So How Badly Does The AI 'Cheat' The Dice?

[Eguzky]

I mean, I already know the Optional objectives are not meant to be completed; When I had 'Kill2 people & take their books', the AI team ran one of the 2 people across the map to ensure I never got to kill them before their team routed.
When I had a mission to 'Steal the enemy flag. Put it in our chest', the AI throw itself at me to keep me from moving. I killed ONE person, and they routed the second I touched the flag.

Optional objectives are meant to make you soak up unholy amounts of damage until you either lose, or win before completing them.

But I've also noticed; When facing the enemy leader on multiple occasions, when the AI is at 1-shot to die, my ToHit was 72%, 83%, 70%, 84%, 77%
..And I missed. 5 times in a row, with those percentages, in three separate battles.

But of course, the AI is laughing as they pound on 3 different people that are surrounding them, and not getting touched for round after round, after round.

So how badly does the AI cheat? Because it feels like they are a 7 year old making up rules to avoid losing.

[PDV]

Based on your example, probably not at all. Runs of extreme results like that happen when the dice are unfudged, and runs of bad luck are remembered much more than the runs of good luck where you get five hits in a row with 60% to hit.

Many games (League of Legends I remember, others I don't) deliberately fudge the dice toward the average - bad luck this round improves your luck next round until you are averaging the expected result, and vice versa for good luck this round. Because if you don't do that and leave your results good-and-proper random, you get frequent complaints like this one.

Next battle you play after reading this, copy out the entire log of every roll made, and split it by which rolls were for you and which were for the AI. I guarantee there won't be a statistically significant difference.

[Eguzky]

Originally posted by PDV:

Based on your example, probably not at all. Runs of extreme results like that happen when the dice are unfudged, and runs of bad luck are remembered much more than the runs of good luck where you get five hits in a row with 60% to hit.

Many games (League of Legends I remember, others I don't) deliberately fudge the dice toward the average - bad luck this round improves your luck next round until you are averaging the expected result, and vice versa for good luck this round. Because if you don't do that and leave your results good-and-proper random, you get frequent complaints like this one.

Next battle you play after reading this, copy out the entire log of every roll made, and split it by which rolls were for you and which were for the AI. I guarantee there won't be a statistically significant difference.

Yeah, you're right. My next match was somewhat in my favor.

It just feels frustrating, ever since we learned that X-Com 2 actually fudge dthe dice; if you got too many hits, the game would say 'There's your 95%, like it should be' but it forced a miss.
People found that in the code. The game would flat out make you miss if you hit too much. So now I get paranoid. :P

[Boris Hauer]

The AI has a little bonus on wounds and damage, and a noticeable bonus on Stupidity tests. That's all. It's all known of, there's no hidden mechanics to alter the rolls.

[GreyHuntr]

Other games use random generators that intentionally minimizes duplicate back to back results. This one does not, therefore allowing the possibility of streaks of good or bad rolls. The bad ones are just more noticeable due to confirmation bias.

[Eguzky]

Originally posted by GreyHuntr:

Other games use random generators that intentionally minimizes duplicate back to back results. This one does not, therefore allowing the possibility of streaks of good or bad rolls. The bad ones are just more noticeable due to confirmation bias.

Like today, missing 10 72% attacks in a row. That one person did NOT want to die. They had 2 bloody HP, 3 people attacking them, and just dodged everything. x.x

Still, I'm getting further as Chaos than I ever did as Skaven.

[GreyHuntr]

Since it has not been mentioned yet in this thread, I must say it just in case. Dodge rolls are made AFTER successful hit rolls are made. Is it possible some of this 72 percent chance hits actually landed, and then the enemy successfully dodged? The logs will tell. It can be confusing since dodge and miss use the same animation, so unless you are watching closely they appear the same.

[Eguzky]

Originally posted by GreyHuntr:

Since it has not been mentioned yet in this thread, I must say it just in case. Dodge rolls are made AFTER successful hit rolls are made. Is it possible some of this 72 percent chance hits actually landed, and then the enemy successfully dodged? The logs will tell. It can be confusing since dodge and miss use the same animation, so unless you are watching closely they appear the same.

It could very well be.

I do wish someone would make an updated, more polished version of this game, no offense to the devs.
The audio stutters when a map hits 90% loaded.
There's victory speeches being talked over by a generic 'You won!' audio line.
The controls are a bit clunky and unintuitive.
The waypoint system seems a bit clunky as well.

[27. Luftwaffle]

Why in hell AI has bonus? That should be optional. You miss enough allready that it feels like AI is really cheating. You want people to have fun or wreck their keyboard and screen? You should really fix that and make difficulty selection for the ones that need more difficulty. I play games to relax not for adrenaline rush.

Putting that a side, You really made awesome game and please keep up the good work. Love war hammer franchise.

[GreyHuntr]

The AI bonus to wounds and damage is directly tied to difficulty. When you select each mission it says Normal, Hard, Brutal, and Deadly. Demons have chance to show up on Brutal and Deadly. Also ambushes. The AI needs the help at higher ranks since it randomly assigns skills and has very sub-optimal builds. You see Marksmen in melee for example.

The bonus to stupidity checks is because the AI can't use the Guidance skill effectively. They still fail them constantly anyway though.

[mltnschroeder]

Playing as The Sisters, in all the dozens of battles my Augur hero has fought and started every round with her free perception ability, she has never once failed it even though she supposedly has a 95% success chance. So sometimes the RNG works in player's favor.

I generally try and rely more on tactics rather than the RNG, because good tactics are 100% under my control and should flatten out the bad luck curve. High init on all my units is always best so I get to move first and set up the situation, rather than reacting to the AI's moves. Also I take the Quick Incision (accuracy) and Demoralize (leadership) skills on every unit and all weapons & helms are enchanted with Rune of Dismay, because those debuffs stack with every hit. That way when I swarm one of his heroes with my mook henchmen, the RNG is working in my favor for him to even pass the All Alone test. And spam Healing Circle as often as possible to regain lost hp from lucky enemy hits. That way, even if I get unlucky for a round or two, I can always keep the situation under control enough to be able to fully recover.

Make sure to carry a Bugmans Ale on all your units, just in case you meet Undead faction or random demons until you can get Hearts of the Gryphon mastery. If you do switch to slower 2-handed weapons in mid-combat to deal more damage, compensate by quaffing a Crimson Shade potion, then switch back to your faster 1-handers when the melee is over to reassert init dominance.

Tactics will help you complete secondary goals as well. In many cases (if you have chosen a strongly unassailable position), once the AI's initial-contact gang-rush is defeated, it is reluctant to attack a strongly defended doorway with its remaining troops and it plays a bit more defensively. Sometimes the AI is fixated on one unit it can see, and as long as that unit remains stationary, the AI stays put as well in a sort of Mexican Standoff. You can then run other units from cover to cover, careful staying out of his movement range, and often scavenge most of the map at that point or look to complete side missions.

Originally posted by Eguzky:

It just feels frustrating, ever since we learned that X-Com 2 actually fudge dthe dice; if you got too many hits, the game would say 'There's your 95%, like it should be' but it forced a miss.
People found that in the code. The game would flat out make you miss if you hit too much. So now I get paranoid. :P

Although I watched a Youtube vid posted by a guy who completed 'an ironman impossible' campaign without ever losing a single soldier. But his amazing hit successes always when needed and the luck he had from aliens missing that should have killed him quickly defied the odds. I don't know what he did, but he cheated somehow, that much is apparent. So maybe the devs caught on and built in a fail-safe to prevent that.

[Carth]

the 5% RNGs are clearly skewed. they happen way to often.

[GreyHuntr]

Originally posted by Carth:

the 5% RNGs are clearly skewed. they happen way to often.

About 1 in 20 times probably.

[Boris Hauer]

Originally posted by 27. Luftwaffle:

Why in hell AI has bonus?

To compensate for its inability to learn and apply tactics, which humans are supposed to do.

[Tlewnipla]

Originally posted by mltnschroeder:

Playing as The Sisters, in all the dozens of battles my Augur hero has fought and started every round with her free perception ability, she has never once failed it even though she supposedly has a 95% success chance. So sometimes the RNG works in player's favor.

That's a bad example. The Augur's unique skill is to pass all perception tests automatically. The highest chance the game can show is 95% (which is the highest to-hit you can get e. g.), although in this particular case, it is actually 100% to pass the perception test.

[Boris Hauer]

Originally posted by Tlewnipla:

Originally posted by mltnschroeder:

Playing as The Sisters, in all the dozens of battles my Augur hero has fought and started every round with her free perception ability, she has never once failed it even though she supposedly has a 95% success chance. So sometimes the RNG works in player's favor.

That's a bad example. The Augur's unique skill is to pass all perception tests automatically. The highest chance the game can show is 95% (which is the highest to-hit you can get e. g.), although in this particular case, it is actually 100% to pass the perception test.

Not really. It can happen to fail a perception test with the Augur or with Chaos heroes with the Featureless Face mutation. It just doesn't matter because in both cases those tests are free, so you can try again and again without wasting Strategy Points. Trust me, it happened.