DAMAGE IS DOUBLED IN .16

[Arrhythmia]

I DEMAND THAT ALL GAMES ARE FORCED TO USE GRUNT_LANG

FUTHERMORE, BALLISTOMYCETES MUST BE DESTROYED

[Pereza0]

For all who might think who this is not serious (its partly serious), from gammafunk:

He everyone,

We'd been noticing that the winrate was higher this tournament
compared to last, and recently wheals noticed that damage was doubled
compared to 0.15. The commit that introduced the bug was found to be:

0.17-a0-22-ge0bdd66
http://s-z.org/neil/git/?p=crawl.git;a= ... bdd66d849c

So player melee damage has been doubled for about two weeks now,
including all of the tournament.

The question is how to resolve this during the tournament. We could:

1) Leave it in-place for the duration of the tournament, and
immediately release 0.16.1 after the tournament ends. We can have the
binaries ready to go hopefully before the actual end so that there's
as little delay as possible.

2) Fix it now, and try to get servers updated quickly. Our South
Korean server (cwz) is hard to get updated since the admin (hong) is
not easy to reach, and we'd certainly have some delay on the Japanese
server (lld), since we have to contact the admin for rebuilds. So if
we did this, it'd be in place on some servers for a while longer
comapred to others, and we're not exactly sure how long it will take
cwz to get updated.

So far elliptic is in favor of (1), reasoning that it's a bit unfair
to change the game so drastically mid-tournament. I don't like leaving
crawl in a broken state for so long, but I don't have any particularly
good reason why we shouldn't do (1).

In either case we'll fix the big in trunk ASAP and will make an
announcement on the tournament page and cdo Wordpress as soon as we've
decided.

So if any developers feel strongly about either option (1) or (2),
please weigh in soon! We'd like to make an announcement for the
benefit of those playing in the tournament who may not have heard
through irc or WebTiles.

-Chris

(Sorry for derailing into something semi-serious)

Would it be completely out of the question to do another tournament later on?

Maybe end this one a bit earlier, make the fixed one a month or some weeks later and then compare the results of the final version to the crazy double damage one. In hindsight it was kind of obvious (it always is isn't it?), short blades suddenly being amazing, a new dude who was posting on /rlg/ transitioned from 0.13 to 0.16 and had a very easy time too

[bel]

Considering that I did not have time to participate in this tournament, a new tournament sounds great to me.

[Berder]

Two thoughts:
1. Wow, that's big news!
2. Nobody even noticed until now. That's what individual "experience" with the game is worth - nobody even noticed a doubling of melee damage for two whole weeks during a tournament! This is a definitive proof of how game and milestone data, for all its flaws, is far more trustworthy than what players think based on their experience.

[Sandman25]

Maybe end this one a bit earlier, make the fixed one a month or some weeks later and then compare the results of the final version to the crazy double damage one. In hindsight it was kind of obvious (it always is isn't it?), short blades suddenly being amazing, a new dude who was posting on /rlg/ transitioned from 0.13 to 0.16 and had a very easy time too

No matter if they fix the bug for the tournament or not, the tournament is already ruined
I am saying it as MuSu winner who didn't benefit from the bug as much as VSxx.

[Bloax]

Yep, damage got fucked by that commit.
Which would explain why I was two-shotting hell sentinels on the MiFE I won, since I found it quite unusual to be that strong.

[twelwe]

I thought this over since last night and concluded that I don`t really mind whether it gets fixed during or after the tournament. The fact of the matter is there will always be an asterisk next to eight the nine wins I have this month in my mind. I was happy to be not only on a good team this tournament but to contribute to it. I realize now that ievery team benefited from this, yet my wins feel cheap to me and that is all that matters. I just lost all desire to play, killed the halfling I had going (the last character my clan needed), and regardless of the timing of the fix I think my time in the tournament, friendly as its supposed to be, is over.

[Sar]

so that is why my OpSu is still alive

[Grudge]

One thing I would like to point out. The dummy commit for 0.17 was made three days before the commit that doubled melee damage.

Actually two things.
Trunk fsim with XL:1 GrFi vs yak

  Code:

               AvHitDam | MaxDam| Accuracy |AvDam | AvTime | AvSpeed | AvEffDam
Attacking:          8.0 |    28 |      66% |  5.3 |    138 |   0.72 |     3.9

0.16 fsim with the same GrFi and the same yak

  Code:

                AvHitDam | MaxDam| Accuracy |AvDam | AvTime | AvSpeed | AvEffDam
Attacking:           4.1 |    14 |      66% |  2.7 |    138 |   0.72 |     2.0

[Moose]

Quite an unfortunate situation. When I found this out, it was strange, I had nodesire to keep playing my current char, and I pretty much suicided it out of apathy.

Now, I'm done with the tournament, people are still having fun, which is great. The statistics for the tourney are relatively meaningless though.

Worst part is, I thought I was actually becoming good at the game, and realize that I still suck.

[Berder]

One thing I would like to point out. The dummy commit for 0.17 was made three days before the commit that doubled melee damage.

Actually two things.
Trunk fsim with XL:1 GrFi vs yak

  Code:

               AvHitDam | MaxDam| Accuracy |AvDam | AvTime | AvSpeed | AvEffDam
Attacking:          8.0 |    28 |      66% |  5.3 |    138 |   0.72 |     3.9

0.16 fsim with the same GrFi and the same yak

  Code:

                AvHitDam | MaxDam| Accuracy |AvDam | AvTime | AvSpeed | AvEffDam
Attacking:           4.1 |    14 |      66% |  2.7 |    138 |   0.72 |     2.0

Does this mean the problem never made it into 0.16, or does it mean it was recently fixed in 0.16?

[Grudge]

I am almost certain that this problem was never in 0.16.

EDIT: Actually was a recent fix. Oops.

[Sar]

Is 0.16 fixed on CXC already?

[gammafunk]

Yes, CXC has been updated, and we're working on getting the rest updated ASAP. We debated leaving the bug in place for the rest of the tournament out of fairness to games played from now on, but it was decided that playing a broken crawl is ultimately less fun and would drive many away from playing. The tournament results are already off, and keeping it in to make all games on parity doesn't fundamentally change this. There will soon be an update to the tournament page and wordpress with an announcement so everyone playing is aware.

Also LOL at the "this proves that gamedata are better..." comment, since it perfectly misunderstands the relationship between game data and "subjective experience".

[wheals]

If QuAd DaMaGe MaKeS eVeRy LeTtEr UpPeRcAsE, sHoUlDn'T dOuBlE dAmAgE mAkE eVeRy OtHeR LeTtEr UpPeRcAsE?

[Sar]

I started a new OpSu and found a +2 hand axe of draining next to Dungeon entrance and now I am killing things so fast it feels like the bug is still in

[Magipi]

I see the future:
Melee damage is halved, then nobody wins anything for a few weeks, then it turns out that damage wasn't double after all.

[mps]

Wow, so melee damage is doubled and most if not all players don't notice the difference?

Says a lot about crawl melee mechanics.

[rchandra]

Wow, so melee damage is doubled and most if not all players don't notice the difference?

Says a lot about crawl melee mechanics.

Lots of people noticed, which is why it was found. It just took a while to be sure.

[gammafunk]

Yes, I think I neglected to mention that in the CRD post, but lots of people were talking about it. wheals says he noticed it when using axe melee that felt overly strong, and another player ##crawl soon asked if people felt like they were doing more damage in melee, so that's when he decided to check with fsim. But even that wasn't the only mention of it and we got lots of responses that people felt their attacks were stronger. It's more that very few people go report bugs, even ones that outright crash their game repeatedly. This one made the player stronger, so it's even less likely to get reported.

Interestingly:

  Code:

 !lg * !meleebug !@bot !boring recent / won
< Sequell> 9048/839371 games for * (!meleebug !@bot !boring recent): N=9048/839371 (1.08%)

!lg * meleebug !@bot !boring recent / won
< Sequell> 1520/41316 games for * (meleebug !@bot !boring recent): N=1520/41316 (3.68%)



meleebug finds game that ended with the bug in either 0.16 or trunk, but can't catch games that benefited partially from it and transferred to a fixed version, so it can't catch everything.

[Berder]

Yes, I think I neglected to mention that in the CRD post, but lots of people were talking about it. wheals says he noticed it when using axe melee that felt overly strong, and another player ##crawl soon asked if people felt like they were doing more damage in melee, so that's when he decided to check with fsim. But even that wasn't the only mention of it and we got lots of responses that people felt their attacks were stronger.

By your own account, you first suspected something was fishy from looking at the tournament winrate. Then wheals eventually noticed it (after playing 20 games) and so did another unnamed person in crawl chat. After raising the issue you also got some responses from people who, after it was pointed out, reflected that they had seemed stronger. Hindsight is 20-20.

What about all the very good players who for two weeks didn't notice it, over 40 thousand games and 1520 wins? Personally after my 6-streak I had just assumed I was getting better at crawl and/or was getting lucky on artifact finds. It's easy to rationalize this kind of thing and it's not easy to notice if you aren't expecting it. Judging from the reactions in the tavern and in crawl chat, most people seemed surprised by this bug rather than expecting something like it.

It's more that very few people go report bugs, even ones that outright crash their game repeatedly. This one made the player stronger, so it's even less likely to get reported.

That's definitely one factor.

I think the main factor is it's just hard to notice. Whether you kill things in 2 hits or 4 hits is not a huge qualitative difference, because you don't have a firm expectation of how many hits it should take for your specific stats, skills, and weapon. Plus it's randomized, so it doesn't always take 2 hits. Improperly handled packs of monsters can still wreck you in short order, so you don't (usually) feel like you're just sailing through. So it's easy to attribute your success to other factors, like your god, your own skill, your lucky gear, etc.

[Sprucery]

If QuAd DaMaGe MaKeS eVeRy LeTtEr UpPeRcAsE, sHoUlDn'T dOuBlE dAmAgE mAkE eVeRy OtHeR LeTtEr UpPeRcAsE?

This reminds me that it's a long time since I played Adom...

[gammafunk]

No, that's a silly way to describe "my own account", since that CRD thread was to describe a problem where we needed to make a decision, not give testimony as to who said what when, which is irrelevant to making said decision. I do recall people mentioning player damage around the time that elliptic first mentioned tournament win rates, but players mention a lot of things. Anyhow it's not "hindsight" if players notice or suspect a problem and just don't report it; we're not exactly paying them to report bugs.

[Hirsch I]

is this issue fixed in trunk?

[Arrhythmia]

is this issue fixed in trunk?

Yes. Unlike the issue of ballistomycetes, which goes on unfixed.

[Hirsch I]

ballistomycetes are ok. Mimics, on the other hand, are not. fuck mimics.

[mps]

Lots of people can notice something and still not constitute most people. Berder's suggestion that it's easy not to notice the difference between 4 hits and 2 hits isn't quite the point, imo. The point is that the likelihood of your attacks that do a nontrivial/nonzero amount of damage actually killing something is improved in a way that's not readily noticeable to many players by doubling the damage they do. Spikiness in combat damage disguises the underlying mechanics to such an extent that doubling the absolute amount of damage is not readily noticed by a typical player.

I think it's strange that doubling the absolute amount of damage a player does in combat is not an immediately obvious effect for most players. I think it's doubly strange that I seem to be in the minority in thinking that.

[duvessa]

Isn't making damage difficult for humans to determine kind of the entire point of Crawl's melee damage mechanics?

[mps]

I can't claim to know what the point of crawl's melee mechanics is. It seems to me problematic to have a system at the heart of a game that has such counterintuitive properties.

[duvessa]

It's counterintuitive that damage is randomized?

[Scalding Steam]

It's counterintuitive that damage is randomized?

Having a damage formula so unbelievably complicated that fsim becomes the easiest way to determine whether melee damage has been doubled sounds rather counterintuitive to me.

[mps]

Exactly. Worse than that, though, most people don't even notice the difference in actual play. It's unreal.

[Sandman25]

Having a damage formula so unbelievably complicated that fsim becomes the easiest way to determine whether melee damage has been doubled sounds rather counterintuitive to me.

It is not really related to formula being complicated IMHO, it is related to too much info hidden by crawl. Consider the simplest form: damage is 1d(WeaponSkill) and it is doubled due to a bug. Now you have 1d20 instead of 1d10 for flail. Will it help you to notice the bug if all you see is "You hit Ogre. Ogre is lightly/heavily wounded"? This can happen with both d10 and d20 especially since Ogre(monster) has HP 16-38 and player is expected to be unaware of that.
The easiest way to notice the bug would be with some lucky rolls if exact damage was shown: "You hit Ogre (19 damage)". What, 19 damage with flail (base damage 10)? That's definitely a bug.

[Pereza0]

I think something should be done along those lines, but I am not so sure about just showing the damage being the way to go. That is how ToME4 and most other roguelikes do it and I kind of dislike it, you just get flooded with numbers and there is usually no point in wasting your time looking at them, and its hard to know how relevant they are when you don't know the exact health of the things you are hitting .

The exclamation mark system in DCSS is just as bad for reasons mentioned in this thread as well. The fact that something as vital as a double damage increase took two weeks to be rectified shows this, with a good system in place such a change would probably have been spotted and rectified within hours. That it took this long basically means that most players (myself included) have no idea of how effective their weapons are or should be.

I think a pip/percentage system could be interesting here as well (it seems to have worked for many things in Crawl right?). Game could give you two statistics while examining a monster, one would be the chance to land a hit on it with your wielded weapon (displayed with percentage) and the other would be the median health taken from that monster upon a successful hit (taking AC and brand into account) displayed through pips to avoid determination of max health.

I also wouldn't want the thread to get derailed by the whole data vs anectdote stuff (has happened before), I see potential for interesting brainstorming here

[Siegurt]

Note that effectively the exclamation points are a type of pip (not of median damage of course)

[Pereza0]

Yeah they are. But in practice its basically displaying the numbers, but instead of just displaying them you make them even more unintelligible by assigning them exclamation marks (based on an undisclosed, arbitrary rule).

Its good in the sense it avoids overwhelming the player with numbers (which is one of the things I like about Crawl). The bad thing is that the information is basically lost when you give it in that format.

[XuaXua]

I smell a quintuple-damage deployment for early April.

[mps]

The problem isn't the information given to the player. Yes, obviously the bug would have been noticed sooner, but the interesting thing is that it's not easy to tell the difference without the numbers. This suggests the mechanics are very loosely tuned, hard to analyze, and therefore difficult to discuss in relation to issues of design and balance.

Discussions on crawl design and advice that devolve into "no true statements about crawl" arguments are a reflection of peculiar random systems like this one.

[Kismet]

The problem isn't the information given to the player. Yes, obviously the bug would have been noticed sooner, but the interesting thing is that it's not easy to tell the difference without the numbers. This suggests the mechanics are very loosely tuned, hard to analyze, and therefore difficult to discuss in relation to issues of design and balance.

Discussions on crawl design and advice that devolve into "no true statements about crawl" arguments are a reflection of peculiar random systems like this one.

It's a bit of an emperor has no clothes moment.

[Sandman25]

I smell a quintuple-damage deployment for early April.

It should be doubled damage for player spells/missiles. Let casters/hunters have some fun too.

[scorpionwarrior]

Yeah they are. But in practice its basically displaying the numbers, but instead of just displaying them you make them even more unintelligible by assigning them exclamation marks (based on an undisclosed, arbitrary rule).

Its good in the sense it avoids overwhelming the player with numbers (which is one of the things I like about Crawl). The bad thing is that the information is basically lost when you give it in that format.

Exactly. Players quickly get used to the ! system being mostly useless, so no one pays attention to whether they're actually getting more !!! than usual or whatever.

[Lasty]

Lots of people can notice something and still not constitute most people. Berder's suggestion that it's easy not to notice the difference between 4 hits and 2 hits isn't quite the point, imo. The point is that the likelihood of your attacks that do a nontrivial/nonzero amount of damage actually killing something is improved in a way that's not readily noticeable to many players by doubling the damage they do. Spikiness in combat damage disguises the underlying mechanics to such an extent that doubling the absolute amount of damage is not readily noticed by a typical player.

I think it's strange that doubling the absolute amount of damage a player does in combat is not an immediately obvious effect for most players. I think it's doubly strange that I seem to be in the minority in thinking that.

I have a hammer. In an unrelated story, look at all the nails around here.

[mps]

I have a hammer. In an unrelated story, look at all the nails around here.

Yes, you do.

[archaeo]

I have a hammer. In an unrelated story, look at all the nails around here.

Yes, you do.

Man you guys are elevating concise and passive-aggressive venom to a whole new level, here, and this is the Tavern we're talking about.

[duvessa]

https://www.youtube.com/watch?v=8_rSzkAWckY
https://youtu.be/DIXyDl56DGQ?t=1m50s
https://youtu.be/GbGE4cLobVo?t=12s
https://www.youtube.com/watch?v=2ZYlRjQhFYY

[axaxaxas]

So which servers have been patched so far? Just CXC?

i like the idea of having another tournament fwiw

[tasonir]

Isn't making damage difficult for humans to determine kind of the entire point of Crawl's melee damage mechanics?

Yes. However, there are rebels amongst us who posit that this is a flaw, not a feature. The password is 'pizza' and meetings are held in the treefort next to the old mill, this is a temporary location until we can find something more secure.

[twelwe]

https://www.youtube.com/watch?v=8_rSzkAWckY

Lol

https://youtu.be/DIXyDl56DGQ?t=1m50s

heh

https://youtu.be/GbGE4cLobVo?t=12s

Lol.....

https://www.youtube.com/watch?v=2ZYlRjQhFYY

heh

[gammafunk]

We did a 0.16.1 bugfix release. Users not busy playing the DC:PT beta (that version was a way bigger disaster imo) are encouraged to upgrade from 0.16.0.

[Tressol]

Is this bug only on the Webtiles or is it in the version of .16 I downloaded to play offline as well?

Edit: Oh, Ninja'd, sort of.