Shafts - Too common now?

[Shard1697]

My memory may be failing me but I'm pretty sure that part of the reason I started this thread was because I got shafted three floors on swamp:1 despite having boots of flying active and died a few turns later to a pack of Hydras/Blink Frogs I got landed next to.

So While I like your idea, flying, somehow, does not prevent you from getting shafted.

I think there's a message like "You get sucked into the shaft".

Those hungry, hungry shafts.

I don't really like that all flight does is let you ignore water(and sometimes lava, aka deep water version 2). Not being able to avoid traps while in the air makes no sense. To me that either means flight should be reflavored to something else, like some sort of aquatic buff that lets you move freely in water specifically, or removed. I don't think something that only lets you ignore water(and/or lava) is very interesting and would rather it was just removed as a player available effect.

[IveGoneSupine]

I get the feeling that flight gives very conflicting messages, but I don't understand the ultimatum of change or remove. It does have a narrow application, but so do things like potions of degeneration, scrolls of torment, and scrolls of random uselessness. Unlike those things, flight gives people who make maps the opportunity to utilize it in interesting ways. It's just unfortunate such an application isn't used more.

Regardless of that, I do like the idea of expanding flight into a means of shaft immunity.

[TheMeInTeam]

Torment has some devastating positive use cases, doesn't belong alongside degeneration (strictly to disincentivize quaff ID) and random uselessness (an ID pressure early game at best). Not only for undead species, but anybody with lig potions also.

[MisterPersonMan]

Just make every tile a shaft.

No more surprises.

[acoolguy]

I do agree that I would prefer deliberate and mandatory hatches for specific branches or potentially even randomly as they are quite a bit more interesting than random shafts, which can range from game ending (early shafts especially with weaker races like mummies), annoying (in the case of late game characters that can simply burn a few magic mapping and make their way to the stairs), annoying (due to multiple shafts on the same floor), or interesting (shafts that put you in tense encounters that may require a few consumables to burn). The main changes I would consider about current shafts is to have a shaft time out of some kind (to prevent chain shafts taking you more than 3 floors or chain shafts where you go back up to the original floor and explore again and get shafted again), tone down the frequency of shafts to an average of 1-3 per game and introduce floors without up/down stair cases that instead have up and down hatches.

[ebering]

I can say this until I'm blue in the face. I just checked the online database.

For 0.23 the current average shafts per won game is 2.24 (down from the 2.5 when I checked on just the tournament games).

[VeryAngryFelid]

I can say this until I'm blue in the face. I just checked the online database.

For 0.23 the current average shafts per won game is 2.24 (down from the 2.5 when I checked on just the tournament games).

IMHO currently you are saying something like average number of deaths per won game is zero and thus we should increase difficulty. I mean when character gets shafted just once from D:1 to D:3, it can easily die. If then it is shafted again from D:3 to D:5 while still being XL 1-2, it is very likely to die. And those are just 2 shafts, below "average".

[Sprucery]

It would be interesting to know the average number of shafts per lost game.

[VeryAngryFelid]

It would be interesting to know the average number of shafts per lost game.

Why? Many (most?) characters die on floor 1-2 without any shafts. Does it show if shafts are dangerous or not?
Probably it is possible to find characters who died to shafts (shafted and then died in 20-30 turns) and compare them in different versions.

[acoolguy]

I can say this until I'm blue in the face. I just checked the online database.

For 0.23 the current average shafts per won game is 2.24 (down from the 2.5 when I checked on just the tournament games).

Hmm. Then maybe the frequency of shafts are ok - It could just be confirmation bias where I have had games with what felt like an excessive amount of shafts. I still do think there should be a some sort of shaft timeout though - I've had instances (seems to occur most commonly in lair branches and elf) where I get shafted go back to the original floor and get shafted again, which mostly just ends up being annoying.

[Turukano]

I can say this until I'm blue in the face.

I found a foto of you!

[Sprucery]

I guess the metric we want is "proportion of games where the character died after shafting before getting back to the original level". There should be an increase in this value, but whether the increase is too big would then be for the devs to decide.

[VeryAngryFelid]

I guess the metric we want is "proportion of games where the character died after shafting before getting back to the original level". There should be an increase in this value, but whether the increase is too big would then be for the devs to decide.

I believe there is no milestone "Returned back to level shafted from" so the metric cannot be obtained.

[TheMeInTeam]

I guess the metric we want is "proportion of games where the character died after shafting before getting back to the original level". There should be an increase in this value, but whether the increase is too big would then be for the devs to decide.

I believe there is no milestone "Returned back to level shafted from" so the metric cannot be obtained.

You'd need something like that though. If you just compare shaft deaths in .23 to .21 or something you're going to get a similar rate of deaths per shaft, because the *result* of a shaft is similar. To really compare we'd need a "percent of shaft deaths compared to all deaths" or some such, between versions...with a way to consistently identify shaft deaths.

For gameplay purposes even that probably wouldn't be good enough. Players can die after a shaft, just as they can die from taking one too many steps forward. The time trap deaths become cancerous/bad for gameplay is when they result in unavoidable deaths. These are rare and it's not easy to get a metric on which scenarios are truly unavoidable. Getting 2x shafted into an OOD cyclops before you could have reasonably read-ID'd the few scrolls you had and dying in 1 hit is unavoidable. Dropping down one floor with the option to use wands or identified scrolls/pots to reliably survive isn't. There's a pretty wide grey area between such extremes.

[IveGoneSupine]

I can say this until I'm blue in the face. I just checked the online database.

For 0.23 the current average shafts per won game is 2.24 (down from the 2.5 when I checked on just the tournament games).

Yeah, that stat addresses the original question of whether shafts are more common, but only in the context of people who won the game. It doesn't address any of the other questions raised in this thread regarding the application of shafts as a mechanic. It also makes it seem like patches are only intended to impact people who are capable of winning the game.

If you really feel like relying on numbers, how about looking at these metrics...

Determine patch impact on difficulty: Losses vs. Wins, Win rate of different calibers of players, Mean/Median/SD Length of Streak, Number of Streaks
Determine shaft impact on difficulty: Turns between Shaft Event and Death, Losses w/ Shaft Event vs. Losses w/o Shaft Event

[duvessa]

Crawl's design philosophy specifically says that having some unavoidable deaths in the game is okay. So even if you manage to find a case where a shaft has actually caused an unavoidable death, that wouldn't be a reason to nerf or remove shafts.

[svendre]

Crawl's design philosophy specifically says that having some unavoidable deaths in the game is okay. So even if you manage to find a case where a shaft has actually caused an unavoidable death, that wouldn't be a reason to nerf or remove shafts.

What I see is that people are discussing the difficulty level. I don't understand your point.

You're saying essentially something like: if there were a game element that forced you to instantly die with a 1 in 3 chance each time you took a step, that because there are unavoidable deaths in crawl, there should be no reason to remove or modify that mechanic.

[duvessa]

There could be plenty of reasons to remove or modify that mechanic, but "this mechanic has the potential to cause an unavoidable death" would not be one of them. And my reason for that isn't "there are unavoidable deaths in crawl already" (I don't even believe there are unavoidable deaths in crawl currently), it's that the design philosophy is very clear that having the potential for unavoidable deaths is fine:

The possibility of unavoidable deaths is a larger topic in computer games.
Ideally, a game like this would be really challenging and have both random
layout and random course of action, yet still be winnable with perfect
play. This goal seems out of reach. Thus, computer games can be soft in the
sense that optimal play ensures a win. Apart from puzzles, though, this
means that the game is solved from the outset; this is where the lack of a
human game-master is obvious. Alternatively, they can be hard in the sense
that unavoidable deaths can occur. We feel that the latter choice provides
much more fun in the long run.

Instead, the arguments against it would be things like "this mechanic causes an excessive number of unavoidable deaths" (shafts don't) or "this mechanic doesn't have any counterplay" (shafts do).

You're right that many people are talking about difficulty instead, but I thought it was already well-known that the devs have no problem with making the game harder, so I didn't feel any need to clarify it.

[IveGoneSupine]

Yeah, but harder for who? ebering just showed that people who win are seeing less shafts per game. Why should the game be easier for people at the top end of the bell curve? Why is that the standard being used?

[svendre]

There is more than one way to make a game harder. Increasing the random variables where death becomes a higher probability is one way, but there is also a way to force players to play better to deal with difficult situations which have more ways to react to them. I do believe there are unavoidable deaths. I've had some games where you start out on D:1 and the first things you encounter are a mob of jackals and other stuff, or you run into a guy with an acid wand and you're just toast. Maybe the random number generator could have produced an amazing set of rolls such that you would have survived, but in the cases where it didn't, and you didn't have any other options because you were cornered without the raw ability to win against the sheer volume of damage, it was for those games, unavoidable.

I think most people accept that a dungeon crawler has a random aspect to it, so unavoidable deaths are part of the genre. If your point was really just that there shouldn't ever be changes made based on the sole fact that there are unavoidable deaths, I guess I didn't understand why that needed to be said. I don't recall anyone saying there should be no unavoidable deaths in crawl??

[TheMeInTeam]

Crawl's design philosophy specifically says that having some unavoidable deaths in the game is okay. So even if you manage to find a case where a shaft has actually caused an unavoidable death, that wouldn't be a reason to nerf or remove shafts.

That's not quite what the philosophy implies. In general, crawl's mechanics/design try to be fair and overwhelmingly succeed. Because there are so many interactions, however, it is impossible to guarantee against extreme statistical edge cases when enough interactions stack up. Rather than damaging the game by trying to guarantee against them the development message effectively says that these rare cases are preferable to the cost of trying to remove them (hence the "out of reach" part of your quote).

That's quite different from intentionally implementing heavy RNG, which is inconsistent with crawl's actual mechanics. Fortunately, that doesn't seem to be what shafts are.

Best as I can tell in actual crawl experience, shafts are a candidate for least dangerous traps after D:1 or D:2 or so (that probably still goes to net traps though). However, if for some reason they were significantly more RNG-lethal there would in fact be a legit case for removing or reworking them. There just isn't evidence to support such an assertion.

[Siegurt]

Crawl's design philosophy specifically says that having some unavoidable deaths in the game is okay. So even if you manage to find a case where a shaft has actually caused an unavoidable death, that wouldn't be a reason to nerf or remove shafts.

That's not quite what the philosophy implies. In general, crawl's mechanics/design try to be fair and overwhelmingly succeed. Because there are so many interactions, however, it is impossible to guarantee against extreme statistical edge cases when enough interactions stack up. Rather than damaging the game by trying to guarantee against them the development message effectively says that these rare cases are preferable to the cost of trying to remove them (hence the "out of reach" part of your quote).

That's quite different from intentionally implementing heavy RNG, which is inconsistent with crawl's actual mechanics. Fortunately, that doesn't seem to be what shafts are.

Best as I can tell in actual crawl experience, shafts are a candidate for least dangerous traps after D:1 or D:2 or so (that probably still goes to net traps though). However, if for some reason they were significantly more RNG-lethal there would in fact be a legit case for removing or reworking them. There just isn't evidence to support such an assertion.

It sounds like you're both saying "Whether shafts are overly leathal or not is irrelevant to the question of whether or not they should be removed"

[TheMeInTeam]

^ I think that's not exactly my argument. I they *were* overly lethal (like despite optimal play 80% of the time you died to them or something) that would not be irrelevant, it would be grounds for removal or a massive rework outright. In reality that number is almost certainly < 5%, probably much less, which is what makes it irrelevant in this case.

That by itself isn't enough to say they're a good mechanic though, just that they don't fail a check on whether they're egregiously out of line with crawl in general. I mostly don't have a problem with them, other than how the change to traps made them more unintuitive. I also wouldn't care if they got removed, they don't add or detract from crawl enough for me to really push for something beyond a trap system where an unspoiled player/spectator isn't very confused.

[sardonica]

You're saying essentially something like: if there were a game element that forced you to instantly die with a 1 in 3 chance each time you took a step, that because there are unavoidable deaths in crawl, there should be no reason to remove or modify that mechanic.

He's not essentially saying that. What a silly exaggeration.

There's literally no way to prevent all unavoidable deaths without diluting a number of systems which keep DCSS fresh through multiple playthroughs.

[svendre]

You're saying essentially something like: if there were a game element that forced you to instantly die with a 1 in 3 chance each time you took a step, that because there are unavoidable deaths in crawl, there should be no reason to remove or modify that mechanic.

He's not essentially saying that. What a silly exaggeration.

There's literally no way to prevent all unavoidable deaths without diluting a number of systems which keep DCSS fresh through multiple playthroughs.

While I can see the point she made after the fact with explanation more clearly, I don't think the original statement was all too clear for two reasons:

1. I did not see anyone saying that a game mechanic should be changed solely on the basis of if it ever created an unavoidable death, so why did it need to be said at all? It did not make a lot of sense to me in the context of the conversation which was about difficulty, not absolutes (which I have already clarified myself). I have consistently stated that I accept there should be some unavoidable deaths in a dungeon crawler. You could have skipped educating me that there is literally (practically?) no way to prevent all unavoidable deaths without messing up the game had you read everything first.

2. Saying that a game mechanic shouldn't be changed or removed because it "actually" caused an unavoidable death isn't accurate if the frequency with which it "actually" causes unavoidable deaths is too high and influences the difficulty level to an undesirable level. The original statement made no explicit mention of frequency, therefore one could read it as saying that unavoidable deaths (of no specified frequency) was not a valid reason to make any changes or remove the mechanic for even (subjectively) unacceptable levels of difficulty. Yes my example was extreme, but only to illustrate the ambiguity.

Getting back to actual constructive conversation, and upon further consideration, I think that my earlier suggestion to prevent multiple shafts from happening within a certain time frame and leaving the rest alone would go the furthest towards limiting the difficulty level to a more consistent and balanced degree.

[sardonica]

Getting back to actual constructive conversation, and upon further consideration, I think that my earlier suggestion to prevent multiple shafts from happening within a certain time frame and leaving the rest alone would go the furthest towards limiting the difficulty level to a more consistent and balanced degree.

Once I know that shafts are limited, I'm incentivized to run around each level until I find a shaft, trying to hit it at full health.

Then, once my shaft is triggered (ugh), I have to keep track of its cooldown. How long until my next shaft chance?

Seems tedious to me.

[duvessa]

also not seeing what problem this even solves? what's wrong with getting shafted multiple times in succession?

[petercordia]

I think people don't like it when they're shafted 3 floors, and then shafted another 3 floors before they've even taken an upstair, because they'll be 6 floors out of depth. For some characters this is really hard to survive.

To avoid the tedium associated with cool-downs pointed out by sardonica, you could instead implement that you can't be shafted again whilst you are above a floor you which shafted to. So if you're shafted D:10->D:13, you could explore D:10,11,12 and 13 in peace, but on D:14 you'd risk shafts again. I think this would offer no perverse incentives.

[tealizard]

I'm not sure I can speak for the devteam on shafts, but the way I read the position (and I think this is correct) is: shafts are good, getting shafted twice is good, getting shafted three times is good. If you got shafted from d:1 to d:3 then to d:5 and so forth all the way down to d:15, that would be good. Maybe you would lose because of it, but it would still be good.

I mean, my experience with crawl convinces me that being able to go upstairs at all is bad. Being effectively shafted on every single descent from one floor to the next is good for the game, creates more interesting and balanced play. So all the whining about how hard shafts are is being directed at a very moderate, milquetoast perspective.

[sanka]

People do not like when an orc priest smites four times in a row or when a lich decides to cast crystal spear three times in a row. There have been proposals to remove "unfair" series in other places of the game. I think simple independent random events are better. I have played many games and double shafting early is not so common as you make it seem, and rarely it is absolutely ok. You do not know how to prepare and survive such situations - not a problem, there need to be some cases which distinguish the best players from average ones. Consider it is good for you: there is still possibilities inside your favorite game to improve! There is a rare case where even elliptic/mikee etc. would die: still not a problem, the design philosophy does not aim to completely remove these cases.

However, I agree that shafts are very bad flavour currently, and it would be better to have some more flavorful mechanic, and removing upstairs everywhere would be even better.

[VeryAngryFelid]

People do not like when an orc priest smites four times in a row or when a lich decides to cast crystal spear three times in a row.

Indeed. Consider you are playing chess vs AI and AI can checkmate you in 1-2 moves. Yet instead of winning AI randomly decides to do a stupid move. Why? Just because. How would you feel about it? In crawl I am frustrated every time it happens because I know I should have died and there is no my merit in survival. Even worse, I suspect I don't recognize all situations when I could have died and thus I don't deserve my wins even when I have never been in danger, I won due to luck with random events, not due to my skill.
My point is that good games use a different model: AI always tries to do its best and what it can do is balanced around that. When AI can kill a unit in age of wonders, warbands, eador etc. it goes and kills it, often without even taking its own losses into account. And it is great, because it was you who prevented the loss and not the stupid AI which took pity and allowed your unit to live.

[tealizard]

Chess is a silly comparison. It's a deterministic game. Crawl is a dice game, more like backgammon. In backgammon there is a clear difference between better and worse players that shows up in statistics across numerous games. The confounding factor is that crawl is absurdly long, so the importance of looking at a long series of games is obscured.

I'm no defender of the dcss design philosophy, but it is absolutely correct to reject the "omg such and such can theoretically happen five times in a row it's not fair" line of criticism.

[VeryAngryFelid]

Chess is a silly comparison. It's a deterministic game. Crawl is a dice game, more like backgammon. In backgammon there is a clear difference between better and worse players that shows up in statistics across numerous games. The confounding factor is that crawl is absurdly long, so the importance of looking at a long series of games is obscured.

I'm no defender of the dcss design philosophy, but it is absolutely correct to reject the "omg such and such can theoretically happen five times in a row it's not fair" line of criticism.

There can be different degrees of determinism. I know crawl is not chess and please note that I don't complain about rolls, if my AC/EV/SH allowed to escape death, I would feel ok because I did something to increase them along with resistances. Monster tried its best, I tried my best, I won the fight and continue to live. But I can't feel ok when something is completely out of my control i.e. orc priest can kill me with 2-3 smites but it does not even try, it decides to cast Cantrip instead. Even converting those cantrips into smites with minimal damage would be better than current situation.

Edit. Actually I am pointing out that 10 smites in a row is a better model than smite-cantrip-smite-smite-smite-cantrip-cantrip-cantrip-cantrip-smite

[sanka]

VeryAngryFelid: you could be right - I like to play chess and go sometimes (I am not very good in either though). But crawl is a very different game, where randomness does play a large role, and part of this role is that even weaker players with bad play can win sometimes. For this, we pay the price that even strong players with good play can lose sometimes (is this even true? I am not even sure. There are still very long streaks!).

I personally think that crawl tries to give a different kind of challenge. Because of the randomness there are no 100% sure strategies - you merely try to get an intuition how to be prepared and how to react to bad streaks of rolls.

In a game like this I really prefer some intuitive simple random model - series of mostly independent random events. So yes, I prefer an orc priest to have some simple behaviour, like choosing an action randomly, independently every time from a pool of actions, instead of having some complex AI with an internal state. Because in the latter case complicated reasoning about the internal state of enemies really blurs the area where the game could shine.

A game with "smart" AI would need to be really different from crawl, so it would be much better to design one from scratch than trying to fit a very different game model into an old game. And my feeling is that it would be really hard to design. If battles are non-random, then in most games you quickly reach a stage where you could predict the outcome with 100% certainty. It is really not easy to come up with battles of the simplicity and deep strategy of chess or go. If the resulting game is not deep enough, it would be much more boring than a random battle after you found the strategy to beat it. If the resulting game is not simple enough, then the battles would not be enjoyable, you will just struggle to keep track of a complexity that feels mechanic and arbitrary.

[VeryAngryFelid]

sanka,

Ok, we are different then, I just shared my feelings. I have about 10 characters in Depths+ suspended forever because I don't enjoy them, every fight feels like a broken RNG machine: if all those monsters decided to cast their best spell with at least average damage, I would die instantly from full HP.

[sanka]

Edit. Actually I am pointing out that 10 smites in a row is a better model than smite-cantrip-smite-smite-smite-cantrip-cantrip-cantrip-cantrip-smite

I am not quite sure. In the first case you can predict a battle outcome with really great certainty from the beginning, so it is boring to play it. In the latter, after every action you should judge the situation again, therefore there is tension. You can try to estimate whether it is better to run from the orc priest because it could smite three times, or try to beat it and use your heal wounds if things go wrong. Since the outcome is probabilistic, it is not easy to judge which is correct.

[VeryAngryFelid]

Probably I didn't explain well enough what I mean. I am not suggesting to use a "some complex AI with an internal state", I am suggesting to use "AI chooses the best action" model. For instance, if monster has fireball and lightning bolt, it checks PC resistances/AC and repeatedly casts spell which is expected to do the highest damage. But yes, I agree, it is probably easier to implement it as a new game. Sorry about off topic.

[VeryAngryFelid]

I am not quite sure. In the first case you can predict a battle outcome with really great certainty from the beginning, so it is boring to play it. In the latter, after every action you should judge the situation again, therefore there is tension. You can try to estimate whether it is better to run from the orc priest because it could smite three times, or try to beat it and use your heal wounds if things go wrong. Since the outcome is probabilistic, it is not easy to judge which is correct.

Orc priest can kill almost any character with 2-3 smites and there is no guarantee that you will kill the orc priest win 2-3 turns even if you lure it behind a corner. I think it is optimal to avoid it until you get more HP from XL. But it takes so much time/key presses to retreat to stairs and re-enter the level via different stairs that many people just gamble (and usually win) especially considering that many other early game monsters like orc wizard or gnoll can do as nasty things.

[sanka]

I have about 10 characters in Depths+ suspended forever because I don't enjoy them, every fight feels like a broken RNG machine: if all those monsters decided to cast their best spell with at least average damage, I would die instantly from full HP.

Well, I very often commit suicide around Depths. We just perceive the reason differently. My problem is not that the monsters do not use their best action. My problem is that they are too week compared to you! (And there are too many of them: popcorns, popcorns, popcorns.) Yes, one way to buff them could be to let them always cast crystal spear, etc. I fear that this kind of buff would not result in an interesting game for the aforementioned problems.

I think that in crawl the best situations when there is genuine danger in the sense of having a possibility for a bad series of bad rolls and dying. For example, you open a door, and three orc priests wait on the other side. It is interesting, exactly because they may kill you. However, the chance that they may kill you depends on your actions! What do you do now? Crawl allows a really bad series (all of them smites you immediately and you die without an action, but it is rare to happen), as a price for allowing a more common scenario: they kill you when you failed to reduce the possibility of the bad rolls enough.

Depths is very boring because there is not much chance to actually die if you play well. Yes, you may perceive that the chance is low because the monsters do not use their best action (stupid). But I feel that smarter monsters would not be a good solution: again there won't be very different chances!

[sanka]

Orc priest can kill almost any character with 2-3 smites and there is no guarantee that you will kill the orc priest win 2-3 turns even if you lure it behind a corner. I think it is optimal to avoid it until you get more HP from XL.

If you use melee there is always a chance that you miss it many times in a row and it decides to smite. (If you cast spells you may miscast, etc.) Of course the probability of this happening goes down exponentially as you get stronger, but exactly that is the aim of the game: get a feeling when it is "safe enough" to fight it, and get a plan what to do when you still get a bad streak of rolls.

It is "safe enough" to fight it when dying due to a bad streak of rolls inside the fight is lower than dying due to too many monsters running around the level that you do not fight. Of course the latter depends on your ability to evade them, and to make your odds better in the fight you may use some consumable.

This question is interesting in the early game. It is not interesting in Depths, because individually (almost) every monster is weak, and you can run around a lot killing them one by one, and you have a ton of options for the rare case when things go wrong.

[VeryAngryFelid]

Well, I very often commit suicide around Depths. We just perceive the reason differently. My problem is not that the monsters do not use their best action. My problem is that they are too week compared to you! (And there are too many of them: popcorns, popcorns, popcorns.) Yes, one way to buff them could be to let them always cast crystal spear, etc. I fear that this kind of buff would not result in an interesting game for the aforementioned problems.

I think your reason for abandoning Depths is very closely related to mine: the monsters feel weak because they do not use their best actions. You fight "popcorn" and then suddenly there is a spike when those popcorn monsters were lucky to roll high with their best attack. Wouldn't crawl be better if we had less popcorn fights and less "let's attack it once and see what happens" which is done dozens of times in a row vs the same monster? I guess my root issue is "Why should player check everything every turn?" so many times even vs "popcorn" monster. Monsters use weak attacks, monsters miss, monsters deal low damage, your AC rolls are high, it results in barely any damage until suddenly you lose half HP to a single attack. Maybe this is why some players like shafts so much, they get into interesting situations without losing half HP and 5 minutes first.

I think that in crawl the best situations when there is genuine danger in the sense of having a possibility for a bad series of bad rolls and dying. For example, you open a door, and three orc priests wait on the other side. It is interesting, exactly because they may kill you. However, the chance that they may kill you depends on your actions! What do you do now? Crawl allows a really bad series (all of them smites you immediately and you die without an action, but it is rare to happen), as a price for allowing a more common scenario: they kill you when you failed to reduce the possibility of the bad rolls enough.

I think it would be more interesting if those orc priests always used smite in such situation (for instance, when adjacent to an orc of any kind or when there is a door/stairs in your view) so you know you must be smart now or you will die next turn.

Depths is very boring because there is not much chance to actually die if you play well. Yes, you may perceive that the chance is low because the monsters do not use their best action (stupid). But I feel that smarter monsters would not be a good solution: again there won't be very different chances!

I think a smaller number of actually dangerous monsters might make depths much better than it is now

[sanka]

I think it would be more interesting if those orc priests always used smite in such situation (for instance, when adjacent to an orc of any kind or when there is a door/stairs in your view) so you know you must be smart now or you will die next turn.

I am not sure that in crawl it would lead to "smart" play. The decision between fight or flee would be much more trivial. (Of course you flee from three priests if three smites kill you, so this situation would not be that different if they always smite, you just cannot get lucky. But there are many other situations where it is not so clear!) If they always smite, then you do not need to develop an intuition about the game since the outcome of the battle would be much more clear, therefore there is no possibility to compare the probability of winning a fight to other probabilities of dying.

I think a smaller number of actually dangerous monsters might make depths much better than it is now

I absolutely agree with this. I may have a different view on how to buff the monsters. Although a few monsters which does simply smite you almost every time (like Saint Roka) would be fine.

[VeryAngryFelid]

If they always smite, then you do not need to develop an intuition about the game since the outcome of the battle would be much more clear, therefore there is no possibility to compare the probability of winning a fight to other probabilities of dying.

Is it bad? I recently started playing a new game, it has all relevant info to make decisions and it feels great. Why can't crawl be consistently won by smart casual players without need to play dozens of hours first to get the intuition?

[sanka]

Well, I feel to achieve that goal would go against the other goal: replayability. I feel that at best it would be interesting the first few times. It would be a puzzle: interesting until you crack it. (I like puzzles, but crawl does not exactly aim to be one.)

Also, I fully agree that crawl should give you all relevant info to make decisions, like the distribution of damage, actions, etc. But I think these distributions should be simple and random.

[VeryAngryFelid]

I think replayability will not suffer, there are dozens of species/backgrounds/gods/spells, every game has different loot, some of which can be game-changing. I don't like replayability when it means I have to fight the same monster 10 times and each time it feels like fighting a different monster despite the monster wields the same weapon, also it contradicts to getting the intuition which you mentioned earlier.

[TheMeInTeam]

When AI can kill a unit in age of wonders, warbands, eador etc. it goes and kills it, often without even taking its own losses into account.

You do have to be careful about this. Such AI can often be baited into awful trades. A lich that only crystal spears in DCSS won't summon nasty stuff instead, and if a player has high EV and dmsl maybe it really shouldn't be only firing crystal spears. It's non trivial to make the AI play perfectly...same lich shouldn't be trying to close distance much at all if it "knows" player has silence for example since that makes them a joke.

Also cantrips were basically introduced for the express purpose of having less average damage output from priest smiting for example. This could be rebalanced, where they do less smiting damage but smite more consistently. Or you could make the game super punishing/more RNG (round corner and priest wakes up while close? Game over.) The problem with making things too predictable/consistent is that those more rare RNG damage spikes do encourage consumable use/more careful play. If the priest always smites you player can basically say "okay, I have this many turns until I die".

VeryAngryFelid: you could be right - I like to play chess and go sometimes (I am not very good in either though). But crawl is a very different game, where randomness does play a large role, and part of this role is that even weaker players with bad play can win sometimes. For this, we pay the price that even strong players with good play can lose sometimes

One of these does not imply the other. You can have a situation where RNG is only positive or negative, though DCSS doesn't do either of those things. True RNG deaths happen but they're rare (you pick a caster build, get no pillar, and jackals aggro and roll well on you). You're substantially more likely to get "carrying" gear, but bad play will lose wearing basically anything in DCSS. Luck can carry modest skill, but only so far before the odds start looking like lotto tickets.

Depths is very boring because there is not much chance to actually die if you play well.

Playing well SHOULD prevent dying. In fact this is the best, most objective feedback for whether you've played well or poorly. In the vast majority of cases, dying means the player misplayed. That's a good thing. Substantially increasing the number of deaths under ideal play would be an overtly poor design choice. DCSS accepts that perfect play can die as a necessary aspect of having so many interactions between mechanics, not as something that is legitimately desirable for the game in its own right.

Shafts are one such mechanic. They basically introduce the same thing as a low-odds streak introduces, though they tend to check against different player resources.

[VeryAngryFelid]

You do have to be careful about this. Such AI can often be baited into awful trades. A lich that only crystal spears in DCSS won't summon nasty stuff instead, and if a player has high EV and dmsl maybe it really shouldn't be only firing crystal spears. It's non trivial to make the AI play perfectly...same lich shouldn't be trying to close distance much at all if it "knows" player has silence for example since that makes them a joke.

That's easy to fix. For instance, first action of lich is to always summon a demon and after that it only casts its best spell (probably as long as the demon is not dead). Or just split Ancient Lich into several kinds of several liches, as a bonus you won't need to xv it to learn which spells this particular lich has, looking at its tile will be enough.

Also cantrips were basically introduced for the express purpose of having less average damage output from priest smiting for example. This could be rebalanced, where they do less smiting damage but smite more consistently. Or you could make the game super punishing/more RNG (round corner and priest wakes up while close? Game over.) The problem with making things too predictable/consistent is that those more rare RNG damage spikes do encourage consumable use/more careful play. If the priest always smites you player can basically say "okay, I have this many turns until I die".

Cantrips just don't work, they allow to survive with bad play and die with optimal play, just because RNG decided so. If you can die to 3 smites, you should assume the priest will smite you 3 times. "I have this many turns" does not apply because smite still has random damage roll.

[TheMeInTeam]

That's easy to fix. For instance, first action of lich is to always summon a demon and after that it only casts its best spell (probably as long as the demon is not dead). Or just split Ancient Lich into several kinds of several liches, as a bonus you won't need to xv it to learn which spells this particular lich has, looking at its tile will be enough.

The code will get complex for each monster, fast. For example such a lich could conceivably be looped into never attacking by casting abjuration, making it less threatening than liches are right now with a simpler system. Even saying "cast best spell" ignores the programming effort of identifying what is, in fact, the "best spell" and why...and also a decision on how much you want to cheat by letting the AI know in advance what the player can do. It's a lot of work for questionable benefit.

Cantrips just don't work, they allow to survive with bad play and die with optimal play, just because RNG decided so. If you can die to 3 smites, you should assume the priest will smite you 3 times. "I have this many turns" does not apply because smite still has random damage roll.

I'm not seeing any reasoning here that supports the notion that cantrips allow players to die with optimal play. They make monsters with them do less expected damage *on average* with a possibility for streaky damage. Yes, players *should* account for the possibility of that streak, same for melee hits. In fact there is very little practical difference between cantrips and a missed attack.

[tealizard]

VAF uses this weird reasoning a lot where he'll say there was an infinitesimal but nonzero chance that the player could have died and therefore he went on to win the game only by luck. This is like saying it's bad luck you didn't get rich by finding a winning powerball ticket in a parking lot. There's no winning game of crawl in which the player was not lucky in this sense. There are always wild possibilities that could kill the player. That the scenarios we actually get to hear about are things like three smites in a row shows a serious lack of imagination.

I think any fair-minded, competent player can look at how crawl plays and agree that whatever flaws it may have, it's reasonably fair. In my opinion, complaints about these unlikely certain death scenarios show a lack of appreciation of the actual probabilities involved.

[sanka]

Playing well SHOULD prevent dying.

I think I did not explain myself clearly enough (or maybe I am wrong). I try again.

I know that deaths with very good play are really, really rare in crawl. I will argue that they are too rare.

I will call a strategy which leads to literally 100% win rate optimal, since I only care about win rate now. (Of course maybe no such strategy exist: then the one with the highest achievable winrate is optimal.) I do not know if anybody knows what an optimal strategy is in crawl - I have absolutely no idea. I merely know that some moves are better than others.

If in a game somebody knows an optimal strategy I think it will not give great replayability for her. There must be some tension not only in the form of a one-armed robber game, where simply rolls determine the outcome (the game will degrade to this if you follow the optimal strategy) but also from not knowing which moves are the best.

Chess and go are good games because nobody really knows what the optimal strategy is. (Yes, we made AI that beats every human - but most likely they still do not play optimally.) And more importantly: I do not know what the best strategy in go is, therefore I keep wondering in every game what to do.

Now how can we get whether your requirement ("Playing well SHOULD prevent dying") is true for a given game? I think the only answer is to make the game shallow, that is, that we do know what an optimal strategy (or at least a strategy we can show is very close to optimal) is, and show that it always (or not always) wins. So we know your condition is true/false for a given game because we know too much about it: it is not as interesting to play anymore!

A good game for me is one where there is a place to improve, so when we are not really sure about the answer for your question.

And I think all of this is still true if we replace "literally 100% winrate" with a winrate which is very, very close to it, since as humans we do not play that many games to really make a difference.

Of course, this goal is not easy to achieve. It is easier to create a game which is more shallow, so there clearly exists an optimal strategy, but make it hard in different ways:
1. More opaque: the game does not gives enough info about the task to the player (spoilers needed)
2. More obscure: there is arbitrary complexity like 500 different monsters with 500 different effects which mostly differ by some numbers, so you need to calculate a lot if you want to know the correct action.

These are not good. A good goal for a game I think:
1. Transparent: gives you info about the game itself, no spoilers needed
2. Simple: the required info fits into a human's head, so no lexicon and constant rule checking (xv) is needed
3. Hard: nobody knows what the optimal move is, so nobody can achieve true 100% winrate

I think that this hard goal is easiest to achieve (or get close) with a random game that does not have a known strategy that gives you 100% winrate. In other words, we do allow deaths with "good" play exactly to create a game where we do not know what the best strategy is: maybe these deaths could be reduced even more.

Now lets look at crawl for some dangerous situations: centaur/killer bee etc. on D2, or Grinder on D3, etc. I do think that there is a non-zero probability that a very good player dies in these situations. However, by allowing it we also created a race: which player can minimalize this probability? Can we find a strategy that gives even better results (higher winrates)? If we would know the answer for your question, that would mean that we know a strategy that always survives these situations, therefore the game is not very interesting anymore.