to_hit

[Derakon]

You can change your to-hit values fairly easily too, remember. For example, chugging !Heroism and reading ?HolyChant usually gives me a 1-3% increased chance to hit, as well as slightly increasing the quality of critical hits I get (as shown by the 'I'nspect display for weapons).

[Magnate]

You can change your to-hit values fairly easily too, remember. For example, chugging !Heroism and reading ?HolyChant usually gives me a 1-3% increased chance to hit, as well as slightly increasing the quality of critical hits I get (as shown by the 'I'nspect display for weapons).

Yes. With hindsight I regret making this info available. Again one for info mode, I think.

[Derakon]

Ehh, personally I like being able to see the impact that changing my gear has on my combat capabilities.

[jens]

Do you have a script that you're generating these numbers with?

If so, it'd be nice to have a CSV file with the same AC ranges for all the monsters (and maybe an even larger monster selection).

If not, don't worry about it.

Nope just kicked up an Excel sheet.

[EpicMan]

Yes. With hindsight I regret making this info available. Again one for info mode, I think.

Why do you regret it?

I thought we were against make things opaque for the player. If we're going to show the player their to-hit numbers we might as well show the hit chance since without that the to-hit number is pretty useless, other than in a 'this is more so that must be better' sense.

If you know the to-hit formula you can calculate the hit chances by hand anyway.

[jens]

I'm afraid I don't really like the idea of inflating monster AC to rebalance to-hit. I fully accept that monster melee needs rebalancing, I just think that Jens's other ideas were better. I think it would be better to inflate monster attack skill by a constant factor (e.g. instead of 3x level + attack_type_constant, make it 5x level and adjust the constants). Since Jens already has the spreadsheet for various monsters it should be reasonably easy to adjust this way.

My suggestion is just the symmetrical consequence of the inflation you did for players. It will not be harder for players to hit the monsters, the effect of their ac will be exactly the same as today inside the formula. The only reason not to do this change is if there are other things that depend on the factors that are changing. I can think of two at the moment. Monster spells call the test_hit function as well, so those that use that function would hit players more often than today, i.e. it would get in line with how it was before the inflation, so should be another improvement (if there are such spells, I did not go through the spell list to check this). Another thing is if there are spells from players that depends on the monster ac in a what that does not depend on the test_hit formula, this I have not researched.

What I'm getting at is that this is not a change. This is a return to the situation that existed before you inlated the AC. So, come to think about it there is another reason not to do this change. If the game balance when it comes to to-hit was worse a year ago than it is today.

The problem by problem solution is good in some ways, but I will only be able to solve a few cases. I will not be able to find where the problem cases are and solve those, because the problem cases are everywhere a player has significantly higher ac than expected at that level. If I solved all the problems, case by case, the situation would be better than today, but the solution would not be as elegant, and there would probably creep up new issues with monsters hitting to well. So the way to go here would be to just solve a few of the problems and ignore the rest. If we just revert the situation we at least know there are several years of testing behind the to-hit values we revert to.

Problem cases are when your ac gets close to the power (3*lvl+effect) of a monster. If ac was used straigth in the formula you would get hit 5% when your ac reaches that power (and the last few points would each be a drastic improvement against that particular monster). In the current situation the limit is 33% higher than your ac, by deflating ac this limit would move to 100% higher than your ac.

At the moment I am a bit pressed for time. I will have time to code/test/push deflation tomorrow. I already have a program that can go through the monster list and modify values in it, so this will just take like 10 minutes to code. The testing and pushing will take longer, and all the thought process is done. I will not have time to get the spot solution done in time before I leave for the weekend. So I propose I do the work I can do tomorrow, then after the weekend, I could try the other aproach. Preferably after you had tried, and discarded the first solution though...

[jens]

Well now I have pushed it to GitHub :-)

When testing I didn't feel any change, which was the expected outcome of my test... I tested a couple of characters, one warrior, one mage.

Just a note on numbers: a change from say 70% to 90% is not a very big change, it means you will, on average, be hit 29% more often. On the other end of the scale however a change from 10% to 30% means that you will be hit 200% more often.

[Magnate]

Why do you regret it?

I thought we were against make things opaque for the player. If we're going to show the player their to-hit numbers we might as well show the hit chance since without that the to-hit number is pretty useless, other than in a 'this is more so that must be better' sense.

If you know the to-hit formula you can calculate the hit chances by hand anyway.

I know. I feel a bit ambivalent about this. OT1H I stopped playing Crawl precisely because of the deliberate opacity - it drove me mad. OTOH I really don't feel that @ should have an kind of iPhone-app availability of detailed knowledge of his combat percentages to a high degree of accuracy. "Oooh, my chance to hit emperor liches in melee just went from 67.3% to 69.4% - cool!".

I'm not sure what the middle ground is, but the more I think about turning the various info-related wizmode/cheat settings into a "detail mode" or "info mode", the more I like it. Takkaria might not (since it means another gameplay option), but I think it would make me more comfortable with having less detailed info in normal play, if there was a "detail mode" available for people who wanted it.

I have long been deeply persuaded by Timo's arguments about the different feel of frog-knows being largely (albeit not entirely) related to the reduced amount of info that was available.

[Netbrian]

I have long been deeply persuaded by Timo's arguments about the different feel of frog-knows being largely (albeit not entirely) related to the reduced amount of info that was available.

I think that to a certain extent, once you've made that information available, it's going to be difficult to get players to accept removing it again.

[Derakon]

Pav added the ability to sort by version to the ladder, so here's some stats from level-50 characters on the ladder:

Version 2.9.x: 169, 171, 200, 157, 180, 172, 167, 293 (!, only 178 from gear, so what's going on here?), 181, 174, 186, 151, 203, 177
Min: 151; max (discounting Mr. 293): 203; mean: 176.0; standard deviation: 14.2

Version 3.2.0: 279, 157, 200, 202, 249, 240, 194, 231, 189, 227, 195, 186, 193
Min: 157; max: 279; mean: 210.9; standard deviation: 31.2

So that's interesting. The mean AC is up about 20% compared to olden days. Here's the guy with an AC of 279; he's getting almost a third of it from Celeborn.

[jens]

So, I might have overshot a bit. I did a quick extension of my spread sheet, adding a col for a deflation of 1.25. I also extremified a few more AC values, to better see the effects close to the edge.

		lvl	Power				|	AC	Current	Def 1.5	Def 1.25
Morgoth		100	60	60	2	15	|	300	34,0%	54,4%	46,2%
		100	60	60	2	15	|	200	54,4%	67,9%	62,5%
		100	60	60	2	15	|	100	74,7%	81,5%	78,7%
Greater Balrog	79	10	10	60	15	|	300	16,9%	42,9%	32,5%
		79	10	10	60	15	|	200	42,9%	60,3%	53,3%
		79	10	10	60	15	|	100	69,0%	77,6%	74,2%
Atlas		76	60	10	60	10	|	300	17,3%	43,2%	32,8%
		76	60	10	60	10	|	200	43,2%	60,5%	53,6%
		76	60	10	60	10	|	100	69,1%	77,7%	74,3%
Master lich	41	5	15	0	0	|	150	16,1%	42,1%	31,6%
		41	5	15	0	0	|	100	42,1%	59,8%	52,7%
		41	5	15	0	0	|	50	68,9%	77,4%	73,9%
Stone giant	33	60	60			|	150	31,6%	52,5%	44,1%
		33	60	60			|	100	52,5%	66,7%	61,0%
		33	60	60			|	50	74,1%	80,8%	78,0%
Uruk		16	60	60			|	100	32,5%	53,3%	45,0%
		16	60	60			|	50	64,2%	74,2%	70,0%
		16	60	60			|	25	80,0%	85,0%	82,5%
Bullroarer	5	60	60			|	75	27,8%	50,6%	41,0%
		5	60	60			|	30	68,6%	77,0%	73,4%
		5	60	60			|	15	81,8%	86,6%	84,2%
Jackal		1	60				|	60	30,7%	52,1%	43,6%
		1	60				|	30	63,6%	73,6%	69,3%
		1	60				|	15	79,3%	85,0%	82,1%
Floating eye	1	2				|	9	5,0%	23,0%	5,0%
		1	2				|	5	41,0%	59,0%	41,0%
		1	2				|	0	95,0%	95,0%	95,0%

As I mentioned earlier, the differences at the high % end do not matter very much. What matters is the low end. So look at the Balrog for a good example:
Today, if you have very good AC the probability of being hit is 16,9%. What would we want it to be? 42,9%, or 32,5%? Keeping in mind that the lower value is close to the point were it too will colapse.

[d_m]

Magnate just committed your change at 1.5. I think 1.25 may work a bit better but I will play test.

If we can fix this object assertion we're going to release another RC nightly for people to try out. Will be good to get balancing feedback on it for changes like this one.

[Derakon]

Keep in mind that overshooting isn't inherently bad, since it makes the game harder. We just want to make AC be meaningful without making it overpowered.

[jens]

Keep in mind that overshooting isn't inherently bad, since it makes the game harder. We just want to make AC be meaningful without making it overpowered.

Precisely. I'd like AC to be as meaningful as possible, without being unbalanced. My feeling at the moment is that 1.25 is probably closer to the goal... Test a while and we'll see :-)

[TJS]

Ehh, personally I like being able to see the impact that changing my gear has on my combat capabilities.

Yep me too. I currently rarely bother using Bless or !Heroism because I can't judge what difference they actually make in combat.

You can change your to-hit values fairly easily too, remember. For example, chugging !Heroism and reading ?HolyChant usually gives me a 1-3% increased chance to hit

So having both only increases the chances of hitting something by only 1-3% ? I'm glad I never bother using them since that seems way underpowered to be worth it.