Balanced Berserker

[Pereza0]

GUYS

BERSERKERS

[triorph]

Yes, I do know how Elo works. Do you? It's used to calculate ratings in player versus player games, which you may know crawl is not. While one person attempted to shoehorn it into crawl as a way to measure player skill, doing so yielded wildly different results based on several arbitrarily-chosen variables. The suggestion that Elo should be considered a "source of knowledge" that one has to do better than is baffling. You might as well ask me to come up with a better source of knowledge than astrology or numerology.

As the person who tried to shoe-horn it into crawl, I would like to echo Lasty's sentiments that its a mostly useless indicator of anything. I made several arbitrary decisions, such as player skill to affect results (5 wins), the K-factor of the rating system, whether or not Char or Race is the better indicator, or maybe some combination of both. It was merely a fun exercise with some mild applicability.

Its also worth noting that the only thing it measures is your ability to win, and is completely thrown off by games where that's not the intention, such as trying to speed-run, trying to learn extended for the first time, trying to mega-zig. It also makes no distinction between dying on D:1 and dying in Zot, or winning with 3 runes or winning with 15.

Its fun to play with, but no serious conclusions should be drawn from it.

[Berder]

I think you misunderstood what I was saying. The object of inquiry that and into described isn't "being good at crawl", so demonstrations of crawl skill are entirely beside the point. The object of inquiry in this case is "the relative value of different gods", or something thereabouts

I've won using most of the gods, so?

What source of knowledge do you have that is better than the Elo ratings, that proves GrFi is not very easy and OpTm is not very hard? Do you know how Elo works? I think it proves that your conceptions about what is very easy and what is very hard require some adjustment. As a guess, since you didn't explain yourself, maybe you think OpTm is easy. Transmuter is the single lowest winrate background among tenpercenters, and octopode is the single lowest winrate race. Combine the two and it's not surprising OpTm has a high Elo.

Yes, I do know how Elo works. Do you? It's used to calculate ratings in player versus player games, which you may know crawl is not. While one person attempted to shoehorn it into crawl as a way to measure player skill, doing so yielded wildly different results based on several arbitrarily-chosen variables. The suggestion that Elo should be considered a "source of knowledge" that one has to do better than is baffling. You might as well ask me to come up with a better source of knowledge than astrology or numerology.

It's perfectly valid to use Elo for player vs enemy (enemy in this case being a character). For example, chess websites use this method (or Glicko, a variation on Elo) for ranking puzzles. The basic idea is that a 400 point difference results in a 10% win chance, and the ratings self correct so that this is true.

triorph, you ought to give yourself more credit. Elo is much better than winrate queries in Sequell since it takes into account the skill of the players who try a given character, as well as how often they succeed. Yes, it has problems with speedrunners etc, but so do winrate queries. As for tuning the K-factor, we already have to do something like that when examining how many versions back to include in a query. (you know, if you treat all games in a given version as from the same "tournament," like described in http://en.wikipedia.org/wiki/Elo_rating ... nce_rating , this would reduce the influence of the K-factor on the results).

As it happens, OpTm is not a particularly hard Op or a particularly hard Tm, and while neither Op is not the easiest species and Tm is not the easiest background, neither are they the hardest. My source of knowledge is playing a lot of crawl and having a pretty thorough understanding of crawl. Or, if it appeals to your love of data more, my source of data is running each character combo through random2(1000) ten million times and then averaging the values.

Op is the lowest winrate race among tenpercenters, and Tm is the lowest winrate class among tenpercenters, so it really makes sense that OpTm would be hard. It's unclear to me how just "playing a lot of crawl" would qualify you to say that Op isn't the hardest race, Tm isn't the hardest class, and OpTm isn't hard. I've played a lot of crawl, too.

It's true that you're not making the argument that Trog is being used to bolster weak combinations. You're the one who needs to demonstrate that there's absolutely no confounding data and therefore all the data you offered before amounts to something other than a complete waste of time. You're also the one who doesn't seem to understand that good use of data, e.g. the kind of empirical study you're trying to emulate, requires you to make an effort to remove all confounding factors, show significance, and, if confounding factors remain that cannot be ruled out, admit them clearly and acknowledge that they may well be confounding your conclusions.

You are here insisting that I perform an exhaustive analysis to eliminate the possibility of weak characters taking trog (when it doesn't appear to be a problem, based on a look at the characters that take trog).

In a previous post you insisted the data (of what characters take trog) was meaningless and begged me not to analyze it.

I really think it would help you to try reading some well-performed, peer-reviewed studies to see how data is correctly used in empirical research, or ideally take a class on the subject.

Fortunately I do have a degree in the subject.

[and into]

True story: Berserker is my go-to background when I am playing Crawl drunk.

(Yes, I know this is "anecdotal.")

[nilsbloodaxe]

Fortunately I do have a degree in the subject.

Well then you should be aware of the adage, "garbage in garbage out". You can't rely on crappy data to get good inferences.

[dpeg]

I think the tone in this thread is unnecessarily hostile. Whatever, I like the !lg queries and their interpretations. I am also someone who's outspoken in favour of keeping the Berserker background, so I am interested whether Trog could see some nerfs (like slower piety gain, or higher piety costs).

[Berder]

The data is adequate to show that Trog is not far ahead of other gods among non-zealots as far as getting three runes is concerned. That's a fairly weak hypothesis and it doesn't take a lot to prove. Just the fact that Trog comes in about equal to somewhere between five and eight other gods makes it quite likely that trog is not the best of the five to eight, and less likely still that it's far ahead of them all.

The eight gods I'm referring to, with a somewhat arbitrary cutoff based on estimates of uncertainty in these figures, are: fedhas, qaz, oka, mak, ru, dith, veh, ash.

  Code:

Elyvilon 0.285714285714 ( 6 / 21 )
Sif Muna 0.31067961165 ( 32 / 103 )
Xom 0.3125 ( 5 / 16 )
Gozag 0.342105263158 ( 13 / 38 )
Cheibriados 0.385245901639 ( 47 / 122 )
Zin 0.4 ( 18 / 45 )
Yredelemnul 0.416666666667 ( 15 / 36 )
Nemele Xobeh 0.431034482759 ( 25 / 58 )
Kikubaaqudgha 0.432835820896 ( 58 / 134 )
Fedhas 0.440476190476 ( 37 / 84 )
Qazlal 0.460526315789 ( 35 / 76 )
Okawaru 0.478125 ( 153 / 320 )
Trog 0.479452054795 ( 35 / 73 )
Makhleb 0.48087431694 ( 88 / 183 )
Ru 0.5 ( 33 / 66 )
Dithmenos 0.507462686567 ( 68 / 134 )
Vehumet 0.537142857143 ( 94 / 175 )
Ashenzari 0.553333333333 ( 83 / 150 )
the Shining One 0.564102564103 ( 22 / 39 )
Beogh 0.636363636364 ( 14 / 22 )
Lugonu 0.769230769231 ( 10 / 13 )
Jiyva 2.22222222222 ( 20 / 9 )


[Siegurt]

Well, and another thing to consider, and which the data doesn't really reveal, is the ease of gaining piety with Trog vs with other gods, and the portions of the game for which he's useful...

Anecdotally, I feel like Trog's very very good early on, and his use wanes late in the game to something middling-to-average (I don't feel like he's ever *bad*)

I'm not sure how that is representable as a data set when compared to another god who will have a different power curve, TSO being the obvious choice (He's much better when you fight a significant number of undead or demons) but most gods have different "powerfulness" in different places in the game. Trying to place them all on a single-dimensional scale is bound to result in disagreements over how to treat the data.

Even racial advantages or disadvantages can play into how "powerful" a good is, merfolk and octopodes can use fedhas's rain ability in ways other races can't. Felids get very little benefit from oka gifts, Formacids have more use for lucy's random blink than other races do.

I think the overall problem is that trying to distill a god's effect on the game to a single-dimensional expression like "powerful" results in a meaningless expression of results (Regardless of how the data is collected and presented).

You could similarly try to extract values for which is the 'best' breakfast cereal and you'd get a host of answers depending on whether a given person values taste, nutrition, feeling full, gluten allergies, pretty colors, magic toy surprise inside etc. etc. as the defining characteristic.

Even if you managed to create a concrete set of critera like "has the greatest positive impact on win rates" you *still* have different results depending on the type of character played, tactics generally employed, race used, branches visited etc. If a player is good enough to consistently win every branch except the Tomb without a god, which god is the most 'powerful' for that player? Is it the same for a player who can only occasionally get to the lair?

The problem isn't with the data or how it's collected, it's that the question itself belies a premise that isn't answerable for the general case, except as general hand waving or as a series of anecdotes about specific situations, specifically because there *is no* general case.

There's not a canonical game of crawl that one can point to and say "that's the normal from which all others are variations" there's *popular types* of games of crawl, but even among the popular combos there's different ways of playing it out.

[xentronium]

All this talk makes me want to go play another berserker.

[XuaXua]

My love for you is ticking clock.

[Lyrick]

On topic, there probably is no such thing as a "balanced berserker", so long as you start with enough piety to use Berserk.

[Berder]

Well, and another thing to consider, and which the data doesn't really reveal, is the ease of gaining piety with Trog vs with other gods, and the portions of the game for which he's useful...

Anecdotally, I feel like Trog's very very good early on, and his use wanes late in the game to something middling-to-average (I don't feel like he's ever *bad*)

I'm not sure how that is representable as a data set when compared to another god who will have a different power curve, TSO being the obvious choice (He's much better when you fight a significant number of undead or demons) but most gods have different "powerfulness" in different places in the game. Trying to place them all on a single-dimensional scale is bound to result in disagreements over how to treat the data.

Even racial advantages or disadvantages can play into how "powerful" a good is, merfolk and octopodes can use fedhas's rain ability in ways other races can't. Felids get very little benefit from oka gifts, Formacids have more use for lucy's random blink than other races do.

I think the overall problem is that trying to distill a god's effect on the game to a single-dimensional expression like "powerful" results in a meaningless expression of results (Regardless of how the data is collected and presented).

You could similarly try to extract values for which is the 'best' breakfast cereal and you'd get a host of answers depending on whether a given person values taste, nutrition, feeling full, gluten allergies, pretty colors, magic toy surprise inside etc. etc. as the defining characteristic.

Even if you managed to create a concrete set of critera like "has the greatest positive impact on win rates" you *still* have different results depending on the type of character played, tactics generally employed, race used, branches visited etc. If a player is good enough to consistently win every branch except the Tomb without a god, which god is the most 'powerful' for that player? Is it the same for a player who can only occasionally get to the lair?

The problem isn't with the data or how it's collected, it's that the question itself belies a premise that isn't answerable for the general case, except as general hand waving or as a series of anecdotes about specific situations, specifically because there *is no* general case.

There's not a canonical game of crawl that one can point to and say "that's the normal from which all others are variations" there's *popular types* of games of crawl, but even among the popular combos there's different ways of playing it out.

All that is true, and yet there is still value in taking an average. On any given day, temperature varies between outdoors and indoors, and between shade and sunlight, and some areas are windier than others so have a different windchill. And yet, you still want to know what the temperature is like on average for your area.

[Siegurt]

All that is true, and yet there is still value in taking an average. On any given day, temperature varies between outdoors and indoors, and between shade and sunlight, and some areas are windier than others so have a different windchill. And yet, you still want to know what the temperature is like on average for your area.

The point I'm trying to make here is that in your analogy, we don't have access to the temerature in my area, or even in a contiguous region, we have access to randomly sampled temperatures at non-regular intervals in various places across the globe in such a way that we can neither verify that the average of any given set of data represent an average temperature for a given area, nor can we verify that the set as a whole represents an average for the whole planet. Unless one of those points is actually where I'm standing, averaging data across several points which may or may not have anything to do with me.

Additionally, temperature is a relatively bad representation as an analogy for crawl games, two people in relatively close geographic proximity to each other will have some similarity of temperature, because temperature has an effect on adjacent locales. Two players playing a similar type of build, will have substantially different experiences of the game, depending on their play style, skill level, what the RNG generates for them, what choices they make for skills etc. The closest thing to geographic proximity that crawl has is "player consistency" a given *player* will have similar experiences across crawl games, but it's not really rational or useful to try to extract "how powerful a given god is for a specific player" from the data (1. because there's not enough of it, 2. because it doesn't transfer well or in a consistent way to other players, it'd be a bit like trying to predict the weather on a given day in Amsterdam from weather data collected in Auckland)

[PleasingFungus]

I think the tone in this thread is unnecessarily hostile. Whatever, I like the !lg queries and their interpretations. I am also someone who's outspoken in favour of keeping the Berserker background, so I am interested whether Trog could see some nerfs (like slower piety gain, or higher piety costs).

I don't really think Trog needs any changes at this time, except maybe tweaking the corpse sacrifice UI to work like Fedhas's.

[Berder]

All that is true, and yet there is still value in taking an average. On any given day, temperature varies between outdoors and indoors, and between shade and sunlight, and some areas are windier than others so have a different windchill. And yet, you still want to know what the temperature is like on average for your area.

The point I'm trying to make here is that in your analogy, we don't have access to the temerature in my area, or even in a contiguous region, we have access to randomly sampled temperatures at non-regular intervals in various places across the globe in such a way that we can neither verify that the average of any given set of data represent an average temperature for a given area, nor can we verify that the set as a whole represents an average for the whole planet. Unless one of those points is actually where I'm standing, averaging data across several points which may or may not have anything to do with me.

Randomly sampled temperatures at non-regular intervals across the globe - like what's used to determine global temperature trends. Weather stations aren't distributed regularly.

How you use the data is up to you. What it says is that Trog does not have a higher rate of getting the third rune among tenpercenters than other gods do.

The original subject of this thread was whether berserkers were "balanced" and what, if anything, should be done to make them balanced. If you take your view, what criteria could you possibly use to judge whether Trog is balanced? If every god is different for every character and player and no meaningful comparison can be made, then you could never judge when any god is balanced or not.

[tedric]

Fun Fact: This thread was never about what "should" be done with Berserkers!

Curious, what does a balanced berserker start look like? If a char starts at the gate naked, with no items nor skills, but has trog, will they still be broken?

Caveat, I am not proposing a change to berserker. I really like the place in the game berserker has. I'm just curious as to how severe a nerf would balance the background.

(Emphasis mine.)

Interestingly, the OP's query about a naked/skill-less Troglodyte was answered (so far uncontroversially) by the very first reply -- using "anecdotal" data, no less! Also interestingly, if you accept that reasoning ("most experienced players agree that Trog, and no other god, is powerful enough to easily salvage the worst starting combos"), then you really have a pretty poor grounding for the rest of this over-serious discussion about whether statistics exist and how we should interpret them -- which stems from your (Berder's) assertion that Trog is already balanced; in other words, that the OP started with a faulty premise that can be corrected with "objective" data.

But the OP wasn't asking "How much of a nerf does Trog need so that Berserker is a balanced start?" -- the question before us is (was) "What would it take, hypothetically, to nerf Berserkers to balance the advantage they get from starting with Trog?" Just as Trog is a clear choice for salvaging poor combos, it seems pretty clear a priori that starting naked and skill-less wouldn't make much of a difference -- since you could just zerk everything on D:1 until you found a weapon and some body armor and then you're pretty much running a normal start. Starting with normal equipment and taking away the starting * of piety might not do it, since Trog piety builds quickly. And starting naked/skill-less with no piety would just make D:1 extremely punishing and force start-scumming. Thus "balance" would probably mean adding a disadvantage that would persist throughout the game, like a malus to all aptitudes, significantly lower starting stats, etc.

[Sandman25]

Also interestingly, if you accept that reasoning ("most experienced players agree that Trog, and no other god, is powerful enough to easily salvage the worst starting combos")

Don't forget that only first 5 winners get the bonus points. Trog is the fastest god because you don't care about which spell to cast/learn and should not channel/restore MP. You even have 2 ways to speed up real time game. It's great when you can kill anything without bothering to escape and can continue pressing 'o' and Tab.

Edit. I think Be can become balanced by moving Berserk to ***** though I would not like it happen.

[gammafunk]

If we consider games where players might actually choose Trog as their first god outside of some kind of challenge or high score attempt, which would be for Fighter and Adventurer class groups (Fi Gl Mo Hu As Ar Wn)

  Code:

!lm tenpercenters recent !boring !experimental cls=ar|fi|gl|as|hu|mo|wn god!= god!=beogh urune=0 xl<=15 br=temple|D lvl<=9 god.worship s=god
<Sequell> 789 milestones for tenpercenters (recent !boring !experimental cls=ar|fi|gl|as|hu|mo|wn god!= god!=beogh urune=0 xl<=15 br=temple|D lvl<=9 god.worship): 174x Okawaru, 85x Makhleb, 67x Dithmenos, 50x Qazlal, 50x Trog, 44x Cheibriados, 41x Fedhas, 38x Nemelex Xobeh, 34x Ru, 30x Ashenzari, 29x The Shining One, 29x Kikubaaqudgha, 25x Zin, 21x Vehumet, 21x Gozag, 17x Yredelemnul, 16x Elyvilon, 8x Xom, 5x Lugonu, 3x Sif Muna, 2x Jiyva
!lm tenpercenters recent !boring !experimental cls=ar|fi|gl|as|hu|mo|wn god!= god!=beogh urune<=3 rune s=god
<Sequell> 1228 milestones for tenpercenters (recent !boring !experimental cls=ar|fi|gl|as|hu|mo|wn god!= god!=beogh urune<=3 rune): 280x Okawaru, 111x Makhleb, 104x Dithmenos, 92x Trog, 90x Qazlal, 58x Cheibriados, 57x Fedhas, 57x Ru, 52x Nemelex Xobeh, 48x Kikubaaqudgha, 46x Ashenzari, 43x Vehumet, 42x The Shining One, 27x Jiyva, 26x Zin, 26x Gozag, 21x Yredelemnul, 17x Lugonu, 13x Elyvilon, 12x Xom, 6x Sif Muna


and then consider only gods with at least 20 games (somewhat arbitrary cutoff, but this doesn't matter much if I'm not trying to answer a question that's well-defined and with data that are sufficiently relevant to begin with), a pattern emerges when we look at "number of runes" / "number of games":

  Code:

        Vehumet            Trog          Qazlal              Ru   Kikubaaqudgha
          2.048           1.840           1.800           1.676           1.655
        Okawaru       Dithmenos       Ashenzari The Shining One          Fedhas
          1.609           1.552           1.533           1.448           1.390
  Nemelex Xobeh     Cheibriados         Makhleb           Gozag             Zin
          1.368           1.318           1.306           1.238           1.040


There, by adding a few more relevant conditions (but by no means doing the necessary work of filtering these games for weird game objectives or silly gameplay) and waving my hands around a bit (I should have just used a cutoff of 25 to ignore Vehumet), I've made the data fit my conclusion that Trog is one of the very best if not the best gods to use for non-zealots who might actually worship Trog. For those convinced by previous Sequell queries showing that Trog is not all that great, proceed to shower me with thanks now that I've opened your eyes.

[Sandman25]

Thank this post if you support previous post by gammafunk but cannot thank his post because you already believed Trog is the best god

[Sar]

thank my post if you are disgusted by blatant thankbait itt

[Sandman25]

thank my post if you are disgusted by blatant thankbait itt

Good catch
Seriously, I was going to thank gammafunk but (unusually for me) I have read the post to the end. Such posts should be thanked! I hope gammafunk will edit his post and I will thank it.

Edit. Also I admire your thankbait skills, I have much to learn from you

[gammafunk]

Just to be clear, I thanked Sandman25's post even though Sif is the best god and Sar's post because I meant to click report.

[tasonir]

Thank this post if you've read this thread at least a little bit but mostly held back from posting because you didn't want to embarrass everyone with how wrong they were when you finally point out that Cheibriados is the best god.

[Berder]

If we consider games where players might actually choose Trog as their first god outside of some kind of challenge or high score attempt, which would be for Fighter and Adventurer class groups (Fi Gl Mo Hu As Ar Wn)

  Code:

!lm tenpercenters recent !boring !experimental cls=ar|fi|gl|as|hu|mo|wn god!= god!=beogh urune=0 xl<=15 br=temple|D lvl<=9 god.worship s=god
<Sequell> 789 milestones for tenpercenters (recent !boring !experimental cls=ar|fi|gl|as|hu|mo|wn god!= god!=beogh urune=0 xl<=15 br=temple|D lvl<=9 god.worship): 174x Okawaru, 85x Makhleb, 67x Dithmenos, 50x Qazlal, 50x Trog, 44x Cheibriados, 41x Fedhas, 38x Nemelex Xobeh, 34x Ru, 30x Ashenzari, 29x The Shining One, 29x Kikubaaqudgha, 25x Zin, 21x Vehumet, 21x Gozag, 17x Yredelemnul, 16x Elyvilon, 8x Xom, 5x Lugonu, 3x Sif Muna, 2x Jiyva
!lm tenpercenters recent !boring !experimental cls=ar|fi|gl|as|hu|mo|wn god!= god!=beogh urune<=3 rune s=god
<Sequell> 1228 milestones for tenpercenters (recent !boring !experimental cls=ar|fi|gl|as|hu|mo|wn god!= god!=beogh urune<=3 rune): 280x Okawaru, 111x Makhleb, 104x Dithmenos, 92x Trog, 90x Qazlal, 58x Cheibriados, 57x Fedhas, 57x Ru, 52x Nemelex Xobeh, 48x Kikubaaqudgha, 46x Ashenzari, 43x Vehumet, 42x The Shining One, 27x Jiyva, 26x Zin, 26x Gozag, 21x Yredelemnul, 17x Lugonu, 13x Elyvilon, 12x Xom, 6x Sif Muna


and then consider only gods with at least 20 games (somewhat arbitrary cutoff, but this doesn't matter much if I'm not trying to answer a question that's well-defined and with data that are sufficiently relevant to begin with), a pattern emerges when we look at "number of runes" / "number of games":

  Code:

        Vehumet            Trog          Qazlal              Ru   Kikubaaqudgha
          2.048           1.840           1.800           1.676           1.655
        Okawaru       Dithmenos       Ashenzari The Shining One          Fedhas
          1.609           1.552           1.533           1.448           1.390
  Nemelex Xobeh     Cheibriados         Makhleb           Gozag             Zin
          1.368           1.318           1.306           1.238           1.040


There, by adding a few more relevant conditions (but by no means doing the necessary work of filtering these games for weird game objectives or silly gameplay) and waving my hands around a bit (I should have just used a cutoff of 25 to ignore Vehumet), I've made the data fit my conclusion that Trog is one of the very best if not the best gods to use for non-zealots who might actually worship Trog. For those convinced by previous Sequell queries showing that Trog is not all that great, proceed to shower me with thanks now that I've opened your eyes.

This is a pretty good approach in terms of restricting to physical fighters. I think you should look only at the 3rd rune instead of the number of runes 1-3. It doesn't matter if you can get the first rune with a god easily if the second and third are more difficult. However, accounting for that, Trog still does very well.

It shows Trog is among the best gods for non-spellbook non-zealot backgrounds. You can figure on around 10% uncertainty for sample sizes like this.

[Sprucery]

Qaz rocks, Makh sucks, clearly the undisputed truth.

[duvessa]

Just to be clear, I thanked Sandman25's post even though Sif is the best god and Sar's post because I meant to click report.

remember that time you thanked one of pleasingfungus' posts when you meant to hit delete?

[PleasingFungus]

remember that time you thanked one of pleasingfungus' posts when you meant to hit delete?

if gammafunk starts deleting my posts, I'm coming out of retirement to remove Sif

[edgefigaro]

Fun Fact: This thread was never about what "should" be done with Berserkers!

Curious, what does a balanced berserker start look like? If a char starts at the gate naked, with no items nor skills, but has trog, will they still be broken?

Caveat, I am not proposing a change to berserker. I really like the place in the game berserker has. I'm just curious as to how severe a nerf would balance the background.

(Emphasis mine.)

Interestingly, the OP's query about a naked/skill-less Troglodyte was answered (so far uncontroversially) by the very first reply -- using "anecdotal" data, no less! Also interestingly, if you accept that reasoning ("most experienced players agree that Trog, and no other god, is powerful enough to easily salvage the worst starting combos"), then you really have a pretty poor grounding for the rest of this over-serious discussion about whether statistics exist and how we should interpret them -- which stems from your (Berder's) assertion that Trog is already balanced; in other words, that the OP started with a faulty premise that can be corrected with "objective" data.

But the OP wasn't asking "How much of a nerf does Trog need so that Berserker is a balanced start?" -- the question before us is (was) "What would it take, hypothetically, to nerf Berserkers to balance the advantage they get from starting with Trog?" Just as Trog is a clear choice for salvaging poor combos, it seems pretty clear a priori that starting naked and skill-less wouldn't make much of a difference -- since you could just zerk everything on D:1 until you found a weapon and some body armor and then you're pretty much running a normal start. Starting with normal equipment and taking away the starting * of piety might not do it, since Trog piety builds quickly. And starting naked/skill-less with no piety would just make D:1 extremely punishing and force start-scumming. Thus "balance" would probably mean adding a disadvantage that would persist throughout the game, like a malus to all aptitudes, significantly lower starting stats, etc.

This is a more proper and complete way of stating my original intent with this thread. Thank you.

Starting with berserk gives you one of the -best- tools in the game for dealing with early threats. An enemy with an elec whip as the first creature encountered? You have a fighting chance with +50%hp, might, and haste available at l1. Adders are consistently regarded a threat worthy of running all sorts fsim's here on this forum, berserkers of all stripes crush them. By the time you see an early grindr you have trogs hand for MR.

Conceptually, the game is larger than just running Be over and over because it is strong. Strictly speaking, I assert that Fi and Gl starts are worse than berserker. Sometimes I play Fi and Gl though cuz I don't want trog, and magic is fun.

As far as the thought experiment of a balanced berserker, there are multiple ways to go about it. A direct no-skill/naked nerf as mentioned in the OP may not be enough, and you could continue on that inelegant path by reducing starting stats and providing a malus to aptitudes. That line of reasoning is an ugly way to go about things.

There are other nerfs though, such as nerfing trog, specifically berserk. The drawback to berserk can be made more severe. Consider adding noise to using the ability. The major advantage of berserk is that you can destroy an early game threat. If using the ability caused enough noise to bring more creatures to you and you end up engaging them when slow, the start would be weaker. A piety cost could be added to the skill, tuned to roughly ~3 berserks over the first two levels. This could be facilitated by altering piety gain specific for low HD creatures, allowing the piety cost to be negligible to a non-berserker character who finds an altar on d4.

Once the char gets to the midgame, things function more as normal. I don't have a lot to say about berserker in the lair-and-beyond environment, though I would welcome such discussion here. The thing about early game advantages is they don't always pan out. I've lost chars that have found stupidly good items, and I've done well on chars that have had hot garbage the whole game.

[tasonir]

Where did tedric's post go? Anyways, it's quoted now, so...The only malus I can think of which would address the early game is a substantial accuracy penalty while berserk is active, and possibly also for the exhaustion cooldown period. If it was made that berserking caused you to miss 60-80% of the time, then you'd have to risk actually landing a hit before your berserk expires and you die. In the late game you'd have enough skill to overcome the flat penalty, so the end game wouldn't be too much harder. This should cover trog being overpowered on D:1 without impacting post first rune gameplay very much at all. I don't think this is a needed change mind you, just answering the theoretical.

[edgefigaro]

(but by no means doing the necessary work of filtering these games for weird game objectives or silly gameplay).

True story: Berserker is my go-to background when I am playing Crawl drunk.

I play Be frequently, and I also do stupid stuff I would never consider on other classes like first rune slime, early depths, ect when I play Be. It is fun. Trog strong.

[edgefigaro]

Where did tedric's post go?

Halfway down the second page, five posts after the "I just came to read the comments" image. Roughly post 62-65.

[edgefigaro]

thank my post if you are disgusted by blatant thankbait itt

Good catch
Seriously, I was going to thank gammafunk but (unusually for me) I have read the post to the end. Such posts should be thanked! I hope gammafunk will edit his post and I will thank it.

Edit. Also I admire your thankbait skills, I have much to learn from you

I split up all my responses into different posts. I don't think it has the same effect though. I believe I have much to learn too.

[Sandman25]

Balanced berserker is called "Gladiator".

Berserker:

  Code:

+0 weapon of choice
+0 animal skin
Bread ration
Fighting: 3
Starting Weapon's Skill: 3
Armour: 2
Dodging: 2
Choosing Berserker adds 9 to your starting Strength, and 4 to your starting Dexterity, but reduces your starting Intelligence by 1.

Gladiator:

  Code:

+0 weapon of choice
+0 leather armour
+0 helmet
3 +0 throwing nets
Bread ration
Fighting: 2
Chosen Weapon's Skill: 3
Throwing: 2
Dodging: 3
Choosing Gladiator adds 7 to your starting Strength and 5 to your starting Dexterity.


[Siegurt]

Balanced berserker is called "Gladiator".

Berserker:

  Code:

+0 weapon of choice
+0 animal skin
Bread ration
Fighting: 3
Starting Weapon's Skill: 3
Armour: 2
Dodging: 2
Choosing Berserker adds 9 to your starting Strength, and 4 to your starting Dexterity, but reduces your starting Intelligence by 1.

Gladiator:

  Code:

+0 weapon of choice
+0 leather armour
+0 helmet
3 +0 throwing nets
Bread ration
Fighting: 2
Chosen Weapon's Skill: 3
Throwing: 2
Dodging: 3
Choosing Gladiator adds 7 to your starting Strength and 5 to your starting Dexterity.


Just because you don't mention it, Gladiator gets a better starting weapon of choice than berserker.

[edgefigaro]

Berserker also has trog, which is kinda the point. You can't have a balanced berserker without trog.

[duvessa]

In case you don't already want to kill yourself after reading this thread, here's another one: https://crawl.develz.org/tavern/viewtopic.php?f=8&t=691

[PleasingFungus]

In case you don't already want to kill yourself after reading this thread, here's another one: https://crawl.develz.org/tavern/viewtopic.php?f=8&t=691

also relevant

[tasonir]

In case you don't already want to kill yourself after reading this thread, here's another one: https://crawl.develz.org/tavern/viewtopic.php?f=8&t=691

also relevant

+Berserkers this space for rent

Answering the important questions.

[TeshiAlair]

Going back on topic a bit: why not stagger the berserk effects?

Level one could be something like:
Adrenaline rush: +X Might

Level two:
Surge of Rage: Everything but the speed boost

Level three:
Berserk: Full berserk

Getting to 3 stars with Trog is really easy, but it would make the first few levels of the game less of a tabstorm.

[Siegurt]

Hmm would the penalties also scale up or remain constant?

[jejorda2]

If I could get might at will with only exhaustion and no slowing or passing out as a penalty at * piety, I might wear and remove an amulet of faith to prevent reaching ** piety.

If there is no exhaustion penalty, then adding another duration of semi-berserk is a pretty good way of staying powerful all the time.

So I guess the blackout is really the only gradable penalty.

[bel]

I would like Trog's corpse sac mechanism to be less painful.

Berder's different queries are definitely a way to analyze Trog. The important thing is to not fixate on one particular query, but take lots of them in different ways, which Berder is correctly doing. Each of them can be criticized in certain ways, but if many of them point to a conclusion, it should be given some weight. This may or may not convince people. That is a problem of rhetoric. I recommend a good essay on this: Rhetoric of Economics by Deirdre Mccloskey.

That said, OP was talking about Berserkers. Zealots, starting with Trog. I don't think that anyone doubts that it is a great advantage. Of course this is CYC, and no harm in the thread drifting somewhat from the original.