--- Log opened Sat May 06 00:00:00 2017 --- Day changed Sat May 06 2017 20170506 00:00:00-!- Appleman1234 [~Appleman1@pl24634.ag1212.nttpc.ne.jp] has quit [Ping timeout: 260 seconds] 20170506 00:17:27-!- sevu [~Unknown@141.39.226.226] has quit [Disconnected by services] 20170506 00:17:27-!- shiki [~Unknown@141.39.226.226] has joined #wesnoth 20170506 00:38:38-!- Narrat [~Narrat@93.237.22.213] has quit [Quit: They say a little knowledge is a dangerous thing, but it's not one half so bad as a lot of ignorance.] 20170506 00:48:19-!- Shiki_ [~Unknown@141.39.226.226] has joined #wesnoth 20170506 00:49:28-!- shiki [~Unknown@141.39.226.226] has quit [Ping timeout: 240 seconds] 20170506 01:07:56-!- ToBeCloud [uid51591@wikimedia/ToBeFree] has joined #wesnoth 20170506 01:18:12-!- shiki [~Unknown@141.39.226.226] has joined #wesnoth 20170506 01:21:40-!- Shiki_ [~Unknown@141.39.226.226] has quit [Ping timeout: 240 seconds] 20170506 01:51:31-!- Shiki_ [~Unknown@141.39.226.226] has joined #wesnoth 20170506 01:52:05-!- shiki [~Unknown@141.39.226.226] has quit [Ping timeout: 240 seconds] 20170506 02:23:21-!- shiki [~Unknown@141.39.226.226] has joined #wesnoth 20170506 02:24:40-!- Shiki_ [~Unknown@141.39.226.226] has quit [Ping timeout: 240 seconds] 20170506 02:36:05-!- aeth_ is now known as aeth 20170506 02:48:11-!- ArneBab [~quassel@55d45d89.access.ecotel.net] has joined #wesnoth 20170506 02:48:11-!- ArneBab [~quassel@55d45d89.access.ecotel.net] has quit [Changing host] 20170506 02:48:11-!- ArneBab [~quassel@freenet/developer/arnebab] has joined #wesnoth 20170506 02:52:48-!- ArneBab_ [~quassel@freenet/developer/arnebab] has quit [Ping timeout: 260 seconds] 20170506 02:55:39-!- shiki [~Unknown@141.39.226.226] has quit [Ping timeout: 260 seconds] 20170506 03:11:26-!- ToBeCloud [uid51591@wikimedia/ToBeFree] has quit [Quit: Connection closed for inactivity] 20170506 03:18:20-!- aeth [~Michael@wesnoth/umc-dev/developer/aethaeryn] has quit [Ping timeout: 260 seconds] 20170506 03:19:50-!- aeth [~Michael@wesnoth/umc-dev/developer/aethaeryn] has joined #wesnoth 20170506 05:02:51-!- edaq [~edaq3@h184-60-58-252.cytnin.broadband.dynamic.tds.net] has joined #wesnoth 20170506 05:35:58-!- Kwandulin [~Kwandulin@p200300760F6D803C90EFE9501B66FE53.dip0.t-ipconnect.de] has joined #wesnoth 20170506 06:26:36-!- claymore [~claymore@unaffiliated/claymore] has joined #wesnoth 20170506 07:03:24-!- zookeeper [~lmsnie@wesnoth/developer/zookeeper] has joined #wesnoth 20170506 07:20:55-!- Kwandulin [~Kwandulin@p200300760F6D803C90EFE9501B66FE53.dip0.t-ipconnect.de] has quit [Read error: Connection reset by peer] 20170506 07:21:40-!- claymore [~claymore@unaffiliated/claymore] has quit [Quit: Leaving] 20170506 07:54:30-!- edaq [~edaq3@h184-60-58-252.cytnin.broadband.dynamic.tds.net] has quit [Quit: Leaving] 20170506 08:02:22-!- Ravana_ [~Ravana@unaffiliated/ravana/x-2327071] has quit [Ping timeout: 240 seconds] 20170506 08:15:47-!- Kwandulin [~Kwandulin@p200300760F6D803C90EFE9501B66FE53.dip0.t-ipconnect.de] has joined #wesnoth 20170506 08:34:32-!- celmin [~celmin@unaffiliated/celticminstrel] has joined #wesnoth 20170506 08:34:32-!- celticminstrel [~celmin@unaffiliated/celticminstrel] has quit [Disconnected by services] 20170506 08:34:32-!- celmin is now known as celticminstrel 20170506 08:38:41-!- m_ [~m@89.77.194.225] has joined #wesnoth 20170506 08:39:14< m_> I know that mermen are supposed to be a little weaker, and that Mer Entangler’s main ability is Slow… But… 20170506 08:39:30< celticminstrel> But? 20170506 08:39:48< m_> Mer Entangler, a Lv 3! unit, has 8-3 melee blunt and 8-3 ranged blunt slow 20170506 08:40:00< m_> Isn’t that a little… excessively weak for a lv 3 unit? 20170506 08:41:31< celticminstrel> I have no idea what's "normal" for a given level. 20170506 08:41:52< zookeeper> don't forget that it also gets 7MP 20170506 08:44:19< m_> Maybe… Well, my feeling (only my feeling) is that a unit at lv. 3 should have at least one attack at 30 total, unless it is intentionally made useless for attacking, like E.Shyde 20170506 08:46:04< zookeeper> the question is, do you often feel like you don't want to get any netcasters (and then entanglers) and would be better off just getting spearmen or whatever other alternatives you have? 20170506 08:46:42< zookeeper> if yes then they might be too weak, but if you still find them worth getting and using, then they're clearly not too bad despite what their stats might suggest in isolation 20170506 08:47:23< m_> Slow is good enough to keep them useful. But, in that case, why don’t simply cap this evo line at lv 2? 20170506 08:48:12< m_> My point is, we could have a lv 3 unit with only one attack, that is 4-2 slow ranged, and it still would be useful if there was no alternative to slow 20170506 08:48:22< m_> But it would still be far to weak to be lv 3 20170506 08:55:25< zookeeper> sure, it is stats-wise particularly weak for a lvl3 unit. but then again the javelineer isn't very impressive either at 8-2 and 10-4. so considering the entangler has slow, they actually seem somewhat equal to me. 20170506 08:56:03< zookeeper> 8-2, really? urgh. that's really lousy for a spear-wielding unit :> 20170506 08:58:00< wesnoth-discord-> "why don’t simply cap this evo line at lv 2?" Level caps are pretty much random, dependent more on historical accidents in development that stick around, and availability of sprites and animations, than balance reasoning 20170506 08:59:00< wesnoth-discord-> Unit stats are balanced by the multiplayer 1v1 / 2v2 people, and they care very little about lvl3s 20170506 08:59:56< zookeeper> back in the good old days no outlaw units (except thieves) reached lvl3 20170506 09:00:49< m_> 10-4 is at least acceptable for a lv.3, and javelineer seems to be intentionally made useless for melee 20170506 09:01:02< wesnoth-discord-> Where is the balance of a unit like merman entangler most important? Probably in those ultra-script-heavy mp scenarios, where you can buy stats 20170506 09:01:31< wesnoth-discord-> But 1) of course you cannot balance a unit for arbitrary script-heavy stuff 20170506 09:02:04< wesnoth-discord-> 2) in such scenarios, where you can buy stats, special abilities are often more valuable than stats 20170506 09:03:20< m_> In campaigns units are often leveled up to lv3 20170506 09:04:05< m_> And this is where this balancement sometimes does matter 20170506 09:05:16< m_> Two more examples: Lv3 Dwarf Sentinel almost useless if compared to Lv3 Dwarvish Lord (or at least this was the case two years ago or so, didn't see if this changed) 20170506 09:05:27-!- Ravana_ [~Ravana@unaffiliated/ravana/x-2327071] has joined #wesnoth 20170506 09:05:37< m_> These two units are often given in campaigns in conjuntion 20170506 09:06:16< m_> This sucks, because in that case there is no reason to recruit Dwarf Guardsmen other than for cannon fodder (a role they excel at) 20170506 09:06:23< m_> There’s no reason to level them up 20170506 09:07:08< m_> Also, in HttT for example, one may be given, if I recall, the Poacher line ALONGSIDE with the Elv. Archer line 20170506 09:07:48< m_> Which sucks again, since every unit in the Poacher line is so much weaker than every unit in the Elv. Archer line its not even funny 20170506 09:08:16< wesnoth-discord-> Guardsmen are useful when you need to hold tiles in campaigns 20170506 09:08:23< wesnoth-discord-> They're somewhat resilient units... 20170506 09:08:35< wesnoth-discord-> I don't remember Httt maps, but I think there were some swaqmpy ones? Poacher has rare swamp defense 20170506 09:08:49< m_> Guardsmen. Not Sentinel. 20170506 09:09:20< m_> The problem, that screws this balancement, is that Dwarf Lords are EVEN better damage sponges than Sentinel. 20170506 09:09:21< wesnoth-discord-> Even in multiplayer, poachers live to fit certain terrain hexes perhaps even more than to shoot at enemies 20170506 09:10:00< m_> Guardsmen are damage sponges. Dwarf Fighters are melee fighters. ANd that's an OK balancement at lv 1. 20170506 09:10:53< m_> But at Lv 3, this is becoming screwed, because Lords are both damage sponges and melee attackers, while Sentines are ONLY damage sponges and in addtition they are usually worse damage sponges than Lords. 20170506 09:11:08< wesnoth-discord-> I agree with that 20170506 09:12:13< wesnoth-discord-> But again - the balancing people don't care about lvl3 at all, while for the script-heavy-rpgish-multiplayer players, abilities tend to be more valuable than stats 20170506 09:13:06< m_> It's not important for script-heavy-rpgish-multiplayer. Its important for single-player campaigns. SInce otherwise there is literally NO reason to level Guardsmen. 20170506 09:13:48< m_> While in campaigns the player should probably level both Guardsmen and Fighters, to make his army more diverse. 20170506 09:16:50< zookeeper> yeah, i agree there's plenty of inconsistency like that. 20170506 09:19:07< m_> My idea for the Poacher line, I agree it might be a little controversial… is like that: Poachers and their evolutions are hunters, right? So give them a nerfed version of Heals: Poachers and Trappers will have Heal +4, but ONLY if they’re standing on a forest tile, and Huntsmen and Rangers will have Heal+8, again Only if they’re standing on a forest tile. 20170506 09:19:20< m_> (they’ll share their game with their mates) 20170506 09:19:38< m_> This will make them not useless compared to Elvish Archer line. 20170506 09:20:20< m_> Also maybe give Ranger 8-4 rather than 7-4? 20170506 09:20:21< wesnoth-discord-> You're giving healing to Knalgans... 20170506 09:20:36< m_> Yeah, I know it’d be controversial. 20170506 09:21:24< m_> It’s just I’m looking for some kinda perk that would make them a tille more than a poor man’s E. Archer. 20170506 09:21:57< loonycyborg> dwarf fighter line gets better defenses in high tiers but still not hearly as good as guardsmen 20170506 09:22:19< loonycyborg> guardsmen are still needed for discouraging charges and stuff :P 20170506 09:23:09< wesnoth-discord-> He was talking about lvl3, not lvl1 20170506 09:23:38< loonycyborg> yes, it still has same dynamic 20170506 09:23:52< m_> nope 20170506 09:23:58< loonycyborg> Lords do get better resists but not nearly as insane as sentinels 20170506 09:24:21< m_> But loonycyborg! Sentinels have worse defenses and worse HP 20170506 09:24:33< m_> Let’s do the math to prove my point 20170506 09:24:48< loonycyborg> hmm I already forgot details 20170506 09:24:57< loonycyborg> guardsmen line had some passive 20170506 09:25:16< loonycyborg> to improve resistances 20170506 09:25:19< m_> They only have better resists if the passive is counted 20170506 09:25:24< m_> and they’re still worse 20170506 09:26:19< m_> The simplified formula for damage received can be considered as such: DmgReceived = IncomingDmg * Resists * Defens 20170506 09:26:49< m_> So, the simplified formula for how much dmg can a unit survive before dying can be considered as such: 20170506 09:27:49< celticminstrel> Um. 20170506 09:28:03< celticminstrel> That formula doesn't look right. 20170506 09:28:17< loonycyborg> lords have only crushing and slashing attacks 20170506 09:28:26< m_> why celtic? 20170506 09:28:33< loonycyborg> while sentinels are needed to provide piercing 20170506 09:28:43< celticminstrel> Well, first of all, defence is not a factor that the damage is multiplied by. 20170506 09:29:09< m_> Prob is, Sentinels' pierce damage SUCKS 20170506 09:29:23< wesnoth-discord-> For piercing, dwarfs have the shooter 20170506 09:29:43< loonycyborg> that shoot attack is extremely unreliable 20170506 09:29:47< loonycyborg> and ranged 20170506 09:30:00< m_> celtic sure. Im sorry, its (1-defs) rather 20170506 09:30:09< celticminstrel> ... 20170506 09:30:10< loonycyborg> not very good against cavalry charge 20170506 09:30:14< m_> Same for resists, (1-resis) 20170506 09:30:23< celticminstrel> That's an irrelevant detail. 20170506 09:30:26< loonycyborg> defense is just plain chance of being hit 20170506 09:31:01< celticminstrel> So let's say you have a sentinel on flat and a sentinel on castle. 20170506 09:31:03< wesnoth-discord-> on average, you can think of it as a multiplier 20170506 09:31:15< m_> My formula is getting better and better the more times a unit is attacked 20170506 09:31:43< m_> It approaches correctness with the amount of clashes. 20170506 09:31:52< celticminstrel> And so I don't have to deal with other units, let's pit them against each other. 20170506 09:32:01< celticminstrel> Using 9x3 spear. 20170506 09:33:06< celticminstrel> The steadfast ability doubles resistances on defense, right? 20170506 09:33:23< wesnoth-discord-> only up to 50% 20170506 09:33:40< m_> loonycyborg: OK, Sentinel vs Lord if charged by Grand Knight 20170506 09:33:51< celticminstrel> Okay, so the actual resistance is max(normal_resistance*2, 50%)? 20170506 09:33:57< celticminstrel> Or min sorry 20170506 09:34:00< m_> Sentinel: 9*3*1.2*2 = 64.8 total damage 20170506 09:34:32< m_> Lord: 15*3*0.8*2 = 72 total damage, oops greater than Sentinel’s 20170506 09:35:00< celticminstrel> Okay, sentinel on flat charged by grand knight, sure. 20170506 09:35:21< loonycyborg> sentinel on fortress charged by grand lord on flat more like 20170506 09:35:27< celticminstrel> At dusk so I don't need to deal with ToD modifiers. 20170506 09:35:38< celticminstrel> ...fine. 20170506 09:36:06< celticminstrel> They're both pierce. 20170506 09:36:38< celticminstrel> Sentinel as 30% resistance, which increases to 50% because steadfast. 20170506 09:36:55< celticminstrel> So it takes 8 damage per successful hit. 20170506 09:37:07< celticminstrel> Oh, but charge, so double that. 20170506 09:37:15< wesnoth-discord-> celticminstrel : min (max (normal_resistance * 2 , 0%) , 50%) 20170506 09:37:28< celticminstrel> I don't know which applies first. Either 16 or 17 damage per hit. 20170506 09:37:30< wesnoth-discord-> not true 20170506 09:37:41< m_> Anyway, the formula for damage sponginess can be, in simplification, described as such (I believe): HP * 1/(1-DEF) * 1/(1-RESIST) 20170506 09:37:45 * celticminstrel will assume 17. 20170506 09:37:47< wesnoth-discord-> steadfast still multiplies negative resists in some conditions 20170506 09:38:06< wesnoth-discord-> oh 20170506 09:38:18< m_> This formula can make such counts easier, I believe, and I also believe it is correct 20170506 09:38:20< wesnoth-discord-> rather than increasing them to 0, or even leaving them unchanged 20170506 09:38:25< m_> The more correct, the more clashes 20170506 09:38:33< celticminstrel> The sentinel also gets doubled damage, plus the knight has -20% resist, so he does 21 damage per hit. 20170506 09:38:48< celticminstrel> If I recall how negative resists work correctly. 20170506 09:39:16< wesnoth-discord-> Err yeah I screwed it up - but I thought it was min (max (normal_resistance * 2 , normal_resistance) , 50%) 20170506 09:39:30< celticminstrel> Sentinel's chance to hit is 60%. Knight's chance to hit is 40%. 20170506 09:39:35< m_> So, Sentinel on Mountains against Pierce: 20170506 09:39:43< wesnoth-discord-> Learn something every day! 20170506 09:41:42< m_> 68 * (1/(1-0.5)) * (1/(1-0.6)) = 340 20170506 09:41:52< m_> Lord on MOuntains against Pierce: 20170506 09:42:15< m_> 79 * (1/(1-0.3)) * (1/(1-0.7)) = 376,19 20170506 09:42:20< celticminstrel> So the sentinel can deal 0, 21, 42, or 63 damage. The knight can deal 0, 17, or 38 damage. 20170506 09:42:22< m_> *376.19 20170506 09:42:53< m_> Lord wins 20170506 09:43:00< m_> Now flat: 20170506 09:43:12< m_> Sentinel on Flat against Pierce: 20170506 09:43:34< m_> 68×(1÷(1−0.5))×(1÷(1−0.4)) = 226.67 20170506 09:43:39< m_> Lord on Flat against Pierce: 20170506 09:44:02< m_> 79×(1÷(1−0.3))×(1÷(1−0.3)) = 161.22 20170506 09:44:14< m_> OK Sentinel Wins, but you don't use dwarves on flat much anyway 20170506 09:44:30< m_> So now Lord on Flat against Blade: 20170506 09:44:33< celticminstrel> For the sentinel, I think E(DMG) = 0 * 0.6^3 + 21 * (0.4^2 + 0.6) * 3 + 42 * (0.4 + 0.6^2) * 3 + 63 * 0.4^3 20170506 09:44:59< Ravana_> I would say than noone would charge either of those units in mountains 20170506 09:45:21< loonycyborg> probably they'd be attacked with magic or poison 20170506 09:45:50< wesnoth-discord-> probably they'd be attacked by AI who does not think 20170506 09:45:57< celticminstrel> I'm not quite sure, did I mess up my probabilities? Something seems off about it. 20170506 09:46:45< loonycyborg> don't try to mess with probabilities 20170506 09:46:55< wesnoth-discord-> yes you messed up 20170506 09:46:59< loonycyborg> listing possible damages is clearer anyway 20170506 09:47:05< m_> I’d say we should give Sentinel max(Lord, Sentinel) defenses on any particular terrains, and Lord should be given Min(Lord, Sentinel) defenses on any particular terrain. Also swap Lord and Sentinels HP. 20170506 09:47:22< m_> This should, hopefully, fix the balancement issue. 20170506 09:47:26< celticminstrel> Possible damage 0, with probability 0.6^3, right? 20170506 09:47:39< wesnoth-discord-> > (0.4^2 + 0.6) 20170506 09:48:00< wesnoth-discord-> might be (0.6^2 * 0.4) instead 20170506 09:48:19< celticminstrel> But that's in the next term? 20170506 09:48:28< celticminstrel> Maybe the *3's are wrong? 20170506 09:48:38< wesnoth-discord-> or if (0.4^2 * 0.6), then 0.6^3 should be 0.4^3 20170506 09:49:40< celticminstrel> Okay, so CTH is 60%, and attacks are independent, so the probability of missing all three is 0.6^2 or 21.6%, right? 20170506 09:50:16< celticminstrel> The probability of missing exactly once is 3 * (0.6^2 + 0.4) or 43.2%. 20170506 09:50:43< celticminstrel> The probability of hitting exactly once is the reverse, 3 * (0.4^2 + 0.6) or 28.8%. 20170506 09:51:15< celticminstrel> Ah, that 0.6^3 is the probability of hitting all three. 20170506 09:51:21< wesnoth-discord-> If CTH is 60%, the chance to miss once is 40% and the chance to miss all three is 0.4^3 == 6,4 %. 20170506 09:51:30< celticminstrel> And that, yeah. 20170506 09:51:39-!- Kranix [~magnus@x1-6-74-44-01-e4-f1-52.cpe.webspeed.dk] has joined #wesnoth 20170506 09:52:31< wesnoth-discord-> 0 hits: 0.4^3 = 0.064 1 hits: 0.4^2*0.6^1*3 = 0.288 2 hits: 0.4^1*0.6^2*3 = 0.432 3 hits: 0.6^3 = 0.216 20170506 09:52:58< celticminstrel> So the P(0) = 6.4%, P(21) = 28.8%, P(42) = 43.3%, P(63) = 21.6% 20170506 09:53:30< celticminstrel> Thus E(DMG) = 0*6.4% + 21*28.8% + 42*43.2% + 63*21.6% 20170506 09:53:54< celticminstrel> Which yields 36. 20170506 09:54:45< celticminstrel> So on average the sentinel deals 36 damage to the knight. 20170506 09:54:48< Ravana_> 37.8 20170506 09:56:19< celticminstrel> For the knight hitting the sentinel, CTH is 40% and there are two strikes, so P(0) = 36%, P(17) = 48%, P(38) = 36%. 20170506 09:56:42< celticminstrel> So E(DMG) = 21. 20170506 09:57:04< celticminstrel> Then suppose the sentinel is replaced by a lord. 20170506 09:57:20< wesnoth-discord-> I dont think core has any attacks that change their damage between hits in same attack 20170506 09:57:31< celticminstrel> ??? 20170506 09:57:37< wesnoth-discord-> 17 and 38 20170506 09:58:01< celticminstrel> 38 is two hits. 20170506 09:58:19< loonycyborg> 34 20170506 09:58:29< celticminstrel> Oh. 20170506 09:58:33< wesnoth-discord-> @Ravana Cuttlefish is core and has Swarm 20170506 09:58:43< wesnoth-discord-> swarm does not affect damage 20170506 09:58:55< wesnoth-discord-> >_> nevermind 20170506 09:58:57< celticminstrel> So E(DMG) = 20 then. 20170506 09:59:15< celticminstrel> The lord's resistance is only 30% now instead of the sentinel's 50%, and it's using blade or impact. I'll suppose blade. 20170506 09:59:49< wesnoth-discord-> well, getting slowed midattack obviously still does 20170506 09:59:58< celticminstrel> So, knight does 17 base damage; modified for resistance and charge, that comes to 10, I think. 20170506 10:00:42< celticminstrel> Or, wait. 20170506 10:00:58< celticminstrel> 27. 20170506 10:01:09< celticminstrel> 20% resistance means multiply it by 80% 20170506 10:01:28< celticminstrel> Oh. 20170506 10:01:36< celticminstrel> 30%, so 23 damage. 20170506 10:01:57< celticminstrel> The lord's base 15 retaliation becomes 24 damage after charge and resistance. 20170506 10:02:48< celticminstrel> For the knight, E(DMG) = 23*48% + 46*36% = 27 20170506 10:04:23< celticminstrel> For the lord, E(DMG) = 24*28.8% + 48*43.2% + 72*21.6% = 43. 20170506 10:04:53< celticminstrel> Compared to the knight's 20 and sentinel's 36 in their match-up. 20170506 10:05:22< celticminstrel> So the lord clearly takes more damage than the sentinel on average. 20170506 10:05:34< celticminstrel> It also deals more damage on average to the knight. 20170506 10:06:09< celticminstrel> If nothing else, this shows that the formula used here is way off: 20170506 10:06:10< celticminstrel> [May 06@05:34:01am] m_: Sentinel: 9*3*1.2*2 = 64.8 total damage 20170506 10:06:11< celticminstrel> [May 06@05:34:32am] m_: Lord: 15*3*0.8*2 = 72 total damage, oops greater than Sentinel’s 20170506 10:07:02< celticminstrel> Though I'm not entirely sure whether that was supposed to be damage taken or damage dealt. 20170506 10:07:36< Ravana_> technically, you would need to apply rounding to each hit separately 20170506 10:07:42< m_> what's wrong with my formula? 20170506 10:07:51-!- vultraz_iOS [uid24821@wesnoth/developer/vultraz] has quit [Quit: Connection closed for inactivity] 20170506 10:08:05-!- Lirion [~m00se@wikimedia-commons/Lirion] has quit [Quit: echo 16i[q]sa[ln0=aln100%Pln100/snlbx]sbA0D4D465452snlb xq |dc] 20170506 10:08:05< celticminstrel> But either way it's off. For damage taken, the average is sentinel = 20, lord = 27; for damage dealt, it's sentinel = 36, lord = 43. 20170506 10:08:39< wesnoth-discord-> @Ravana "steadfast still multiplies negative resists in some conditions" are you sure? I just tested it (1.12.6) - I bought Steadfast in the Colosseum addon for a saurian, and vulnerabilities are not modified. Similarly Help says that Steadfast does not modify vulnerabilities. It seems to be: min (max (normal_resistance * 2 , normal_resistance) , 50%) 20170506 10:08:57< celticminstrel> m_: Your formula isn't correctly accounting for the defense. 20170506 10:08:59< wesnoth-discord-> I did say in some condiations, not that the conditions are found in core 20170506 10:09:32< wesnoth-discord-> oh. 20170506 10:09:47< celticminstrel> m_: I'm not sure, there may be other problems too. For example, did you account for charge doubling damage for both combatants? 20170506 10:10:16< loonycyborg> nonetheless it's weird that lord gets higher base hp than sentinel 20170506 10:10:28< loonycyborg> I'd expect the other way around 20170506 10:10:31< celticminstrel> Or the increased resists resulting from steadfast? 20170506 10:10:50< m_> celtic: I purposefully ignored all conditions that are SAME for Sentinel and Lord and therefore not useful to compare the two unis. 20170506 10:11:03< celticminstrel> (Does the attack predictions dialog calculate the expected value for you?) 20170506 10:11:16< m_> So, when counting the damage output, I purposefully ignored Knight’s defenses and charge. 20170506 10:11:16< wesnoth-discord-> it does 20170506 10:11:19< celticminstrel> m_: Well, the only condition that's the same is charge. 20170506 10:11:20< wesnoth-discord-> celticminstrel: No. 20170506 10:11:21< wesnoth-discord-> but I dont think it displays it 20170506 10:11:34< celticminstrel> @jyrkive It doesn't calculate it? 20170506 10:11:36< wesnoth-discord-> it has to calculate it, to choose preselected attack 20170506 10:11:52< celticminstrel> It calculates damage distribution. 20170506 10:11:59< m_> When counting damage sponginess, I took into account Sentinel and Lord’s HP, resists and defense. 20170506 10:12:08< celticminstrel> Which I guess... is one step away from the actual expected value. 20170506 10:12:32< m_> I maintain that Sentinel is better than Lord, in defense, on Flat. Worse on Mountains. ABout the same elsewhere. 20170506 10:12:58< m_> I also maintain that: 20170506 10:12:59< m_> I’d say we should give Sentinel max(Lord, Sentinel) defenses on any particular terrains, and Lord should be given Min(Lord, Sentinel) defenses on any particular terrain. Also swap Lord and Sentinels HP. 20170506 10:13:04< celticminstrel> That sounds pretty doubtful. 20170506 10:13:25< celticminstrel> Swapping their HP sounds okay, the other stuff seems silly. 20170506 10:13:40< wesnoth-discord-> celticminstrel, @Ravana : to select the default attack, the mouse event handler does someting more complex than merely calculating the average damage. 20170506 10:13:46< m_> why should Lord have greater defenses on Mountains than Sentinel. 20170506 10:13:52< wesnoth-discord-> The primary criterion is chance to kill. 20170506 10:14:15< loonycyborg> iirc sentinels had 70% on mountains and forts before 20170506 10:14:26< celticminstrel> m_: Why should sentinel have freater defenses on mountains than lord? 20170506 10:14:30< loonycyborg> either I'm misremembering or they got nerfed 20170506 10:14:47< wesnoth-discord-> https://github.com/wesnoth/wesnoth/blob/7d68b4ffba8c752e62994e8fc5b9ae25387b8019/src/actions/attack.cpp#L445 20170506 10:14:52< m_> celticminstrel: Because it’s Sentinel’s job to be a damage sponge, and not Lord’s. 20170506 10:14:54< celticminstrel> @jyrkive But IIUC you can fairly easily calculate the expected value just from the graph, right? Well, provided it's not truncated. 20170506 10:15:08< celticminstrel> m_: But it has the high resists for that. 20170506 10:15:29< loonycyborg> hehe lord has almost the same 20170506 10:15:32< loonycyborg> and more hp 20170506 10:15:32< m_> But for them to be effective they must not be countered by worse defenses. 20170506 10:15:50< wesnoth-discord-> celticminstrel: Yes, the expected damage can be calculated from the distribution. 20170506 10:15:51< m_> Also the difference between 60% and 70% is pretty large. 20170506 10:16:00< celticminstrel> On defense, lord has 40% blade, 30% pierce and impact, 10% energy, 20170506 10:16:04< wesnoth-discord-> As @Ravana said, it's not shown anywhere, though. 20170506 10:16:13< m_> Larger than between 50% and 60%. 20170506 10:16:47< celticminstrel> Whereas the sentinel has 50% blade and pierce, 40% impact, 20% energy. 20170506 10:17:07< m_> I maintain that this formula is a good approximation: EffHP = BaseHP * (1/(1-Def)) * (1/(1-Resist)) 20170506 10:17:17< celticminstrel> So its resistances are better all around when steadfast is factored in, and even when it's not, they're only slightly worse. 20170506 10:17:28< loonycyborg> very little difference 20170506 10:17:44< celticminstrel> I don't think using the defense as a multiplier could possibly give you a good approximation. 20170506 10:17:50-!- Lirion [~m00se@wikimedia-commons/Lirion] has joined #wesnoth 20170506 10:18:08< loonycyborg> I agree, probabilitys can be confusing 20170506 10:18:18< celticminstrel> Also why reciprocals. 20170506 10:18:28< wesnoth-discord-> Gotta go, continue the discussion without me. (I'm telling that because the IRC-Discord bridge doesn't tell when I leave.) 20170506 10:18:29< m_> celticminstrel: again, it approaches correctness with the number of hits 20170506 10:18:56< celticminstrel> That sounds quite doubtful. 20170506 10:19:34< m_> celticminstrel: If you have a coin, there’s 50% chance you get tails and 50% chance you get heads, right? 20170506 10:19:45< celticminstrel> Yeah? 20170506 10:19:54< loonycyborg> the best overview would be if you calculated theoretical max 20170506 10:20:01< m_> So, we can presume, if you toss a coin enough times, we’ll get about the same amount of tails and heads. 20170506 10:20:23< m_> But now, someone tosses a coin two times and gets two heads. 20170506 10:20:28< m_> Bang, m_ is wrong! 20170506 10:20:58< m_> The problem is, that it is mathematically correct that given enough tosses, the number of tails WILL be about the same as the number of heads. 20170506 10:21:15< m_> It might be a bad approximation for two tosses. 20170506 10:21:21< celticminstrel> Not wrong, you just didn't toss it enough times. :P 20170506 10:21:32< m_> It will be a good approximation for two hundred tosses. 20170506 10:21:50< m_> I think it should be same for applying defenses. 20170506 10:23:57< m_> As for the reciprocals… The unit dies, when HP = DmgOutput * (1-Defs) * (1-Resists) 20170506 10:24:42< m_> So, equivalently… the unit dies when HP * (1/(1-Defs)) * (1/(1-Resists)) = EnemyDmgOutput 20170506 10:25:07< celticminstrel> Ah. 20170506 10:25:32< m_> The unit dies when the enemy, counting the Defs and Resists, has dealt as much damage as the unit has HP. 20170506 10:25:49< celticminstrel> I think your coin analogy is flawed though. 20170506 10:26:05< m_> It’s an approximation only. 20170506 10:26:40< m_> A unit with 70% defs can withstand TONS of dmg if lucky, but very little dmg if unlucky. 20170506 10:27:30< m_> But an army of 70% defs units will, on average, withstand just as much damage as an army of units who are always being hit (no random) but have 70% resists. 20170506 10:28:08< celticminstrel> That sounds wrong. 20170506 10:28:17< m_> why? 20170506 10:28:29< celticminstrel> A unit that has 0% defense but 70% resists will take 70% of the damage, exactly. 20170506 10:29:24< celticminstrel> For a unit with 70% defense byt 0% resists... 20170506 10:29:34< wesnoth-discord-> there are more attacks that can bypass defense than those that can bypass resists 20170506 10:29:35< celticminstrel> ^but 20170506 10:30:21< celticminstrel> For each hit, the unit either takes full damage or no damage. 20170506 10:31:00< celticminstrel> So if they take 100 hits, the expected value should be... 20170506 10:31:31< celticminstrel> 50d*0.7 + 50d*0.3? 20170506 10:32:05< celticminstrel> No. 20170506 10:32:13< celticminstrel> That's max damage which comes out to 100d. 20170506 10:34:30< m_> OK, I’ll grant you that. 20170506 10:35:26< m_> But that’s the only way to count the unit’s "damage sponginess" without taking into account every possible enemy. 20170506 10:35:44-!- vultraz_iOS [uid24821@wesnoth/developer/vultraz] has joined #wesnoth 20170506 10:36:08< celticminstrel> I guess it's a binomial distribution. 20170506 10:36:16< celticminstrel> Or something. 20170506 10:37:02< celticminstrel> I think it ends up as some (x + d)^100 type thing. 20170506 10:37:43< m_> This will get tough, since then we must take into account enemy’s dmg per hit and number of hits 20170506 10:39:06< celticminstrel> Yeah. 20170506 10:39:38< m_> OK… My formula holds when the unit can’t be killed in one turn, which is a very idealistic scenario. Granted, if a unit can be killed in one turn, my formula proves useless for things like strategic unit placement on board. 20170506 10:40:53< celticminstrel> I wonder just how approximate your formula really is. 20170506 10:41:09< m_> I think it is still not useless for unit balancement, since it allows to consider a unit in isolation from other conditions. 20170506 10:41:14< celticminstrel> But I can't be bothered to remember how to do all the calculations, let alone actually do them. 20170506 10:41:31< wesnoth-discord-> eg Slow and Drain help "spongyness" 20170506 10:42:03< wesnoth-discord-> there's probably a lot of stuff that you could count into it, but that are really hard to quantify for a general case 20170506 11:09:49-!- chiec [~chiec@2a01:c50e:3509:1f00:1f05:155b:5053:7345] has joined #wesnoth 20170506 12:24:04-!- Kwandulin [~Kwandulin@p200300760F6D803C90EFE9501B66FE53.dip0.t-ipconnect.de] has quit [Ping timeout: 255 seconds] 20170506 12:24:27-!- minzbonbon [~min@meta23.net] has quit [Ping timeout: 240 seconds] 20170506 12:26:42-!- minzbonbon [~min@meta23.net] has joined #wesnoth 20170506 13:06:13-!- m_ [~m@89.77.194.225] has quit [Quit: Leaving] 20170506 13:29:44-!- mattsc [~mattsc@wesnoth/developer/mattsc] has joined #wesnoth 20170506 13:32:28-!- Smedles [~quassel@CPE-58-160-77-59.ssqn1.lon.bigpond.net.au] has quit [Ping timeout: 260 seconds] 20170506 13:34:51-!- Smedles [~quassel@CPE-58-160-77-59.ssqn1.lon.bigpond.net.au] has joined #wesnoth 20170506 13:53:35-!- minzbonbon [~min@meta23.net] has quit [Ping timeout: 240 seconds] 20170506 13:54:31-!- minzbonbon [~min@meta23.net] has joined #wesnoth 20170506 13:58:19-!- prophile [~alynn@oftn/oswg-member/prophile] has joined #wesnoth 20170506 14:43:55-!- Kwandulin [~Kwandulin@p200300760F6D801B055C4A8B6D11E63D.dip0.t-ipconnect.de] has joined #wesnoth 20170506 14:52:30-!- THERetroGamerNY [THERetroGa@pool-68-133-48-135.bflony.fios.verizon.net] has joined #wesnoth 20170506 16:01:42-!- claymore [~claymore@unaffiliated/claymore] has joined #wesnoth 20170506 16:09:34-!- Appleman1234_ is now known as Appleman1234 20170506 16:15:27-!- janebot [~Gambot@unaffiliated/gambit/bot/gambot] has quit [Remote host closed the connection] 20170506 16:15:33-!- janebot [~Gambot@unaffiliated/gambit/bot/gambot] has joined #wesnoth 20170506 16:16:38-!- mattsc [~mattsc@wesnoth/developer/mattsc] has quit [Quit: So long and thanks for all the fish.] 20170506 16:18:45-!- THERetroGamerNY [THERetroGa@pool-68-133-48-135.bflony.fios.verizon.net] has quit [Quit: Be Blessed!] 20170506 17:07:51-!- vultraz_iOS [uid24821@wesnoth/developer/vultraz] has quit [Quit: Connection closed for inactivity] 20170506 17:35:30-!- DeFender1031 [~DeFender1@46-116-209-76.bb.netvision.net.il] has joined #wesnoth 20170506 18:03:15-!- ToBeCloud [uid51591@wikimedia/ToBeFree] has joined #wesnoth 20170506 20:11:27-!- ToBeCloud [uid51591@wikimedia/ToBeFree] has quit [Quit: Connection closed for inactivity] 20170506 20:25:03-!- claymore [~claymore@unaffiliated/claymore] has quit [Quit: Leaving] 20170506 21:06:46-!- prophile [~alynn@oftn/oswg-member/prophile] has quit [Quit: The Game] 20170506 21:11:46-!- janebot [~Gambot@unaffiliated/gambit/bot/gambot] has quit [Remote host closed the connection] 20170506 21:11:52-!- janebot [~Gambot@unaffiliated/gambit/bot/gambot] has joined #wesnoth 20170506 21:43:18-!- Appleman1234 [~Appleman1@pl26691.ag1212.nttpc.ne.jp] has quit [Ping timeout: 260 seconds] 20170506 21:54:35-!- Smedles [~quassel@CPE-58-160-77-59.ssqn1.lon.bigpond.net.au] has quit [Ping timeout: 255 seconds] 20170506 21:57:11-!- Smedles [~quassel@CPE-58-160-77-59.ssqn1.lon.bigpond.net.au] has joined #wesnoth 20170506 22:10:54-!- chiec [~chiec@2a01:c50e:3509:1f00:1f05:155b:5053:7345] has quit [Remote host closed the connection] 20170506 22:15:15-!- Kwandulin [~Kwandulin@p200300760F6D801B055C4A8B6D11E63D.dip0.t-ipconnect.de] has quit [Read error: Connection reset by peer] 20170506 22:36:48-!- vultraz_iOS [uid24821@wesnoth/developer/vultraz] has joined #wesnoth 20170506 22:56:21-!- zookeeper [~lmsnie@wesnoth/developer/zookeeper] has quit [Ping timeout: 240 seconds] 20170506 23:09:15-!- Appleman1234 [~Appleman1@pl7533.ag1212.nttpc.ne.jp] has joined #wesnoth 20170506 23:11:20-!- sevu [~Unknown@p548545AE.dip0.t-ipconnect.de] has joined #wesnoth 20170506 23:22:05-!- sevu [~Unknown@p548545AE.dip0.t-ipconnect.de] has quit [Remote host closed the connection] 20170506 23:50:18-!- prophile [~alynn@oftn/oswg-member/prophile] has joined #wesnoth --- Log closed Sun May 07 00:00:52 2017