Fight Nights Results and Schedule

This is a general open discussion for all ALFA, Neverwinter Nights, and Dungeons & Dragons topics.

Moderator: ALFA Administrators

Post Reply
Zelknolf
Chosen of Forumamus, God of Forums
Posts: 6139
Joined: Tue Jul 05, 2005 7:04 pm

Fight Nights Results and Schedule

Post by Zelknolf »

So the previous thread was getting cluttered. We can still chat there, but I'm going to use this one to post schedules, results/observations, paths to .erfs for builders interested, etc.

Next Fight Night:
Saturday, October 27 at 1600 U.S. Eastern (1300 Pacific, 2000 UTC)

Next Fight Night Builds:
Level 6, 16900 gold, Level Adjustment race required, no Prestige Classes. Levelup lever is in the start area.

Server Status:
Up and on the ADL.
Of note that the test server redeploys with the latest under-development haks server every night, even though it tells connecting clients to use the live version haks; this might result in quirky displays while creating characters. As a rule, you shouldn't take anything that looks suspicious.

(this won't be entirely consistent from week to week; we have some players who would like to participate for whom convenient times are mutually exclusive with convenient times for others interested)

Summary: On some nights, I host a small event on the ALFA Test Server, configuring it to point to the live version autodownloader for the evening and pitting fake/test characters against spawns. I try to make use of variety in these fights, so that I don't host an 'all undead' or 'all trolls' night, though the time it takes to set and reset fights means that an event will usually have three or four 'primary' fights, with resetting the fights with adjustments or coaching to see how it affects the outcomes. There are a few expected advantages of the event:
  • It is fun in the actual sense of the word-- consequence-free violence in a familiar setting provides light-hearted amusement.
  • It provides practice in combat for the player participants, making it less likely that they suffer a real character loss due to lacking player proficiency (as distinct from lacking character proficiency, which is a decidedly-more-valid reason to die).
  • It provides insight into the behavior of the AI, both from Obsidian and the ACR, allowing for improvements and debugging of the behaviors in realistic scenarios.
  • It provides a reasonably-controlled method to assess the actual challenge of known creatures from Dungeons and Dragons once they're translated to the specific medium that we play in (that is, not everything can be translated perfectly into the Aurora engine, and even if we do there are fundamental changes, like having real-time combat; the best we can say is that we'll try to get it close to what Forgotten Realms says it is).
  • It produces templates for our DM teams (that is, I'm making things for fight night; I don't intend to keep them to myself).
Erfs Produced:
-- Currently none.

Erfs to Come (and why they're not available yet):
-- Low-level hobgoblin adventurers (the tanks only meaningfully restrict one person's mobility)
-- Wraiths (they don't like using their touch attack; their touch attack will be a shadow's touch attack until the v1.89 release)
-- Zombified Soldiers (not quite as mindless as I would like)
-- Colliseum-style arena (scripts to prevent monsters in the pit from attacking people in the stands aren't working)
-- Hydras (require scripted way to get more than the default attacks/rd)

Results from previous fights in the replies, for interested parties.
Zelknolf
Chosen of Forumamus, God of Forums
Posts: 6139
Joined: Tue Jul 05, 2005 7:04 pm

Re: Fight Nights Results and Schedule

Post by Zelknolf »

Results from October 15
Builds: Level 5 PCs, 11700 gold no Prestige Classes

Large Elementals vs. 2 lv.5 PCs
Spoiler:
Side 1:
NPC Large Earth Elemental (CR 5), Obsidian AI
OR
NPC Large Fire Elemental (CR 5), Obsidian AI
OR
NPC Large Water Elemental (CR 5), Obsidian AI

Side 2:
PC H-orc Barbarian 5
PC Human FS 1/Clr 4

These fights were straightforward. Elementals, having a finite set of actions available to them, do not do anything particularly cleverly, nor do they have many options to do so. These three fights resulted in simple injury done to the party, and an expected depletion of spells/day to restore lost health. CR 5 is plausible

Side 1:
NPC Large Air Elemental (CR 5), Obsidian AI

Side 2:
PC H-orc Barbarian 5
PC Human FS 1/Clr 4

This fight ended quickly with little effect from the elemental. It spent its actions on its pulse attack, which was ineffective. Having insufficient AC/ DR/ hit points to survive the barbarian with his two-handed weapon, it fell with little resources expended. Implied CR might be questioned, depending on the details of the pulse attack.
All of the Baby Wyverns in the World vs. 2 lv.5 PCs
Spoiler:
Side 1:
NPC wyvern hatchling (CR 1), x20

Side 2:
PC H-orc Barbarian 5
PC Human FS 1/Clr 4

The wyverns simply did not have enough attack bonus to be much of a contest against heavy armor, and were ineffective. Hard to say what this would mean for a more-sensibly-leveled party facing a wyvern hatchling.
Wyvern vs. 2 lv.5 PCs
Spoiler:
Side 1:
NPC adult wyvern (CR 6), Obsidian AI

Side 2:
PC H-orc Barbarian 5
PC Human FS 1/Clr 4

The dice gods favored the PCs in this fight, and the wyvern was unable to make use of its most effective attack (poison). PCs were victorious, but the fight was quick and thus did not have enough dice thrown to account for randomness.
Young Green Dragon vs. 2 lv.5 PCs
Spoiler:
Side 1:
NPC Young Green Dragon (CR 5), Obsidian AI

Side 2:
PC H-orc Barbarian 5
PC Human FS 1/Clr 4

This fight was notably harder than the one against the wyvern, with the dragon's primary attack being successful and doing enough damage to visibly wear on the PCs. However, it is of note that the dragon did not attempt to use its mobility to its advantage, and did not make use of its breath weapon (the latter being by far its most powerful attack). CR feels a little low, but would seek more samples before saying for sure.
Troll vs. 2 lv.5 PCs
Spoiler:
Side 1:
NPC Troll (CR 5), Obsidian AI

Side 2:
PC H-orc Barbarian 5
PC Human FS 1/Clr 4

The dice gods hated this troll with much intensity, and the PCs killed it before enough dice were thrown to normalize for randomness. Hard to say how valid the CR is.
Bulette vs. 2 lv.5 PCs.
Spoiler:
Side 1:
NPC Bulette (CR 7), Obsidian AI

Side 2:
PC H-orc Barbarian 5
PC Human FS 1/Clr 4

PCs were utterly destroyed. The Obsidian AI was very insistant on keeping its original target down, and thus once one PC was on the ground, the other was forced to either solo the creature or heal (the latter only buying the healer one round, as the bulette returned to killing the healed one immediately, and was typically successful). Implication is that CR 7 is plausible.
Bulette vs. 3 lv.5 PCs.
Spoiler:
Side 1:
NPC Bulette (CR 7), Obsidian AI

Side 2:
PC H-orc Bariarian 5
PC Human FS 1/Clr 4
PC Dwarven Ftr 5 (tanky)

PCs were tapped, and required the -7 hp floor, but were victorous. The tanky fighter, when he was able to attract the bulette's ire, was able to hold it long enough to see the others healed, and its inability to do much other than chew on things meant it was eventually beaten. Implication is that CR 7 is plausible.
5 hobgoblins vs. 5 lv.5 PCs
Spoiler:
Side 1:
NPC Hobgoblin Fighter 3 (CR 3), ACR AI tank, x2
NPC Hobgoblin Ranger 3 (CR 3), ACR AI skirmish
NPC Hobgoblin Sorcerer 3 (CR 3), ACR AI nuke
NPC Hobgoblin Favored Soul 3 (CR 3), ACR AI medic

Side 2:
PC H-orc Bariarian 5
PC Human FS 1/Clr 4
PC Dwarven Ftr 5 (tanky)
PC Dwarven Ftr 2/Clr 3 (tanky)
PC Barbarian 5 (throwing weapons)

Fight was laughably short; some wounds were taken, but the NPC party was dispatched with little resistance. Implies that the hobgoblins' CR was lower than estimated.
10 hobgoblins vs. 5 lv.5 PCs
Spoiler:
Side 1:
NPC Hobgoblin Fighter 3 (CR 3), ACR AI tank, x4
NPC Hobgoblin Ranger 3 (CR 3), ACR AI skirmish, x2
NPC Hobgoblin Sorcerer 3 (CR 3), ACR AI nuke
NPC Hobgoblin Favored Soul 3 (CR 3), ACR AI medic

Side 2:
PC H-orc Bariarian 5
PC Human FS 1/Clr 4
PC Dwarven Ftr 5 (tanky)
PC Dwarven Ftr 2/Clr 3 (tanky)
PC Barbarian 5 (throwing weapons)

The AI didn't intelligently distribute the tanks, and that seems to have been part of this party's downfall; the unrestricted movement of the PC barbarian spelled a quick end to their squishier pals, and clumped on the dwarves (who were well-equipped to defend against swords). Some wounds taken and spells used, but ultimately had the feel of a buildup or filler fight.
15 hobgoblins vs. 5 lv.5 PCs
Spoiler:
Side 1:
NPC Hobgoblin Fighter 3 (CR 3), ACR AI tank, x5
NPC Hobgoblin Ranger 3 (CR 3), ACR AI skirmish, x4
NPC Hobgoblin Sorcerer 3 (CR 3), ACR AI nuke, x3
NPC Hobgoblin Favored Soul 3 (CR 3), ACR AI medic, x3

Side 2:
PC H-orc Bariarian 5
PC Human FS 1/Clr 4
PC Dwarven Ftr 5 (tanky)
PC Dwarven Ftr 2/Clr 3 (tanky)
PC Barbarian 5 (throwing weapons)

Again the tanks did not intelligently distribute, but this time clumped on the PC healer. Their greater numbers and the PCs' difficulty rescuing the healer gave the NPC spellcasters more time to put healing and protection on their front line. Fight looked to have been a moderate challenge-- implying along with the previous two fights that these hobgoblins are in fact CR 2.
40 hobgoblins vs. 5 lv.5 PCs
Spoiler:
Side 1:
NPC Hobgoblin Fighter 3 (CR 3), ACR AI tank, x10
NPC Hobgoblin Ranger 3 (CR 3), ACR AI skirmish, x10
NPC Hobgoblin Sorcerer 3 (CR 3), ACR AI nuke, x10
NPC Hobgoblin Favored Soul 3 (CR 3), ACR AI medic, x10

Side 2:
PC H-orc Bariarian 5
PC Human FS 1/Clr 4
PC Dwarven Ftr 5 (tanky)
PC Dwarven Ftr 2/Clr 3 (tanky)
PC Barbarian 5 (throwing weapons)

The chaos of this fight, combined with the presence of 20 NPCs capable of attacking at range (10 of them flawlessly), spelled a quick end to the majority of the party. The spellcasters ran out of spells before the tank was finished, though, and his possession of large quantities of simple healing consumables saved his hide for a good stretch after the fight was over for the rest of he party. However, it would have overwhelmed the supplies of all live ALFA PCs I know of, and retreat would have been the most plausible route for such a group to take. These numbers continue to imply CR 2, as about 32 CR 2 creatures would typically be enough to overwhelm a level 5 party.
1 wraith vs. 4 lv.5 PCs
Spoiler:
Side 1:
NPC Wraith (CR 5), ACR AI

Side 2:
PC Dwarven Ftr 5 (tanky)
PC Halfling Rgr 3/Rog 2
PC Half-orc Ranger 5 (melee)
PC Halfling Sorcerer 5

Wraith was bugged, plain and simple. It required possession to operate correctly, but its incorrect vulnerability to sneak attacks spelled its doom. Data not very useful; template needs fixing.
50 zombies vs. 4 lv.5 PCs.
Spoiler:
Side 1:
NPC Zombie (CR 0.5), equipped with sword and shield, ACR AI Mindless, x 20
NPC Zombie (CR 0.5), equipped with short spear and shield, ACR AI Mindless, x20
NPC Zombie (CR 0.5), equipped with halberd, ACR AI Mindless, x10

Side 2:
PC Dwarven Ftr 5 (tanky)
PC Halfling Rgr 3/Rog 2
PC Half-orc Ranger 3 (melee)
PC Halfling Sorcerer 5

Terrain seemed to make this fight read a little more difficult. The sheer number of bodies in the arena (circular, 2 tiles wide) kept the party from maneuvering wide enough to really use their superior movement to their advantage. Of course, this is a demonstration of zombies used tactically well (in huge quantities and tight quarters), and it took the party some time to leverage the mindlessness to their advantage (having found that the creatures will cluster and line up when presented with a choke point, and won't try to change tactics to gain access to those delicious brains). The party did win the fight, though, and did not lose any people, but at the cost of most of their use/day abilities. Total CR was likely about 6, which implies that arming zombies and deploying them like an intelligent master wants to get the most out of them inflates their CR to about 1 a piece.
Zelknolf
Chosen of Forumamus, God of Forums
Posts: 6139
Joined: Tue Jul 05, 2005 7:04 pm

Re: Fight Nights Results and Schedule

Post by Zelknolf »

Results from October 22
Builds: Level 5 PCs, 11700 gold no Prestige Classes

8 Hobgoblins vs. 5 lv.5 PCs
Spoiler:
Enemies:
NPC Hobgoblin Tank, level 3 (CR 2), ACR AI x3
NPC Hobgoblin Archer, level 3 (CR 2), ACR AI x3
NPC Hobgoblin Medic, level 3 (CR 2), ACR AI x2
NPC Hobgoblin Nuke, level 3 (CR 2), ACR AI x2

Adjustments made to the AI to make their reflexes and reaction times largely backfired with these hobgoblins, resulting in some of them simply opting to not fight while their friends were butchered.
20 Hobgoblins vs. 5 lvl 5 PCs
Spoiler:
Numbers matter very little, as the results were more severe than with the previous group; NPCs were easily dispatched.
1 Wraith vs. 5 lvl 5 PCs
Spoiler:
Wraith as last time, with new logic; wraith seemed to attempt its touch attack, but was easily derailed after taking an attack of opportunity. It then lingered on its melee attack until slain, and was a totally-underwhelming challenge as a result.
Hydras vs. 4 lvl 5 PCs
Spoiler:
A series of tests vs. hydras with various numbers of heads reveal that they are not particularly threatening without their full battery of attacks. Hydras will need to be updated before they're worth really using and represent the threat that they really should.
20 Zombies vs. 3 lvl 5 PCs
Spoiler:
Though zombies are correctly mindlessly pursuing the nearest source of brains still, the changes that sabotaged hobgoblins also made the back ranks of zombies pretty listless and anemic when it came to trying to reach those tasty tasty brains. On the plus side, not even trying to pathfind is more efficient than trying to pathfind smart-- it's just not very effective; they were beaten with little effort.
1 Great Wyrm Red Dragon vs. 3 lvl 5 PCs
Spoiler:
Unsurpisingly, this was mostly answered by running away. An ancient red dragon is serious enough serious business that it didn't even get any amusing cat and mouse going.

Not really useful data gained from this run, on account of more-than-typical quantities of lost rounds (OE AI does this to a certain degree, where large groups will simply fail to find actions and derp around for a bit) and some obvious losses in acquiring delays. Server logs also show unhandled exceptions in perception/ hearing events-- obvious performance loss there, and probably contributing to NPCs' unresponsive behavior.
Zelknolf
Chosen of Forumamus, God of Forums
Posts: 6139
Joined: Tue Jul 05, 2005 7:04 pm

Re: Fight Nights Results and Schedule

Post by Zelknolf »

October 27 was canceled; insufficient attendance to do anything..
Post Reply