Increase scoring formula

[Berder]

To compensate for the cTele removal for speedrunners, I suggest scaling the scoring formula upwards so that the new high scores would be similar to pre-cTele removal high scores. This will allow speedruns to continue in trunk and compete with old speedruns.

To find the correct scaling amount it will be necessary to estimate the number of turns saved by cTele in a 15-rune, X. Expert speedrunners may have some idea of what X should be, or this could be done by observation of how current speedruns compare to old ones. Then use turncount - X * runes / 15 in the scoring formula, instead of turncount.

[tabstorm]

I estimate the turns saved around 2k-4k (depending on when ctele is found) in a 15 runer, fwiw. But, I do not think this is a needed change, really. Previous versions have had some broken features and are still on the all-time scoreboard. See: 0.10 summoning. The most fair thing would probably be scoring by version, I guess.

[Kate]

Crawl changes a lot all the time and games from different versions are never particularly comparable if you care about it that much, this isn't going to happen.

[Berder]

I saw in the other thread:

It's not even about speedruns (which are dead because no cTele)

and it sounded like a problem.

[tabstorm]

It's not really a problem, you can still speedrun, it's just that you won't be getting turncounts of under 25k turns on a 15 runer anymore. The game changes between versions.

[archaeo]

I mean, if people really want to challenge the top scores, all the old versions are still up online, after all. It's pretty common in the speedrunning community for people to continue using previous versions of games in order to continue taking advantage of glitches patched out in later versions; SGDQ just showed off several games where players purposefully played old versions, for example.

edited to add: of course, one of the issues with speedrunning Crawl is how unbearable it is; getting a sub-25k turncount requires things like "memorize/look up every pan rune vault to ctele into unexplored tiles" or "breadswing to save a few hundred turns," and that's assuming the RNG even lets you speedrun that game, or if you need to play another hundred games to get the same required items.

[bananaken]

I think the best way to accommodate competitive Crawl players would be to differentiate any top scores and speedruns according to major revisions (i.e. 0.15, 0.16, etc.). Trunk could also have its own leaderboard, but since this is constantly changing it's inherently not fair and probably less important to differentiate.

I watch speedrunners all the time, and even hardcore speedrunners tend to take a more pragmatic approach to running their games, so optimizing for turncount in a version where breadswinging constantly is optimal is probably going to be phased out in favor of versions that have more interesting methods. Differentiating leaderboards by version just makes this transition easier, IMO.

Edit: Looking at the leaderboards on Akrasiac, versions are already tracked, so it's just a matter of re-organizing the existing data, so this probably doesn't require touching Crawl's code!

[asdu]

optimizing for turncount in a version where breadswinging constantly is optimal is probably going to be phased out in favor of versions that have more interesting methods.

Isn't breadswinging optimal in every existing version of crawl? Or do you mean breadswinging is going to be "fixed" in a future version (yes please)?

[bananaken]

optimizing for turncount in a version where breadswinging constantly is optimal is probably going to be phased out in favor of versions that have more interesting methods.

Isn't breadswinging optimal in every existing version of crawl? Or do you mean breadswinging is going to be "fixed" in a future version (yes please)?

Yeah, I'm assuming it's just a matter of time for it to become obsolete somehow, sorry I wasn't clear.

[archaeo]

The problem with "just a matter of time" is that the mechanics involved in breadswinging are pretty complicated, from what I understand; the only serious reform I've heard about would require removing "turns" from the game and moving entirely to aut scoring.

[Berder]

To eliminate breadswinging there is a simple solution: resting with 5 occurs in turns that last 15 auts instead of 10.

The simple solution has the problem that resting becomes slightly more dangerous in normal games. I don't think this problem is serious. This problem could be solved (if you think it's serious, which I don't) by letting every "rest" turn be composed of two smaller turns, and your total turn count only increments by 1 instead of 2 at the end of it.

[archaeo]

That seems pretty unintuitive, Berder; even though it's not like, hard to grasp or anything, it's still a weird oddity you have to explain to players. I think you might also be underselling how weird it would be to give the enemies 5 auts whenever they interrupt your rest, and the proposed correction to that means some turns don't count as turns, etc., which only just makes it more complicated.

It also doesn't really fix "breadswinging," since speedrunners will just switch to using heavy untrained weapons instead of bread or other non-weapons. I assume you can change how many turns you spend resting with 5 in the .rc options, otherwise that presents another problem to your proposal.

[duvessa]

yeah let's fix speedruns by making the 5 command completely useless, that sounds good

wtf

[asdu]

One possible solution would be to limit the amount of regeneration from actions that last longer than 10 aut to that of a 10 aut action. Slow movement speed would have to be adjusted for (and so chei-/nagastepping would replace breadswinging where applicable) and maybe the slow status too. It would also slightly nerf the regeneration when swinging weapons (or bread) slower than 10 aut, which would probably have negligible consequences.
This would also result in a loss of regeneration when taking actions >10aut like switching armour, eating, going up and down stairs, learning spells, dropping/picking up (?), what else? I don't think this would have serious implications but maybe I'm overlooking something obvious.
It's not a very elegant solution I guess, but neither is breadswinging, which I hate with a passion and would love to see die.

[Aule]

optimizing for turncount in a version where breadswinging constantly is optimal is probably going to be phased out in favor of versions that have more interesting methods.

Isn't breadswinging optimal in every existing version of crawl? Or do you mean breadswinging is going to be "fixed" in a future version (yes please)?

Yeah, I'm assuming it's just a matter of time for it to become obsolete somehow, sorry I wasn't clear.

If it's true that breadswinging is really a thing (apparently so), and if it's also true that devs dismiss metagame issues in development (I'm skeptical), then it should never be "fixed" because it's a non-game issue (which it really is).

[tedric]

That's rather rigid logic. Meta-game issues may be de-prioritized in favor of actual-game issues, but that's not the same as everyone pretending that those issues don't exist.

[Rast]

Just make Hell shorter to compensate.

My suggestion is to make it a single nine-level branch, with runes on levels 3,5,7,9.

[Aule]

That's rather rigid logic. Meta-game issues may be de-prioritized in favor of actual-game issues, but that's not the same as everyone pretending that those issues don't exist.

All functional if-then statements must be rigidly logical to perform correctly.

So, you say it is not, in fact, true that devs dismiss meta issues. If you are correct, then that second part of my "rigidly logical" conditional expression is evaluated false, and therefore the conclusion is not reached. The statement still stands as a reasonable expression, regardless of how its terms evaluated.

I mean, really, I wouldn't pretend to know the unified hive mind of a number of people I do not know and have never met, would you? Perhaps this might elicit an official position on the matter on which there is apparent disagreement between you and yourself, the author of the following which I paraphrased as the basis for second conditional in the expression:

... highscores, streaking -- basically any player goal except "grab runes, win game" -- are not important design concerns. They're a meta-game that the player base invented...

So you see, it's your own damn fault!

Anyway, until an official position is made public, I remain skeptical that there is any genuine demarcation between actual game and meta game issues in its development, since the "player base" could also be said to have "invented" its current incarnation and direction of development, given that the devs were themselves first players, and to my knowledge remain so.

If my skepticism is justified, then the second conditional fails, and the conclusion (breadswinging is never fixed) is not reached. I suppose an else clause should have been added: breadswinging is fixed (and scores are adjusted for fairness, and optimal play is a consideration in design issues, et al.), but then there should also be a third conditional; namely, "if the devs get around to it, or right people gripe about it."

Personally, I don't believe meta issues should have any impact on game design whatsoever, but that's an impossible ideal in practice, where there are real meat and bone humans at work.

[Piginabag]

what the fuck is breadswinging

[Aule]

what the fuck is breadswinging

Exactly!

From what I have gathered, it's a conduct of certain players who focus entirely on turn counts in their games that takes advantage of an internal mechanic to reduce the total number of turns.

[dowan]

Breadswinging, from what I understand, is a way to take advantage of the fact that there are two measures of time in the game. Auts are the one displayed in game. If you press . you wait for 1 aut, if you attack with a min delay greatsword, it takes 0.7 aut, if you attack with a non-weapon (say a loaf of bread) it takes 1.5 auts. The amount you heal goes by auts, meaning you will heal almost the same amount if you attack twice with a min delay greatsword as if you attack once with a loaf of bread.

The other measure of time is turns. Any action takes 1 turn, regardless of the amount of auts it takes. Turns are how scoring is done. Swinging a loaf of bread takes 1 turn, swinging a min-delay great sword takes 1 turn, pressing . takes 1 turn.

What this means is you can heal more per turn by doing things that take as many auts as possible in those turns. I can heal 9 auts of damage in 9 turns by pressing . 9 times, in ~13 turns by swinging a min delay great sword 13 times, or in 6 turns by swinging a loaf of bread 6 times.

I don't think you can actually equip bread anymore though, so I'm not sure if this is actually even doable anymore.

[Sprucery]

I don't think you can actually equip bread anymore though, so I'm not sure if this is actually even doable anymore.

So you just have to to use the slowest weapon you have unskilled. Of course, finding a bardiche can be hard, especially for a speedrunner.

[Zooty]

You still can wield bread, and even if you couldn't, there's plenty of glaives and battleaxes around.

[tedric]

how formal logic works

There was nothing formally wrong with your construction; the problem was one of semantics, particularly of over-interpretation. You seemed to be using the phrase "dismiss meta-game issues" to mean "refuse ever to think or do anything about meta-game issues," which is stronger than anything I (or, as far as I know, any devs) have ever said.

I think it would be more reasonable, based on past dev comments over the years I have been lurking and participating in this forum, to conclude that they (1) would rather not spend effort on changes that only affect the meta-game; (2) are unwilling to make changes that improve the meta-game at the expense of the actual-game; and (3) are quite willing to make changes that improve the actual-game at the expense of the meta-game, or that improve the meta-game alongside the actual-game.

In practice, that would mean that breadswinging might get fixed/removed incidentally at some point; but also that if there's a "slow dev day" with no pressing actual-game issues to be addressed, and someone can convince a dev to spend a few minutes on their pet project to fix/remove breadswinging, and there would be no negative effects on the actual-game, then breadswinging could get fixed/removed. In practice, of course, "slow dev days" are probably not that common, and it sounds like there would be tons of work to rewrite huge fundamental sections of the code and possibly a lot of negative follow-on effects involved in any effort to address breadswinging. So it's unlikely, but possible. QED.

Of course I'm running the risk of putting words in the devs' mouths here, and I would be happy to be corrected by any devs who want to weigh in on their individual or collective positions...but it seems like a small risk. And it's certainly not "hive-mind-reading" to observe others' statements and actions and then draw conclusions about what they believe. As long as we're quoting things I said in that other thread:

Everybody knows the devs aren't making Dungeon Crawl: Speedrun Soup (or Streak Soup, or Highscore Soup).

[Aule]

how formal logic works

Of course I'm running the risk of putting words in the devs' mouths here, and I would be happy to be corrected by any devs who want to weigh in on their individual or collective positions...but it seems like a small risk. And it's certainly not "hive-mind-reading" to observe others' statements and actions and then draw conclusions about what they believe. As long as we're quoting things I said in that other thread:

Everybody knows the devs aren't making Dungeon Crawl: Speedrun Soup (or Streak Soup, or Highscore Soup).

LOL, thanks. It's all semantics unless these things are clarified, which is a genuine motivation to elicit such clarity.

In fact, I can't understand why the mechanic of wielding allows such things in the first place. It is certainly atypical, and results in mistaken wielding of bread simply by mistaken side-stroke keypresses; i.e., bread assigned to 'e' is accidentally wielded when one mistakenly hits 'e' along with a sideswipe of 'w'. I know I've been in situations swinging away at something with a scroll or another equally useless weapon. I would be more than pleased if it were fixed not as a meta game issue but as a real game issue, by eliminating the ability to wield anything you can't do damage with. (And, if bread can actually do damage, let's just get real and get rid of that, too.)

To say it's spoilery would not be inaccurate, I don't think, so on that basis alone it could use some work.

[tasonir]

Breadswinging, from what I understand, is a way to take advantage of the fact that there are two measures of time in the game. Auts are the one displayed in game. If you press . you wait for 1 aut, if you attack with a min delay greatsword, it takes 0.7 aut, if you attack with a non-weapon (say a loaf of bread) it takes 1.5 auts. The amount you heal goes by auts, meaning you will heal almost the same amount if you attack twice with a min delay greatsword as if you attack once with a loaf of bread.

The other measure of time is turns. Any action takes 1 turn, regardless of the amount of auts it takes. Turns are how scoring is done. Swinging a loaf of bread takes 1 turn, swinging a min-delay great sword takes 1 turn, pressing . takes 1 turn.

What this means is you can heal more per turn by doing things that take as many auts as possible in those turns. I can heal 9 auts of damage in 9 turns by pressing . 9 times, in ~13 turns by swinging a min delay great sword 13 times, or in 6 turns by swinging a loaf of bread 6 times.

I don't think you can actually equip bread anymore though, so I'm not sure if this is actually even doable anymore.

So close! But no, pressing . will wait for 10 auts. Auts are the smallest unit of time (aut stands for absolute unit of time), a turn is usually considered to be 10 auts, but it isn't *always* 10 auts. Swinging bread takes 15 auts, and hence you get more regeneration from it than from standing still (10 auts). Aside from the 10:1 factor of how long an aut is, the rest is correct.

Ya'll need to chei statue form more, movement takes 30 auts.

[Siegurt]

I believe it is arbitrary unit of time.

[Sandman25]

Right, everyone knows time is relative, not absolute.

[bananaken]

So an early dark maul find is the holy grail for breadswinging?

[sanka]

I still do not really get it. Isn't breadswinging only affects your final score and nothing else inside the game? Then why are these complex and really strange proposals to change game mechanics, instead of changing the scoring formula to use auts (or something entirely different)? I think speedrunners should be asked what kind of scoring formula would they prefer, as they are the only ones who cares about score I guess.

[dowan]

Yeah sanka, outside of extremely contrived edge cases, it's generally only a concern to speedrunners and high score seekers (I assume this is one and the same, but maybe there's some separation)

The extremely contrived edge case has to do with poison:

Since poison damage only ticks at the beginning of your turn (I think) one could possibly use breadswinging to cause themselves to take an extra long turn, possibly resulting in just enough health regen to survive the final tick of what would be fatal poisoning. So I guess if you're going to die of poison anyway, maybe swinging a loaf of bread makes more sense than doing nothing.

[johlstei]

I still do not really get it. Isn't breadswinging only affects your final score and nothing else inside the game? Then why are these complex and really strange proposals to change game mechanics, instead of changing the scoring formula to use auts (or something entirely different)? I think speedrunners should be asked what kind of scoring formula would they prefer, as they are the only ones who cares about score I guess.

Wielding bread is a strange thing to have in the game at all - it sounds like something that one is much more likely to do accidentally than do on purpose. I don't care about speedrunners at all but I think we should remove wielding bread, and anything that isn't a weapon (or needed for a spell.) This does not address the speedrun issue at all, since they could still swing a slow weapon, but it addresses the average person player crawl accidentally wielding a phial of floods.

[ajon]

How does swinging bread increase one's final score?

[sanka]

How does swinging bread increase one's final score?

Final score depends on the number of player actions (less is better). Regeneration depends on actual game time. To reduce the number of actions while waiting to regenerate hp back it is optimal to repeat the slowest action you can do, because this means that the most game time will ellapse peraction. So you wield a bread and attack an empty square. I think.

[ajon]

Ok makes sense. Why does swinging bread count as less actions than just hitting "5" which is one action?

[Sandman25]

Ok makes sense. Why does swinging bread count as less actions than just hitting "5" which is one action?

That would solve breadswinging problem indeed. If player was able to get into a safe place for resting, he deserves a bonus to score.

[sanka]

Ok makes sense. Why does swinging bread count as less actions than just hitting "5" which is one action?

One swing is one action, but more time (so more regeneration) then one rest action. Pressin 5 is simply repeats the rest action in a loop, it is not one action, similary as autoexplore is not one action, but as many as steps it actualy takes.

[tasonir]

Ok makes sense. Why does swinging bread count as less actions than just hitting "5" which is one action?

5 is a shortcut that says "rest here for many turns". If you have the default limit of 100 turns of resting, that counts as 100 turns for scoring, it's a huge amount of time to wait in one place on a speedrun. You could swing bread 66 times and heal just as much (ignoring rounding). I believe that some other multi-turn actions count all of their turns, like putting on plate armor counts as 15 turns, eating a ration counts as 3 turns, etc, but I haven't confirmed that. In any case, all melee attacks/movement count as one action.

The reason we don't just switch to aut scoring is that would completely destroy many types of speed running characters from ever being competitive again. Casters as a whole would be nerfed (10 aut casting speed vs 7 aut 2h melee weapons), no one could speed run with Chei anymore (Chei is the #1 god in terms of species high scores, a good enough proxy for "best speedrun god"). It would be extremely problematic and still vulnerable to being gamed in other ways - characters who find that early wand of hasting and then learn the haste spell are suddenly far superior to anyone who didn't...

[Sandman25]

It would be extremely problematic and still vulnerable to being gamed in other ways - characters who find that early wand of hasting and then learn the haste spell are suddenly far superior to anyone who didn't...

Similar to those who find vampiric lajatang in current speedruns?
I don't think it is possible to "break" speedrunning. Some characters are extremely good in speedrunning (VSBe?), some are extremely bad (SpSu?). There is nothing wrong if those characters would change in next version.

[Kate]

The reason we don't just switch to aut scoring is that would completely destroy many types of speed running characters from ever being competitive again. Casters as a whole would be nerfed (10 aut casting speed vs 7 aut 2h melee weapons), no one could speed run with Chei anymore (Chei is the #1 god in terms of species high scores, a good enough proxy for "best speedrun god"). It would be extremely problematic and still vulnerable to being gamed in other ways - characters who find that early wand of hasting and then learn the haste spell are suddenly far superior to anyone who didn't...

Haste spell is the only thing there that's anything close to being an actual problem for aut scoring. Aut scoring has been discussed plenty in the past (to resolve the problem of there being two different types of turncount instead of one), as far as I'm aware it could quite feasibly become a thing especially if haste spell should ever have some kind of accident.

[sanka]

tasonir: So you are saying that if the scoring formula would use auts that would make characters that are actually strong better at the scoring instead of other characters that happen to be adjusted to the scoring formula instead of actual game power? You really must like Chei to see this as a bad thing....

[sanka]

MarvinPA: I do not understand. What is the problem with haste and aut scoring? It is good to be hasted, scoring or not, so it seems logical to get more score...

[Siegurt]

MarvinPA: I do not understand. What is the problem with haste and aut scoring? It is good to be hasted, scoring or not, so it seems logical to get more score...

It would make it optimal to haste out of combat for score, while resting and traveling and what not, creating tedium for no purpose other than scoring (or rather replacing one sort of tedium for a different one, for a non-trivial cost in development time)

[Sandman25]

It would make it optimal to haste out of combat for score, while resting and traveling and what not, creating tedium for no purpose other than scoring (or rather replacing one sort of tedium for a different one, for a non-trivial cost in development time)

It is not a problem IMHO. Even undead races cannot be hasted indefinitely. Getting a couple of bad mutations after you haste for actual fighting or escape is a good price for this score increase.

[Siegurt]

It would make it optimal to haste out of combat for score, while resting and traveling and what not, creating tedium for no purpose other than scoring (or rather replacing one sort of tedium for a different one, for a non-trivial cost in development time)

It is not a problem IMHO. Even undead races cannot be hasted indefinitely. Getting a couple of bad mutations after you haste for actual fighting or escape is a good price for this score increase.

Well, but you could simply haste every time you rested off the contamination with no penalties, that's what I had in mind for 'tedious but optimal' particularly while traveling.

[Sandman25]

Well, but you could simply haste every time you rested off the contamination with no penalties, that's what I had in mind for 'tedious but optimal' particularly while traveling.

And I meant that you can meet a dangerous monster around next corner right after haste has expired and you have full contamination from previous cast.

[Siegurt]

Well, but you could simply haste every time you rested off the contamination with no penalties, that's what I had in mind for 'tedious but optimal' particularly while traveling.

And I meant that you can meet a dangerous monster around next corner right after haste has expired and you have full contamination from previous cast.

Well, that's not very likely if you're using haste to travel back to a shop, or a stash or to an earlier branch or something, or if you're just using it to rest in a safe area. I don't think it'd be optimal to haste while exploring for exactly that reason.

[bananaken]

I still think speedrunning with haste instead of breadswinging would improve the situation due to:

• Both haste and "breadswinging" rely on RNG to varying extents. One requires you to find haste spell (or absurd amounts of haste charges/potions), the other requires you to find a slow weapon. The difference is that in a "godly" run you wouldn't train M&F and find a dark maul, while in a "haste speedrun" you just find haste, which depending on how you approach it actually has more possibilities (Sif Muna, Book Acquirement, etc.)
• Haste makes more logical sense IMO, you go faster to speedrun, instead of the whole "spend less player turns but burn more auts to regen faster" thing.

[duvessa]

you guys do realize breadswinging isn't just about score right? it also reduces piety decay

[Sandman25]

you guys do realize breadswinging isn't just about score right? it also reduces piety decay

Really? So it is optimal for everyone Haste problem with aut scoring is less annoying then IMHO