Were the Statistics and Various Calculations Actually Tested?

[YeahThatGuy]

A week or so ago, after almost a year of not touching Blades of Avernum, I regained my interest in the game. While my resurrected interest in the game was more for playing the game, when I began to create new characters and ponder the paths they'll take it led me to question what the real, actual effects of the various races, skills, traits, et cetera have on the game.

 

After careful testing with a quickly-created scenario and various 1-1 damage weapons I discovered a few surprises. I searched the Blades of Avernum boards on this forum to see what other people had discovered while I was testing everything and noticed a few differences on what people were saying, and my conclusions.

 

The only possible answers for these differences is that either we're playing different games or everyone here is wrong and I am your savior for pointing out your sins and leading you back to the holy road.

 

An example of our differences: Make a new party of two characters of the same race and custom type, with no skills raised and no traits. Raise one character's dexterity to one hundred. Raise the other character's sharpshooter to one hundred. Then equip them with the same type of bow and same type of arrows. By the information I've seen posted here, the character with the sharpshooter skill should out-do the other in damage. From my testing, however, the character with the one hundred dexterity out-did the other character in damage dealt to my unfortunate enemies.

 

Where is your god now?

 

Oh, and at least sharpshooter's one added die of damage per two skill levels is not as bad as blademaster's one per three skill levels ( making me wonder if everyone's love for that skill stems from the assumtion that the harder it is to obtain the prize, the better that prize is ).

[Enraged Slith]

How many trials did you run?

[Lilith]

Some statistical tests have been done, but a lot of the information we have is based on the game's documentation, which is spotty and sometimes internally inconsistent. If you have hard data, we'd be very happy to see it.

 

Note that Blades of Exile has had much more extensive tests done on it, and they seem to confirm that we are, in fact, "playing different games" in some cases. Different people's copies of that game can behave differently in bizarre but generally replicable ways, even if they have the same version number and are on the same platform.

[YeahThatGuy]

Originally Posted By: Enraged Slith

How many trials did you run?

You only need to do it once if you elimate the random factors. That's why I mentioned using weapons that do one to one damage ( always one ) per level.

[Nioca]

Originally Posted By: YeahThatGuy

Originally Posted By: Enraged Slith

How many trials did you run?

You only need to do it once if you elimate the random factors. That's why I mentioned using weapons that do one to one damage ( always one ) per level.

That's the problem, though. It's impossible to remove all the random factors. That's why multiple tests are needed. Heck, I did over 100 tests on luck before I made my findings known.

[Niemand]

That's not the only random factor, depending on your setup. Armor, for instance, would really add a lot of confusion, since it has both a random chance of taking effect and a random blocking range. Unless you were very careful you will need multiple trials.

 

I replicated your results, with the following setup:

I created a party of 2 human custom characters, named Gerald and Harald. I spent no skill points on either.

I added 100 Dexterity to Gerald and 100 Sharpshooter to Harald.

(So Gerald has 102 dexterity and 0 Sharpshooter, Harald has 2 Dexterity and 100 Sharpshooter)

I gave each a 'Weak Bow' and 10 'Weak Arrows", defined as follows:

Code:

begindefineitem 445;	import = 90;	it_name = "Bow";	it_full_name = "Weak Bow";	it_damage_per_level = 1;	it_bonus = 1;begindefineitem 446;	import = 100;	it_name = "Arrows";	it_full_name = "Weak Arrows";

 

I then conducted ten trials, each consisting of Gerald and Harald each shooting one arrow at a level 50 goblin (which was otherwise standard). The goblin was automatically healed completely after each round.

 

Results:

Every arrow hit and the damage inflicted was recorded as follows:

Code:

Gerald9897100981019997999999Harald49484848504746494747

 

Clearly Gerald inflicts more damage; the mean damage for 100 Dexterity is 98.7 with a standard deviation of 1.19, and the mean damage for 100 Sharpshooter is 47.9 with a standard deviation of 1.14. So, it seems that your result is supported.

[Enraged Slith]

Hmm... I guess my dexterity tank idea is a much more viable than I thought.

 

Also, what about 100 to bow and 100 to dexterity?

[Niemand]

Using the same setup but replacing Sharpshooter with Bows, I find that they are equivalent. So, when training an archer, train in whichever of Bows or Dexterity is cheaper, or in Sharpshooter if it's half as expensive as the next best.

[Aoslare]

Thuryl -- now that the BoE source is out, has anybody located the source of these misbehavings?

 

I know the source is rather sprawling, but the bits involving combat mechanics seemed pretty straightforward to me, even if the code wasn't particularly tight.

[Lilith]

My suspicion is that the copy of the source we got isn't the only copy that was ever used. Different copies of the same version may actually be different builds. Jeff mentioned that he had difficulty finding the complete source code for BoE and that delayed its release, so I doubt he was ever very organised with it. I don't think anyone's tested it extensively, though.

[YeahThatGuy]

Originally Posted By: Niemand

That's not the only random factor, depending on your setup. Armor, for instance, would really add a lot of confusion, since it has both a random chance of taking effect and a random blocking range. Unless you were very careful you will need multiple trials.

I also went ahead and created a bare-minimum creature that wouldn't have natural armor or items ( the default goblin has a chance of having a number of armor pieces ). Try attacking this creature and tell me if you see any change in your damage.

Code:

begindefinecreature 234;	clear;	cr_name = "Humanoid Dummy";	cr_level = 1;	cr_hp_bonus = 1000;	cr_which_sheet = 1521;	cr_default_attitude = 4;	cr_species = 1;	//cr_start_item 0 = 445;	//cr_start_item_chance 0 = 100;

Also, about armor. You said that it has "a random chance of taking effect". Going solely by your numbers ( since no one including myself has posted any ) Neither of your characters did either 102 ( Gerald ) or 52 ( Harald ) which would be, according to my findings is what they should be doing ( if the creature had no armor ). Obviously one can argue that the random numbers were biased but from my testing ( commented-out item in the above code was a 1-1 damage sword ) my character, wearing only pants, always took one point of damage less than what he should have ( they protect 1-1 ). When I took them off of him ( dirty ), he then always took the full damage. Niemand is correct.

I wanted to dump all of the specifics of each race, skill, trait, et cetera but unfortunately there are some random factors I cannot go around and I have not had the time to make a great number of trials for everything. For now, I'm settling on simply providing the effects of the species.

Human

• No effect

Slith

• Experience: 20% penalty.
• Fire Resistance: 30% added.
• Pole Damage: 2 + 1/8 level added dice.
• Pole Accuracy: ( 3 + 1/8 level ) * 5% added.

Nephil

• Experience: 10% penalty.
• Dexterity: 2 + 1/7 level added.
• Missile Damage: 2 + 1/10 level added dice.
• Missile Accuracy: ( 3 + 1/10 level ) * 5% added.

While creatures also get these bonuses, I have not checked the other species using one. Now that I think about it, that would be important for any scenario designers ( although I doubt their advantages go beyond what is stated in the scenario editor document ).

The experience penalties were easy to figure-out, as was the Nephil's bonus to dexterity since that is one of the few stat bonuses that actually show. The missile and pole bonuses were deducted from again using weapons that do 1-1 damage and no skill points placed into any skills. A human using either a bow or pole weapon will have a forty-five chance to hit against a creature at level one like my dummy creature ( base of forty, plus ten percent because of the two points in dexterity, minus five for the one point of dexterity a level one creature has ). I then used another character of a different race to determine that race's base bonuses, and then leveled them one level at a time using the High Level Party-Maker and noting when that character's accuracy or damage-dealt increased.

Also, please remember that in addition to the Nephil's bonus to damage and accuracy, their bonus to dexterity further adds to that. So a Nephil with Natural Warrior who never puts a single point into a missile-related skill could easily maintain decent damage and accuracy with a bow to keep it as a secondary weapon and can then focus on other things. At higher levels accuracy may become a slight issue with the rate that creatures gain dexterity... a level forty-two character ( built as described ) attacking an equally leveled creature would have seventy-five percent accuracy and only an average of eighty-seven damage with a blessed crossbow. Okay, maybe a few skill points would be needed.

[Niemand]

Quote:

Also, about armor. You said that it has "a random chance of taking effect"

Take a look at the documentation of the it_variety parameter for items on page 61 of the Editor Documentation. Note that each equipable item type has in its description a sentence like "When wearer is struck, has X% chance of blocking some of the damage." I am not certain, but assume that this applies equally to PCs and NPCs.

[YeahThatGuy]

Yes, you are completely correct. I've edited my post to keep others from assuming otherwise.

[MadScientist]

First of all, I really appreciate your work!

You might be interested in some tests I did quite some time ago about some skills, especially Quick Strike and Magical Efficiency (Clicky)

 

MS

[YeahThatGuy]

Yes, I am interested, especially since those effects involve elements of randomness that I cannot get around.

[MadScientist]

Unlike the stat bonus provided by Divinely Touched, the Dex bonus from being a Nephil counts toward the base Tool Use (Dex/2). Both Divinely Touched and Nephil boni are not counted as requirements for the special skills.

[Nioca]

Just found something interesting. I decided to take a look at how diseased works, and found something surprising. If your party is weak enough, then the disease actually strengthens over time, rather than weakening like most other statuses. How it decreases/increases seems to be based on some kind of random number arrangement correlated to the character's statistics. I'll post more when I know more.

[Dikiyoba]

What, like VotDT isn't bad enough already?

 

Dikiyoba.

[Callie]

Yeah, and that wand doesn't help too much either.