resources like wands.

Also don't forget savescumming ...

Erm, I'm not sure of what you're saying here: are you suggesting that I'm save-scumming in games like Pixel Dungeon (I'm not), or are you recommending it (which seems unlikely, given the rest of your post)? o_0

Note, if you "fix it" the game will be not casual anymore, and that's probably the core of the issue (devs don't find it attractive to fix it).

I suspect that this depends somewhat on how you "fix" it. One could, for example, swap out random combat resolution for a simple, quick minigame (this specific change is off the top of my head, and so may not actually be a good idea) and end up with something that's still "casual", but which replaces highly-randomised combat resolution with player skill.

I'm not convinced that being a "casual" game is the problem; if there's a difference between such roguelikes and others, then I feel that it's more likely a design issue than a genre issue.

I think you simply tried the wrong games.

I would welcome recommendations! ^_^

However, if you do recommend anything, I do have a few restrictions on what I'm likely to enjoy. (I apologise if this makes the request onerous. :/) My restrictions (off the top of my head) are these:

• I'm not looking for something into which to sink enormous amounts of time.
• I don't really want anything with the sheer massive complexity of NetHack.
  • Rogue or Pixel Dungeon might be examples of the sort of complexity that I might look for.
• As to OS, Android would be preferred, but Linux is also fine.
  • I do have Windows XP, so I could theoretically play Windows games, but I have a habit of forgetting about games installed there, with a few exceptions (such as Shadowgate).
• I do like graphics--whether pixel or otherwise. ASCII games are not entirely unacceptable, but it might be harder to convince me to play them.
• Ideally, I'd prefer a free game, but I might consider games up to, say US$2.
• It's possible that I'll turn something down on a matter of personal taste, ethics or beliefs, etc.

But if bad streaks reduce the fun of a game, would it not be better to remove that extraneous source of bad streaks altogether? After all, player mistakes and level randomisation can still cause runs of untoward outcomes.

(/Devil's Advocate.)

I think it's not good for the player to experience too much failure, but failure is one means to teach the player that failure is POSSIBLE.

Knowing the failure is POSSIBLE is what makes success that much more worthy.

Of course, if players believed the developer that failure is possible, but that failure was not possible, then players would ultimately enjoy the game much more, but in my experience, players have to face such failure at least once to truly value their successes.

Furthermore, failures insure the player has learned gameplay mechanics (generally speaking). As such, in super mario, you can't clear the first section until you've mastered the jump. Even if your miraculously/accidentally jump over the first enemy, you still need to jump over the first pipe. Failure to do so would result in time out or being hit by an enemy (failure).

In the context of a roguelike, a bad streak at rolling teaches the player that odds are still odds, and you can still wake up one day and lose a fight against an inferior opponent because the conditions were difficult. While you had a sword and armor, the wolf managed to kill you because x or y.

It is not desirable that such outcome occur often, but having the player go through one such failure would not be sufficient to have them complain that the game balance was off, and yet it would add value to their exploits (knowing danger lurks).

Hmm... I take it that a "hard" encounter would be one in which the player's probability of landing a hit is low; in that case, surely even a normal distribution will yield runs of misses?

That said, I do agree that a good distribution is likely to rather improve the results "rolled".

I would argue that a balanced encounter should have at least 75% hit (and hopefully more).

In an RNG system I had created for a prototype, I actually ended up cheating the odds. I kept scores on the last 50 or so hits, tallying the real hits vs misses ratio. If I would reach a point where this was going downhill, I would automatically connect a hit. Likewise, if the value went too high, I would automatically enforce a miss. This was an added layer that simply removed both undesirable scenarios: someone that hits or misses all the time.

I infer then that you're saying that the reduction of bad streaks produces a more palatable result--is that correct?

I think that what creates more palatable results isn't specifically the concept of bad or good streaks, but the actual breadth of the values being returned.

Let me explain with numbers.

Assume a D&D environment using a d20, rolling 10 times.

You might get these results:

2, 17, 10, 4, 16, 19, 4, 9, 20, 4

When looking at that distribution, the only logical conclusion is that they are all numbers between 1 and 20 (inclusively) without much coherence.

In a 3d6 environment, you're more likely to see:

(3-18)

9, 7, 11, 15, 12, 6, 10, ...

I'm exaggerating here obviously, and the range of 3d6 is marginally lower than 1d20, but it goes to examplify my thought pattern.

While these numbers may be hidden to the player, they will still form expectations. Assume that a 10 allows them to hit an armored soldier, and a 6 allows them to hit a wolf, whereas a 15 is required to hit a demon.

They'll come to think that they can reliably hit wolves, because 6 and less is very hard to roll on 3d6.

The armored soldier will be a bit of a 50%/50% standing roughly at the center of the normal distribution. If it were only 1 point lower, it would get hit a lot more suddenly.

The demon would be almost impossible to hit, whereas in a d20 universe, it would still occur more often.

In essence, 3d6 diminishes the odds of something "crazy happening", and returns more "normal" numbers. There's a stats term for this, but I'm afraid I don't know the english word equivalent...

I don't really have anything of my own to contribute to the discussion at this point, but I wanted to post to add a couple of relevant links:

• Understanding randomness in terms of mastery
• "Not so random randomness" in game design and programming
• Probability and games: damage rolls < be sure to play around with the interactive diagrams!

The last link in particular goes into quite some detail on what Orymus3 has been talking about with demonstrations of different probability curves and different adjustments that can be made. Hope those are helpful! :)

I haven't played ADOM, I don't think. As to Crawl, do you mean this game? If so, then it looks like something different--if not (as I'm guessing is the case), could you clarify which game you mean, please?

The world I knew ends :) For sure :D

Crawl: http://www.dungeoncrawl.org/

Like, say, Rogue itself? That, as I recall, was where I first noted this issue.

Well... Rogue was the very first one. So it was not really the best one...

Hum... I'd like to be clear, to avoid misunderstanding: when you refer to "watering down", are you referring specifically to the combat-resolution mechanics? The mechanics in Pixel Dungeon seem pretty much the same as those in Rogue: you bump into an enemy (or press an on-screen button) to attack; the result of that attack is based on random numbers and character statistics. There's space for some tactical decision-making, such as retreating beyond a door, quaffing a potion or when and how to use resources like wands.

Sigh... Definitely, the world I knew has ended :) To compare Pixel Dungeon with a real roguelike...

You know, I realized that nowadays people visualize somethings different when they say "roguelike" than we did back then.

Sure, the mechanics are there, but... it's not the same. It's not how you would describe ADOM/Crawl/Nethack. To make it clear, random number during a combat was NEVER the cause of you losing the game. It was due to your other decisions (when to use resources, proper risk taking, skill, etc). I rarely felt "stupid RNG killed me" insted I thought when died "how could I be that stupid AGAIN" :)

It was lack of focus that was killing me most of the time, not RNG.

Erm, I'm not sure of what you're saying here: are you suggesting that I'm save-scumming in games like Pixel Dungeon (I'm not), or are you recommending it (which seems unlikely, given the rest of your post)? o_0

Save scumming is considered "breaking the game" in traditional roguelikes which are meant (and actually makes sens only) with permadeath.

You know, I feel talking about it is pointless. You need to feel it :) Play Nethack, Crawl or ADOM and see how you feel yourself.

I think it's not good for the player to experience too much failure, but failure is one means to teach the player that failure is POSSIBLE.

...

Furthermore, failures insure the player has learned gameplay mechanics (generally speaking).

I agree, but that doesn't mean that any form of failure is good: I'm inclined to argue that it's still important that the mechanic that produces the failure be one that isn't frustrating to the player.

For an extreme example--and intentionally moving away from random numbers for a moment--let's say that at arbitrary (but set) points, we simply have the player character suddenly keel over dead. There's no enemy involved and no player input: the character just dies, albeit in a manner similar to a death in battle. We also make a point of having one of these points be very early in the game. Since this provides failures, and the early instance demonstrates that failure is possible, is it a good mechanic?

They'll come to think that they can reliably hit wolves, because 6 and less is very hard to roll on 3d6.

...

True, but this still leaves us with the problem that I can't hit something "just because the dice say so": just as I'm frustrated fighting ordinary enemies in Pixel Dungeon, I imagine your third enemy as being frustrating to encounter; I suspect that my response to that foe would be likely to be less "Ah, a challenge!" and more "Argh, not one of these things!".

In TV Tropes terms, they sound (appropriately) like what that wiki calls "Demonic Spiders".

One could perhaps ensure that the player only encounters them when sufficiently levelled--but if we ensure that the player only encounters them when they're capable of scoring hits, then why not simply give the player a set damage range (perhaps scaling up as the character levels) and provide tougher enemies with one or more of higher health, greater damage, more abilities, or better AI?

On the other hand, admittedly, if there were some way to prepare reliable means of combatting them (stocking up on Holy Water, or keeping in reserve a sword that hits them more often, or whatever), and they're not apt to spring out of nowhere in just any level, then that might be fine: it becomes a challenge of preparedness, with a frustrating fight being the result of not having thought ahead.

(For example by analogy, there's a (somewhat annoying) enemy in Delver that has a long-range attack and a habit of both moving away from the player and weaving from one side to another. If the player had no access to ranged weapons, this would likely be frustrating. However, wands are not uncommon in Delver, and don't require identification, so I've learned to keep some in reserve for use against these foes.)

Similarly, if the player couldn't hit the thing at all, and either required something special (a magic sword or a plot item, perhaps) or was required to simply run away, then that too would seem more palatable to me.

... and you can still wake up one day and lose a fight against an inferior opponent because the conditions were difficult.

But in the cases that I'm finding frustrating (and it should be pointed out here that others have indicated that I've perhaps simply been playing the wrong games), I'm not losing because of untoward conditions, but because of arbitrary dice-rolls. I would expect that a hypothetical "perfect" player might be able to adapt to hostile conditions, but even a "perfect" player will presumably not beat the dice if the rolls run against them.

In an RNG system I had created for a prototype, I actually ended up cheating the odds. ...

That's an interesting system, actually! ^_^

I think that what creates more palatable results isn't specifically the concept of bad or good streaks, but the actual breadth of the values being returned.

Ah, I see, and do agree, I believe--thank you for the clarification. ^_^

The last link in particular goes into quite some detail on what Orymus3 has been talking about with demonstrations of different probability curves and different adjustments that can be made.

Thank you for those. ^_^

For what it's worth, I do think that I follow the mathematical arguments that Orymus is presenting. My uncertainty was more with what conclusions were being drawn (which has now been clarified) than with the supporting arguments themselves, I believe.

Crawl: http://www.dungeoncrawl.org/

Ah, yes--I've heard of it under its full name, but I don't think that I would have likely guessed it from the abbreviated form. (I think that my second guess was Dungeon Crawl: Stone Soup.)

Well... Rogue was the very first one. So it was not really the best one...

Heh, fair enough, and a good point!

(Although as an aside, according to TV Tropes it's not quite the first game of its type: TV Tropes lists dnd (an adaptation of Dungeons and Dragons, not the tabletop game itself) as an earlier game containing much of what would become the "roguelike" genre--although Rogue is still the "Troper Namer" and "Trope Maker".)

Nevertheless, Rogue is still presumably a member of the genre, and thus a similar game (such as Pixel Dungeon) surely isn't disqualified from the genre for elements that are present in Rogue.

To compare Pixel Dungeon with a real roguelike...

I'd rather not go too deeply into the definition of the word "roguelike"; suffice it to say that I don't like the stricter definitions, and feel that too-strict a definition of a genre can straight-jacket it, constricting its development.

As a reference point, I'm on the fence about whether Delver (which uses action-RPG combat mechanics) is a roguelike or roguelite; my intuition is to call it the former (if a relatively easy one), but the combat leaves me equivocal, perhaps leaning towards roguelite. As far as I'm concerned Pixel Dungeon is very much a roguelike--whether it's a good one is another question.

I'll note that outside of this thread I seem to keep seeing praise for Pixel Dungeon, both in Play Store reviews and otherwise. I even happened to stumble upon a recommendation of it when looking for something else recently.

To make it clear, random number during a combat was NEVER the cause of you losing the game. It was due to your other decisions (when to use resources, proper risk taking, skill, etc).

That, admittedly, I'm glad to read. ^_^

Save scumming is considered "breaking the game" in traditional roguelikes which are meant (and actually makes sens only) with permadeath.

That was my impression, which was why I was unsure of what you were saying: as I read it, your comment seemed almost to be recommending save-scumming, or suggesting that I was engaging in it. (I haven't save-scummed in a roguelike for a long time, as far as I recall.)

Play Nethack, Crawl or ADOM and see how you feel yourself.

I may do that. Nethack, as I've mentioned, I recall finding overwhelming in the number of actions available (albeit that this was quite some years ago now, I believe). I'm not sure of whether I've played Linley's Dungeon Crawl before, but I might give it a shot now that you've pointed it out. Thank you! ^_^

When I was playing roguelikes my decisions/dilemmas were like that:

"do I put this unidentified amulet? It can be of strangling which would kill me (or do I have means of removing it? like droping acind on oneself or something), how many of amulets I already identified (if I already identified the of strangling then I'm safe of course). But still, if I'm not sure, maybe I still shout put it on since it might grant some critical survival traits?"

"where is this stupid beholder so I can slash it and eat it so I can get the ESP" (Nethack)

"do I go to rescue the puppy?" (ADOM - some quests were time based, you simply could not go for all quests, not to mention you were time pressed due to corruption system)

"how can I secure the second source of fire resistance so I can visit the fire temple?" (ADOM)

" I was swallowed by a vortex! (the screen changes and I'm inside a moster) I need to kill it fast!" (Nethack)

"oh no, the bugs/ants would they overwhelm me of do I kill them all before they breed" (Crawl)

"do I want to polymorph into something?" (Nethack)

As you see, the combat RNG was not an issue here at all Never.

I mean, really, when I see games where you move around some dungeon quaffing potions and killing monsters and claiming it's anything like the old roguelikes... nope, they are not. The basic mechanic was not what these games were about.

That being said, note you have not instantly recoginzed "Crawl", one of the best & most popular roguelikes ever, while, if I mentioned some watered down casual "roguelike" that is stretching the genre to the limits like "Diablo", you would instantly know what game I'm taking about

Good roguelikes are a tiny niche, these don't sell. Make something stupid like Diablo, it will sell.

A player who finds NetHack too complex and Rogue too random is on the wrong side of the learning curve and possibly unaware that there is one; and of course a player who considers the threat of dying of bad luck a "bug" rather than a thrilling motivation to manage risk shouldn't play roguelike games.