Terrible design choice and service

[dowan]

Yeah, more forks is always better, get some different groups trying stuff out. The proof is in the pudding after all, and then we can merge the good stuff together. Except duvessa crawl, because it will only have things removed, never added. But maybe some removals would turn out well.

Dowan crawl is the best crawl, but I can't distribute it due to some issues with pizza tornado copyright. I'd take it out, but it's sort of integral to the game now...

[Sandman25]

agreed, shafts are great



no but seriously I actually tend to like them a lot. trying to find your way back up out of a scary depth is cool+fun

The cool+fun are subjective and thus should be in the same optional category where player ghosts and random god altars are.

[Sandman25]

You know what? I feel a lot better about myself now. Thank you.

That hostility reminded me why I left the forum so many times. Do you really think there are no good players who hate shafts? You missed the point, get out.

[Sandman25]

I survived this situation just fine though.

So what was the challenge then? Is reading tele and then fear and spamming potion of heal wounds cool+fun? That's basically optimal answer for 95+% of shafts which land you in situation like this.

[Sandman25]

Being forcibly placed out-of-position into an unexplored floor full of out of depth monsters (comparative to your level) and being forced to travel back up three staircases, probably triggering timed portals on the way, all because "well you didn't passively detect that shaft trap", is not my idea of fun

70% of the time nothing happens, 15% of the time something mildly scary happens, 14% you get double shafted and die, and then the last 1% is when it actually feels like an enjoyable experience (almost never)

True. Let's add -scroll and -potion for 10 turns when shaft triggers. It will make it equally dangerous to all characters, currently it works as scroll of mapping and scrolls of tele tax. As bonus it will be more cool+fun. No sarcasm, I am serious.

[dpeg]

You know what? I feel a lot better about myself now. Thank you.

That hostility reminded me why I left the forum so many times. Do you really think there are no good players who hate shafts? You missed the point, get out.

The original posting is so badly written, it has hostility spewing out of every word. Look at the thread's title!

It's amazing how far this thread has come, given the atrocious start.

[dynast]

It's amazing how far this thread has come, given the atrocious start.

Id like to take some credit for that, and by that i mean im so sorry.

[PleasingFungus]

Not that the original post implying that appreciating forks reveals a conspiracy by the Developer Illuminati was anything but trolling, but of course the devs behind a fork could take the time to implement something useful for DCSS that we don't have the time/expertise to implement ourselves. The vast majority of changes in the fork might be things the original project doesn't want, but they can incorporate only those changes in the fork they do want, and then the fork has been valuable to the original project.

We'd take reasonable changes/features/implementations from anywhere, including a fork like dcss-ca with very different notions for what's good design. That dev might just find the time to implement some interface, refactoring, or functionality we've been wanting but that took very substantial work or that we simply hadn't thought of in the first place.

yeah, i already tried to steal some code from dcss-ca (shield swapping). that particular code turned out to be buggy and unworkable, but i still want to go through and plunder other things at some point

[Sar]

That hostility reminded me why I left the forum so many times. Do you really think there are no good players who hate shafts? You missed the point, get out.

Come on man, regardless of how you feel on shafts the opening post was obviously worded in a very inflammatory, even trollish, manner. If whoever wrote it expected a serious reply, he probably should've put more thought in the post, and maybe less expletives.

[Sandman25]

I agree with OP that having shafts for new players is terrible design. I believe that disabling shafts before first rune of a player and Zot traps before entering Zot for the first time would be much better. This way new players would concentrate on learning the game and would quit playing less often.
I understand how new players feel about unfair deaths, it cannot be compared to cool+fun of lucky escapes.

No comment about service since I haven't registered on other popular forums for a long time, I am still in 2002

[dpeg]

I agree with OP that having shafts for new players is terrible design. I believe that disabling shafts before first rune of a player and Zot traps before entering Zot for the first time would be much better. This way new players would concentrate on learning the game and would quit playing less often.

I don't think so. It is very hard to estimate how really new players experience the game, or how easily they quit (although we know from the survey that many players keep playing despite never getting a rune).

However, I am sure that while some new players may be scared off forever from something like a paralysis death, or banishment, or a shaft incident. However, other new players will be intrigued by this. We have no chance of knowing which behaviour occurs more often. That's why I think it is best if we (developers) stick to what consider fun. As always, the basic idea is that if it's fun for us, it'll be also fun for enough players.

[Azarkon]

If a feature never kills anybody (assuming good play), then it didn't really exist.

I think everyone can agree that strictly unavoidable deaths, even with perfect play, are bad, because they are a waste of time, and you learn nothing from them.

I think everyone can also agree that the game should be difficult, such that even the best players can and will mess up and die, and everyone is always learning.

These two principles lead to the observation that the only way to achieving both design goals is to increase the skill difference, such that no one can, practically speaking, obtain "perfect play," yet "perfect play" will lead to a win every or almost every time.

In games of all sorts, this is usually done by increasing the width and length of the decision tree, such that it is harder to calculate an optimal path through the game. Since the Crawl decision tree, measured by the amount of turns it takes to reach the end, is already ridiculously long, we can think of this problem instead as reducing the amount of equivalent states and increasing the amount of decisions that a player can make at any given state.

Falling down the stairs doesn't contribute to either of the above in any obvious way, but does cause a lot of unavoidable deaths. Logically then, it should be removed.

It is funny, though, and an argument could be made for it just because of that.

[duvessa]

While shafts add nothing useful to the game, it's ridiculous to say that they cause a lot of unavoidable deaths. You yourself pointed out how much entropy the game produces. For a shaft-related death to have a significant chance of being unavoidable, it would need to occur within the first few turns of the game, and if the player chooses they can go a very long time before stepping on a potentially-shaft-containing becomes necessary.

[dynast]

I think everyone can agree that strictly unavoidable deaths, even with perfect play, are bad, because they are a waste of time, and you learn nothing from them.

Nope. The moment you know beforehand that you gonna win a game you have no purpose playing it.

I think everyone can also agree that the game should be difficult, such that even the best players can and will mess up and die, and everyone is always learning.

Nope. The game can be easy(as crawl indeed is) and have a huge learning curve for the sake of those who want to improve after they already beat it, so that the casuals dont suffer.

These two principles lead to the observation that the only way to achieving both design goals is to increase the skill difference, such that no one can, practically speaking, obtain "perfect play," yet "perfect play" will lead to a win every or almost every time.

Crawl's "perfect play" starts by picking the most viable combo, which already narrows down the game variety a lot, with the other combos becoming "self imposed challenges", so i dare say "perfect play" is not a design goal for crawl.

In games of all sorts, this is usually done by increasing the width and length of the decision tree, such that it is harder to calculate an optimal path through the game. Since the Crawl decision tree, measured by the amount of turns it takes to reach the end, is already ridiculously long, we can think of this problem instead as reducing the amount of equivalent states and increasing the amount of decisions that a player can make at any given state.

Not sure what you mean here, my guess is that you are talking about all the paddling the game have, in which case i agree.

Falling down the stairs doesn't contribute to either of the above in any obvious way, but does cause a lot of unavoidable deaths. Logically then, it should be removed.

As duvessa said, falling shaftes does not cause a lot of unavoidable deaths because thats rare, but a lot of the times that you fall down shafts you end up dying because of the upside down trap detection system.

[any1]

I think everyone can agree that strictly unavoidable deaths, even with perfect play, are bad, because they are a waste of time, and you learn nothing from them.

Nope. The moment you know beforehand that you gonna win a game you have no purpose playing it.

Winning alone isn't everything. Plenty of people have never won and still enjoy playing. Isn't knowing that you're not going to win basically the same as knowing you will? Why do these players play? There are also players good enough that they will almost always win on their first or second try; why do they play? Obviously they enjoy playing the game, which is a much more lasting and fulfulling attitude to have. If you only care about winning, you're going to be either very frustrated or very bored. If random unavoidable losses are enjoyable to you, go play a slot machine or something. A game that is about tactics, strategy, skill and knowledge should not have random unavoidable deaths, even if that means some people can win almost every time. Because again, winning isn't everything, or even the main thing: what happens matters.

Back to the main topic. I think that shafts should not be in the game as they currently are. The reason is not to do with random deaths but because of the amount of tedium that playing around the possibility of a shafting induces. As a player who started in 0.10, took a few years off and started again with 0.18 and current trunk, I am impressed with how much less tedious the game is now, compared to back then. The removals and changes have added up to an experience where playing sufficiently well to win, if not exactly "good," is a much better experience than it was.

I think this is a good direction for the game to be going. There hasn't been an obvious reduction of depth, breadth or difficulty, and removing tedium is always excellent. That's why shafts should be removed or changed. Most of the time the tactics to deal with them are just generally good things to do anyway, like exploring on full hp/mp. But the tactic of walking only where you've already stepped is a level of tedium that seems inconsistent with the overall direction that the game has been going, which, IMO, has been a steady improvement. So to continue in the same direction they should be changed or removed.

Currently shafts act as a "situation generator:" they put you in a different tactical situation than ordinary exploration, which might require different play to resolve. This is in general a good thing: putting the player in difficult and genuinely different situations is a core element of the game. Only the way in which it is accomplished is wrong. I'd argue that the game generates enough difficult situations already that shafts could be removed to no ill effect, but a change which keeps the core idea intact while not encouraging tedious play would be ideal.

Slightly off-topic, but related to removing tedium: One situation that comes up too often is "there's a lot of enemies that I can't deal with all at once." The standard tactic is to try to split them up and take them on one or two at a time. In my experience this situation gets repetitive and therefore tedious throughout a game. Maybe it's not that this comes up very often in a numerical sense, but that these situations take too long to resolve, so most of your time is spent resolving these situations rather than others. In any case, I'd like to see fewer cases of many-monsters and more of interesting tactical limitations that encourage you to play differently, rather than use the same tactic over and over. But overall the game is in a pretty good place and moving in a good direction, and it's been enjoyable to play these recent versions.

[goodcoolguy]

However, I am sure that while some new players may be scared off forever from something like a paralysis death, or banishment, or a shaft incident. However, other new players will be intrigued by this. We have no chance of knowing which behaviour occurs more often. That's why I think it is best if we (developers) stick to what consider fun. As always, the basic idea is that if it's fun for us, it'll be also fun for enough players.

Sorry about the thank, wrong button.

[dpeg]

sillybadguy: This is trolling. You can unthank by pressing the Thank button again.

[Sandman25]

It is very hard to estimate how really new players experience the game, or how easily they quit (although we know from the survey that many players keep playing despite never getting a rune).

However, I am sure that while some new players may be scared off forever from something like a paralysis death, or banishment, or a shaft incident. However, other new players will be intrigued by this. We have no chance of knowing which behaviour occurs more often. That's why I think it is best if we (developers) stick to what consider fun. As always, the basic idea is that if it's fun for us, it'll be also fun for enough players.

I think we agree that there are different new players so it can be optimal to attract both categories. For example, tutorial might disable shafts before obtaining first rune or entering Lair. So new players might play either tutorial MiBe or standard MiBe.

[goodcoolguy]

sillybadguy: This is trolling. You can unthank by pressing the Thank button again.

Oh, great. Thanks for the tip.

[dpeg]

I think we agree that there are different new players so it can be optimal to attract both [soft and hard] categories. For example, tutorial might disable shafts before obtaining first rune or entering Lair. So new players might play either tutorial MiBe or standard MiBe.

There are different new players, but I am still not convinced. Why make the game pretend it's nicer than it actually is?

I believe that the underlying question is something different: what is the goal of developing a (free, open) game like this?

If the answer is "to have as many players as possible", then you are correct, and one should really think about how to not-lose any new players. But since no money is involved, we can afford to keep all the stuff we consider fun. This will lose potential players, but it'll also mean that the "design identity" (for lack of a better term) is more pronounced, which has good effects in itself.

[Psieye]

But since no money is involved, we can afford to keep all the stuff we consider fun. This will lose potential players, but it'll also mean that the "design identity" (for lack of a better term) is more pronounced, which has good effects in itself.

When you are small and free, better to create a polarised reaction than a homogenously lukewarm reaction. Those who don't like crawl's core identity will leave. There is nothing wrong with this.

[dynast]

Winning alone isn't everything. Plenty of people have never won and still enjoy playing. Isn't knowing that you're not going to win basically the same as knowing you will?

Im not even gonna bother reading the rest of your post since you pulled a strawman right off the bat. Players who never won the game play it because they think they can win, or get further than they did last time. Knowing that you are not going to win is the same thing as knowing that you will, but you dont start a game knowing that you are not going to win but thinking that you can win or not, so you have a purpose playing to see the final result.

[dpeg]

dynast: That's a good point. I had fun playing the "unwinnable" 4.2 by Brent Ross (which eventually was won much later by syllogism, but everyone really believed it to be unwinnable). But that just moved the goal: we were aiming to get as far as possible. If you want to, you could argue with high-scores.

I really think that current Crawl is too easy, and that it'd be better (for everyone) if not 99.9% of all games could be won. It would be better to increase difficulty without removing that the game is in principle won at start, but I have no idea how to achieve that. And the idea is not to double monster damage. It's even hard to make the game hard properly!

[dynast]

If you really want the game to be harder you could start by removing exploitable tedious features like "kiting", replacing it with a common consumable(potion of lightfeet) until players get used to it. Of course thats a huge can of worms that is easier said than done. Replace stair dancing with a scroll of magic rope and so on...

[dpeg]

dynast: I've thought a lot about kiting, and discussed it with people. I don't think there are simple solution. Hopefully, there will be *some* solution.

[Rast]

I believe that the underlying question is something different: what is the goal of developing a (free, open) game like this?

If the answer is "to have as many players as possible", then you are correct,

So make the game more like Candy Crush Saga.

[dowan]

I think everyone can agree that strictly unavoidable deaths, even with perfect play, are bad, because they are a waste of time, and you learn nothing from them.

Nope. The moment you know beforehand that you gonna win a game you have no purpose playing it.

You disagreed with something he didn't say. One statement is "If I play perfectly, I should win". The other statement is "If you are guaranteed to win, it's pointless to play". Those are totally different.

Did you know that in chess, you never roll a die to see if the game is unwinnable for you? Both players go into the game knowing with good enough play, they can win. Yet people still play chess!

The challenge of crawl is making it through the game, not whether or not an invisible dice roll meant you were screwed from the get-go.

[dynast]

You disagreed with something he didn't say. One statement is "If I play perfectly, I should win". The other statement is "If you are guaranteed to win, it's pointless to play". Those are totally different.

Those are very open for interpretation, but if i believe "i should win" i expect the "guaranteed win", if i dont know, i can only wish to win.

Did you know that in chess, you never roll a die to see if the game is unwinnable for you? Both players go into the game knowing with good enough play, they can win. Yet people still play chess!

Did you know that chess is completely different from crawl? Also, the assumption that i will actually bother playing chess. Finally, "can win" is totally different from "should win" and "guaranteed win", according to yourself.

The challenge of crawl is making it through the game, not whether or not an invisible dice roll meant you were screwed from the get-go.

Making through the game IS rolling the dice, so i have no clue what you are trying to say.

[removeelyvilon]

I'm terribly sorry but what I parse from this thread is

"Ghosts are in the game because."

and

"If you don't like how we conduct business, go make your own game. With blackjack and hookers.
We are just the lich king on the frozen throne waiting for a rag-tag band of misfit heroes to overthrow us anyway."

I mean, really?

[Sar]

I'm terribly sorry but what I parse from your post is "I'm really bad at parsing".

[dowan]

The original statement was "I think everyone can agree that strictly unavoidable deaths, even with perfect play, are bad, because they are a waste of time, and you learn nothing from them."

If you play the game perfectly, you should win, in an ideal world. The game rolls dice to generate situations to which you have to respond. The idea is the game should never generate a situation to which there is no response that doesn't end in death. If you respond perfectly to every situation, you should win.

[dynast]

The original statement was "I think everyone can agree that strictly unavoidable deaths, even with perfect play, are bad, because they are a waste of time, and you learn nothing from them."

If you play the game perfectly, you should win, in an ideal world. The game rolls dice to generate situations to which you have to respond. The idea is the game should never generate a situation to which there is no response that doesn't end in death. If you respond perfectly to every situation, you should win.

Thats a very naive perspective that should never be adopted, the only ways to always have a situation you can respond to is to make dice rolling trivial(or just get rid of it) so you can calculate before hand if you gonna win or lose, or give the player a option where he doesnt have to roll the dice(see "kiting"), or just expect that the player will develop clairvoyance.

[removeelyvilon]

I'm terribly sorry but what I parse from your post is "I'm really bad at parsing".

https://www.youtube.com/watch?v=TAryFIuRxmQ

[dowan]

The original statement was "I think everyone can agree that strictly unavoidable deaths, even with perfect play, are bad, because they are a waste of time, and you learn nothing from them."

If you play the game perfectly, you should win, in an ideal world. The game rolls dice to generate situations to which you have to respond. The idea is the game should never generate a situation to which there is no response that doesn't end in death. If you respond perfectly to every situation, you should win.

Thats a very naive perspective that should never be adopted, the only ways to always have a situation you can respond to is to make dice rolling trivial(or just get rid of it) so you can calculate before hand if you gonna win or lose, or give the player a option where he doesnt have to roll the dice(see "kiting"), or just expect that the player will develop clairvoyance.

Ok, so right now, by some accepted estimates, perfect play will result in a 95% win rate or better. Yet nobody has such a win rate, because nobody is perfect, and they make mistakes. If that 95% was 100% instead, that wouldn't make the game drastically easier, it would just mean that theoretically perfect player will have 100% win rate instead of 95%. For us humans, the only difference would be that we know for a fact every time we die we could have avoided it.

[genericpseudonym]

I think everyone can agree that strictly unavoidable deaths, even with perfect play, are bad, because they are a waste of time, and you learn nothing from them.

Nope. The moment you know beforehand that you gonna win a game you have no purpose playing it.

Are you saying that you would play the game if you could win every time except when something unavoidably killed you? I don't think the possibility of your 5-year-old niece beating you makes the Candyland board game any more strategically interesting.

A roguelike is interesting because of the possibility to die from imperfect play. People want to look back on how they died and say "I'll do better next time". A game should be hard not because perfect players still lose but because being a perfect player is extremely difficult/impossible. Unavoidable deaths make a game just as boring as unscrewupable wins.

I don't think shafts are "unavoidable deaths" though, and I think "very hard to avoid deaths" are fine. I just really object to the argument that "strictly unavoidable deaths" are necessary to make a game interesting.

[goodcoolguy]

Ok, so right now, by some accepted estimates, perfect play will result in a 95% win rate or better. Yet nobody has such a win rate, because nobody is perfect, and they make mistakes. If that 95% was 100% instead, that wouldn't make the game drastically easier, it would just mean that theoretically perfect player will have 100% win rate instead of 95%. For us humans, the only difference would be that we know for a fact every time we die we could have avoided it.

No, the reason people don't have these super high win rates is not that they are unachievable due to human frailty or whatever. People don't maintain these kinds of win rates because the way you have to play crawl to get them is not fun. It's more like work than playing a video game, except you don't get paid and only a handful of people give a damn that you're doing it.

And if you don't think people who can achieve 95% win rates over a reasonable number of games exist, you're not paying attention.

[dynast]

Ok, so right now, by some accepted estimates, perfect play will result in a 95% win rate or better. Yet nobody has such a win rate, because nobody is perfect, and they make mistakes.

Then why do you accept that estimate? Also, why would you compare winrate with chances of winning? Those are two different things, i have seen a player new at the game have 100% winrate bulding up his streak, that doesnt mean the game is 100% winnable. The best part is that if this game is 100% winnable you will never know, so again, why make estimates?

If that 95% was 100% instead, that wouldn't make the game drastically easier, it would just mean that theoretically perfect player will have 100% win rate instead of 95%. For us humans, the only difference would be that we know for a fact every time we die we could have avoided it.

Again, if the game is 100% winnable you will never know, all you know right now is that the game has unavoidable deaths and that the game is easy, maybe you should think about that instead of building a platonic vision of crawl.

The moment you know beforehand that you gonna win a game you have no purpose playing it.

Are you saying that you would play the game if you could win every time except when something unavoidably killed you?

Trying to put words into my mouth? There is no other way to interpret what i said, it means exactly that "i will play the game if i dont know if i can win". You could argue that playing the game just for the sake of rolling the dice is silly and i will give you that.

I don't think the possibility of your 5-year-old niece beating you makes the Candyland board game any more strategically interesting.

Is this bait? This is worse than comparing crawl to chess.

A roguelike is interesting because of the possibility to die from imperfect play. People want to look back on how they died and say "I'll do better next time". A game should be hard not because perfect players still lose but because being a perfect player is extremely difficult/impossible. Unavoidable deaths make a game just as boring as unscrewupable wins.

Tell me about those wonderful roguelikes and where i can find them.

[genericpseudonym]

The moment you know beforehand that you gonna win a game you have no purpose playing it.

Are you saying that you would play the game if you could win every time except when something unavoidably killed you?

Trying to put words into my mouth? There is no other way to interpret what i said, it means exactly that "i will play the game if i dont know if i can win". You could argue that playing the game just for the sake of rolling the dice is silly and i will give you that.

Even if the game was 100% winnable by a perfect player, you are not a perfect player. (And neither am I, or anyone else.) The fun is in knowing that you aren't perfect and challenging yourself to succeed anyways. Victory should come from personal skill, and so death should come from personal failures. If a game has a significant number of truly unavoidable deaths, then it becomes less a test of your own skill and more a test of your favor with the Random Number God.

I don't think the possibility of your 5-year-old niece beating you makes the Candyland board game any more strategically interesting.

Is this bait? This is worse than comparing crawl to chess.

Azarkon said that "strictly unavoidable deaths, even with perfect play, are bad."
You disagreed, and said that this possibility of unavoidable death is necessary.

Candyland is a board game where all losses are unavoidable losses and all wins are unscrewupable wins. I used it as an extreme example to illustrate the idea that adding unavoidable losses to a game that you would otherwise always win doesn't make it any more interesting.

A roguelike is interesting because of the possibility to die from imperfect play. People want to look back on how they died and say "I'll do better next time". A game should be hard not because perfect players still lose but because being a perfect player is extremely difficult/impossible. Unavoidable deaths make a game just as boring as unscrewupable wins.

Tell me about those wonderful roguelikes and where i can find them.

I'm not saying that there are games where the player can never unavoidably die. But the chance of things like that should be minimized. Like all things, there's a balance, and you eventually reach a point where all the changes you have to make to remove some one-in-a-million chance make the game worse than just leaving it in.
When a player dies, 99 times out of 100 it should be because that player failed to play perfectly. This is true in most popular roguelikes. The gnome-with-a-wand-of-death-on-D:1 is possible but extremely unlikely, which is good enough for me. I still think the small chance of instantly dying to a death ray on your first turn is a bad thing, but it's uncommon enough that it doesn't make the game noticeably worse.

[tabstorm]

dynast: That's a good point. I had fun playing the "unwinnable" 4.2 by Brent Ross (which eventually was won much later by syllogism, but everyone really believed it to be unwinnable). But that just moved the goal: we were aiming to get as far as possible. If you want to, you could argue with high-scores.

I really think that current Crawl is too easy, and that it'd be better (for everyone) if not 99.9% of all games could be won. It would be better to increase difficulty without removing that the game is in principle won at start, but I have no idea how to achieve that. And the idea is not to double monster damage. It's even hard to make the game hard properly!

My 3-step plan to making crawl harder, in a non-annoying way:

1. Remove Depths. It is just a place to rack up XP and artefacts at this point. There is too much XP in the late-game.
2. Make Vaults and Zot harder. They are two of the game's easiest branches at this point. Place some of the easier Depths monsters into Vaults, and some of the harder ones (shrikes, etc.) into Zot. Remove weaker dragons from Zot. Find somewhere to put Hell/Pan/Abyss entrances.
3. Have an actual food clock that is just that - a clock, that does not interact with spells, abilities, and so on.

[dowan]

Speaking of removal of food, I just played a DSGl of gozag, who I found on D2. The stress wasn't about starving, but about having to spend my money on a food shop, but even still, it was actually meaningful when I found a stack of 9 royal jellies.

Now, a DsGl is not a hard combo, so just because I could survive on permafood doesn't mean all characters could, but I think the game really isn't that far from being quite possible with just permafood.

If that was done, food costs should just be removed entirely from all actions, and then food can be the clock it's been pretending to be all this time.

[dynast]

Even if the game was 100% winnable by a perfect player, you are not a perfect player. (And neither am I, or anyone else.) The fun is in knowing that you aren't perfect and challenging yourself to succeed anyways. Victory should come from personal skill, and so death should come from personal failures. If a game has a significant number of truly unavoidable deaths, then it becomes less a test of your own skill and more a test of your favor with the Random Number God.

Sorry, i am a perfect player. I can come up with a fail-proof way of winning a game and execute it, therefore i need the rng. If you gonna give yourself the privilege of saying "a game should do this and that" without demonstrating HOW THE ****, then i have the privilege of considering myself a perfect player without having to demonstrate that aswell.

Candyland is a board game where all losses are unavoidable losses and all wins are unscrewupable wins. I used it as an extreme example to illustrate the idea that adding unavoidable losses to a game that you would otherwise always win doesn't make it any more interesting.

Sorry i never heard of candyland until now.

I'm not saying that there are games where the player can never unavoidably die. But the chance of things like that should be minimized. Like all things, there's a balance, and you eventually reach a point where all the changes you have to make to remove some one-in-a-million chance make the game worse than just leaving it in.
When a player dies, 99 times out of 100 it should be because that player failed to play perfectly. This is true in most popular roguelikes. The gnome-with-a-wand-of-death-on-D:1 is possible but extremely unlikely, which is good enough for me. I still think the small chance of instantly dying to a death ray on your first turn is a bad thing, but it's uncommon enough that it doesn't make the game noticeably worse.

What popular roguelike?

[dynast]

Speaking of removal of food, I just played a DSGl of gozag, who I found on D2.

You could have waited for d:4 to join him. Bad investment.

[dowan]

If you gonna give yourself the privilege of saying "a game should do this and that" without demonstrating HOW THE ****, then i have the privilege of considering myself a perfect player without having to demonstrate that aswell.

Well, I don't know about everyone else, but I've been very careful to use words like "ideally" and "in a perfect world" when talking about a 100% unavoidable death free game. Because clearly I have no idea how to do it, and it's probably not possible without ruining the game in the process (the dev document spells this out pretty well). It's just a general statement that with all else equal, one should avoid introducing more unavoidable deaths, and where possible and prudent, one should remove unavoidable deaths. So, in a perfect world every single crawl game would be winnable with perfect play. It'll never happen, but that doesn't mean it isn't a good goal.

If you are a perfect player, then by playing crawl you're just waiting to see the dice roll of whether that particular game was actually winnable. You can save yourself a lot of time by just rolling some dice if that's what you're after. I personally want to see if I'm up to the task of winning each individual game, not whether the RNG allowed me to win.

[dpeg]

I wrote that paragraph about unavoidable deaths in the philosophy section. That's really old by now [1], but I still don't think there is a good way to create a game with the following tenets: (1) randomly generated challenges, (2) winnable if perfectly played, (3) challenging for players (this is vague, I mean game depth is as high as possible, i.e. game allows for many levels of player skill).

You can achieve (1) and (2) by making the game easy, and one can argue that this is where Crawl sits right now. Classical design has no randomness, then you can go for (2) and (3).

[1] I wouldn't mind updating the philosophy section, if anyone has ideas on it. It's really old by now. You can smell the dust when browsing it.

[genericpseudonym]

Sorry, i am a perfect player. I can come up with a fail-proof way of winning a game and execute it, therefore i need the rng. If you gonna give yourself the privilege of saying "a game should do this and that" without demonstrating HOW THE ****, then i have the privilege of considering myself a perfect player without having to demonstrate that aswell.

Huh? I'm not saying "roguelikes should remove any possibility of unavoidable death, no matter how minute". As dpeg says, it's probably impossible to make a hard, randomly generated game that's still always perfectly winnable. I'm just saying "unavoidable deaths are bad, and games should make an effort to reduce them."

If it's easy to achieve perfect play in a game, then it isn't a very interesting game. But the solution isn't to add (as an extreme example of unavoidable death) "roll 1d20, on a 1 you have a heart attack and die". The solution is to make the game harder, to create situations that require smarter thinking and more thorough decisionmaking in order to survive.

What popular roguelike?

Well I was talking about nethack specifically there, but this is also true about Crawl, Angband, Brogue, Sil, DoomRL, and really most popular roguelikes. A game doesn't really get popular if it has a ton of BS deaths in it. (Unless its name is SotS:The Pit, or so I've heard.)

[dowan]

Well I think that part pretty well states how unavoidable deaths should be handled in roguelike development. If you want sufficient randomness to keep gameplay fresh, you create the risk of outliers that are too much, thus creating an unwinnable game. However, if you dial the randomness back enough to totally avoid that, you might end up making the game too easy. So the philosophy of doing what's possible to minimize unavoidable deaths without compromising the required randomness is a good one.

I think it's a great section to point to when people complain about dying to something unavoidable, and it's a great section to point to when someone says unavoidable deaths in and of themselves are desirable. Unavoidable deaths are a byproduct of a sufficiently random and difficult game. However, none of that prevents making changes like no jackal packs on d1, or things like that, so it's sometimes still quite possible to eliminate some unavoidable deaths without compromising difficulty.

In fact, special code to prevent an unwinnable situation on D1 means you can afford to dial up the randomness, making everything else harder, because the only reason it wasn't harder in the first place was that it created too many unavoidable deaths in the very early game.

You're probably right that it's impossible to have all this at once, but clearly there's a balance. A game where you lose 99% of the time after your first move, but that 1% where you don't is winnable, still isn't a good game.

[dynast]

If you are a perfect player, then by playing crawl you're just waiting to see the dice roll of whether that particular game was actually winnable. You can save yourself a lot of time by just rolling some dice if that's what you're after. I personally want to see if I'm up to the task of winning each individual game, not whether the RNG allowed me to win.

I play the same way you do, but i respect the unfairness of the rng because im aware it brings more game variety and depth, also because it IS a dice game. I cant see this game as a methodical road to success, at least i cannot see that and being fun at the same time. What i can see is places where the rng fails to deliver game depth and variety and thats what i argue to be changed/removed(in this topic, shafts).

The devs can have whatever goal they want(as they should), but until we get there i dont see it.

Well I was talking about nethack specifically there, but this is also true about Crawl, Angband, Brogue, Sil, DoomRL, and really most popular roguelikes. A game doesn't really get popular if it has a ton of BS deaths in it. (Unless its name is SotS:The Pit, or so I've heard.)

I only tried Sil and SoTs:The pit and i honestly hope crawl stays away from becoming that.

[CanOfWorms]

if you want a randomized game that's always winnable i recommend solving the sudokus in your local newspaper (remember to use a pen)

[CypherZel]

Hhioii

[Shtopit]

if you want a randomized game that's always winnable i recommend solving the sudokus in your local newspaper (remember to use a pen)

I just realized you can't solve a sudoku with a solvent, but only the ink with which it's written. Let me bask in the radiance of this epiphany.