Imoen learns 1/5 spells I try to have her learn

triclops41

Is this a bug? I am not exaggerating when I say she fails to learn spells of her current level or lower 4/5 times. I thought her intelligence was high enough for her to learn most spells. I have to reload several times to get her to learn a 1st level spell and she is a level 6 mage!

Madhax

Try quaffing a Potion of Genius, they're plentiful. I doubt it's a bug, you're either experiencing a run of bad luck or confirmation bias.

[Deleted User]

The user and all related content has been deleted.

Freche

If you reload anyway you can either lower the difficulty while reading scrolls or mod the success chance to 100%.

triclops41

Madhax said:

Try quaffing a Potion of Genius, they're plentiful. I doubt it's a bug, you're either experiencing a run of bad luck or confirmation bias.

I am highly doubtful its confirmation bias. I think I will get some potions of genius.

ZanathKariashi

Hence why int above 13 is pointless, except to allow dualing. At least until they fix the bug stopping Max-castable from working. (Still wouldn't effect bards, unless they fix their spell progression to allow up to 8th, as normal, since 13 int (bard minimum) allows up to 6th level anyway).

mlnevese

We had some discussions about this in the forum in the past. One of the forum members made a scripted test with 3000 attempts to learn spells from scrolls that showed that the curve was compatible with the expected results. I'll try to dig out those threads, although I'm not that good with thread necromancy

Awong124

mlnevese said:

We had some discussions about this in the forum in the past. One of the forum members made a scripted test with 3000 attempts to learn spells from scrolls that showed that the curve was compatible with the expected results. I'll try to dig out those threads, although I'm not that good with thread necromancy

Here you go:
http://forum.baldursgate.com/discussion/12514/chance-to-learn-spell-is-bs/p1

Dexter

I agree with @triclops41 to the point that I pile up 10 - 15 scrolls and write them down after drinking 2 potions of genius. There is a reason for the red potion after all

Battlehamster

Also, I can't confirm this for sure, but it seems to me from personal experience if
your caster level < Spell level...
spells are harder to write.

TJ_Hooker

I vaguely remember reading somewhere that if the spell you are trying to learn is too high a level for you to currently cast, you receive a penalty when trying to scribe it (not sure if this is what @Battlehamster is referring to). Also, I think a specialist mage receives a bonus when scribing his favoured spell school and a penalty for scribing all others. Not sure if either of these things are actually implemented and functioning though.

triclops41

I'm wondering when the pass/fail check is rolled on learning the spell. It just seems weird that an Imoen level 5 or 6 mage would still fail several times in a row on several occasions to learn level 1 and 2 spells.

Jarrakul

The going theory from the last thread was that reloading the game reloads a bad random number seed, or something. I'm not really sure how the details work, but I think we had to control for reloading before the failure rate was within expected error of what it should be. Basically, reloading constantly is biasing your numbers towards failure. Most likely, Imoen succeeds at about the correct rate on the first try, but reloads after failed attempts are much more likely to lead to failure. Without knowing now the RNG in BG works, I can't really give anything more than speculation as to why that might be.

moody_mage

ZanathKariashi said:

Hence why int above 13 is pointless, except to allow dualing. At least until they fix the bug stopping Max-castable from working.

Dualing requires a base score of 15 for the initial class and 17 for the subsequent class.

triclops41

Jarrakul said:

The going theory from the last thread was that reloading the game reloads a bad random number seed, or something. I'm not really sure how the details work, but I think we had to control for reloading before the failure rate was within expected error of what it should be. Basically, reloading constantly is biasing your numbers towards failure. Most likely, Imoen succeeds at about the correct rate on the first try, but reloads after failed attempts are much more likely to lead to failure. Without knowing now the RNG in BG works, I can't really give anything more than speculation as to why that might be.

That makes sense, and would explain the behavior I'm getting.

lamaros

triclops41 said:

Madhax said:

Try quaffing a Potion of Genius, they're plentiful. I doubt it's a bug, you're either experiencing a run of bad luck or confirmation bias.

I am highly doubtful its confirmation bias. I think I will get some potions of genius.

If you can reload and get a different result then it is confirmation bias, as no one else has evidence of such a fundamental bug at this stage in the game's life.

Lord_Tansheron

I faintly remember such a discussion about a decade ago... I do think the "result" people arrived at is that the whole RNG-bias mumbo-jumbo is essentially conjecture and anecdotal evidence.

Don't nail me down on that, though.

Jarrakul

@lamaros, how would reloading and getting a different result create/indicate confirmation bias? If the result were truly random, and loading the game had no impact on the odds (which I posit is not the case, but that's neither here nor there), we would not expect reloading to create anything more than a repetition of the same random event. While it's interesting to note that the reloads only occur on failures, they shouldn't impact the success/failure ratio if they have the same probabilities as the initial event. Even if they did, confirmation bias is about people seeing what they expect to see, or what is most emotionally salient, as being more frequent than actual evidence would indicate, due to a bias in the way people remember events. I do apologize if I'm missing some important point of yours, but I can't for the life of me see what that has to do with reloading.

Also, I think the fact that this isn't the first thread on this phenomenon means that other people have (anecdotal) evidence that this bug does exist. See my previous post for what I took as the general consensus conclusion last time this issue came up.

Sorry if it sounds like I'm jumping down your throat. I don't mean to, but probability, statistics, and bias are kind of passions of mine, and I can get pretty pedantic about things I like.

lamaros

If you're reloading and getting different results then it's not a seed issue, it's actually random. Given that there is no real evidence to indicate there are any issues with the randomness generally it would point to confirmation bias.

mlnevese

As I mentioned earlier a test was made with 3000 rolls and the curve fits in what would be expected given the chance to learn a spell.

sarevok57

when I learn spells, I just put the difficulty down to minimum, its way faster than reloading, and the fact that mages should even be able to fail is just whack, already being limited on the amount of spells they are allowed to know based on INT is brutal enough, they don't also need a failure option as well ( or at least make is so that if you INT is 17+ they cant fail and for every point below 17, make a 10% penalty or something)

triclops41

lamaros said:

triclops41 said:

Madhax said:

Try quaffing a Potion of Genius, they're plentiful. I doubt it's a bug, you're either experiencing a run of bad luck or confirmation bias.

I am highly doubtful its confirmation bias. I think I will get some potions of genius.

If you can reload and get a different result then it is confirmation bias, as no one else has evidence of such a fundamental bug at this stage in the game's life.

I have a degree in psychology and am quite familiar with confirmation bias. My first question was addressing that. If I reload after the RNG chooses, reloading wouldn't change whether it is a success/fail state. That's why I'm trying to understand the conditions of the RNG.
I don't have the time to test it, but my number of consecutive failures with imoen when she should be at .75 is unusually high.
Perhaps I'm just getting lucky, but resting before i try to memorize the spells seems to bring me more appropriate results.

I've looked at those tests, and while helpful, are hardly comprehensive. Is feel more satisfied with a save/reload test.

Jarrakul

@triclops41, hurray for psychology degrees! I have one also. Working on my masters now.

@lamaros, clearly there's more going on than the game already deciding on the outcome. In fact, I can pretty much guarantee that the seed isn't identical across reloads. That simply does not fit the data. Instead, the seed is probably based on either real-world time or playtime. But depending on what BG's RNG uses as a source, it's possible for it to be highly non-random in ranges with similar seeds, and so if the seed is based on playtime, it's very possible that reloading could get a similar seed which would bias the results do to inherent biases of the RNG (remember, no RNG is ever actually random unless it's measuring subatomic particles; your computer just takes a number from a list at a certain point depending on the seed; the list is hopefully highly unpredictable, but this is not always the case). But all that is just theory. It's a neat explanation, but it doesn't address the issue of whether reloading is actually having an effect. From my post on the previous thread:

Jarrakul said:

That said, I'm surprised no one's done any actual statistical analysis on the numbers people have been throwing out. Without such analysis, even the most convincing numbers are entirely meaningless, because we don't know how likely those results actually are. For example, Awong, your data on the first page. I did a couple quick (1-tailed) t-tests, and your first test [with reloading] has a .23% chance (p = .0023) to occur if the chance of learning a spell is actually 75%. Since scientific convention generally uses 5% (p = .05) as the realistic cutoff point between reasonably likely and highly unlikely, we can tentatively conclude that your first set had less than a 75% chance of success. Your second set, the one without reloading, tells a different story. It has a 10.73% chance (p = .1073) of occurring randomly, which is above our 5% threshold. Therefore, we conclude that your second result is within the realm of reasonable likelihood for something with a 75% success rate. From this data, I surmise that your theory about getting a bad random number seed on the first test was correct, and that there is no reason to believe that the chance isn't 75%.

So I'm not just wildly speculating. I actually ran the stats on the data, with and without reloading. If people want to provide me with more trials, I'd be happy to run the stats on those also. Replicate my own work, if you will. But right now, the theory that best fits the data is that reloading does something weird to the random number seed, and messes up the observed frequencies.

lamaros

If you're re-loading so you don't fail, just lower the difficulty level. I really don't get why this is a thing unless it's an issue for people actually playing by the rules.

Jarrakul

It's a thing because it's weird, and unexpected. For me, at least, that's enough to make it interesting. It literally doesn't effect me, because I use a 100% learn spell mod, but it doesn't matter because the number are behaving in an interesting way and I want to know why. Barring that, I want to codify what we know about the patterns of weirdness, at the very least.

lamaros

I can understand the desire to solve a little mystery.

triclops41

lamaros said:

If you're re-loading so you don't fail, just lower the difficulty level. I really don't get why this is a thing unless it's an issue for people actually playing by the rules.

The difference between expecting a .75 success rate vs what I was getting was puzzling. It took me a while to stop reloading and just change the difficulty because I spent a looooong time thinking, "It is just a fluke, keep reloading and it will be normal the next time". Because the other systems in the game seem to work within expected ranges, I assumed that I was just having bad luck, and that I would just power through it.

PugPug

Jarrakul said:

The going theory from the last thread was that reloading the game reloads a bad random number seed, or something. I'm not really sure how the details work, but I think we had to control for reloading before the failure rate was within expected error of what it should be. Basically, reloading constantly is biasing your numbers towards failure. Most likely, Imoen succeeds at about the correct rate on the first try, but reloads after failed attempts are much more likely to lead to failure. Without knowing now the RNG in BG works, I can't really give anything more than speculation as to why that might be.

I know that reloading to try to save Melicamp does not work, presumably because doing so does not generate a new seed.

Awong124

PugPug said:

Jarrakul said:

The going theory from the last thread was that reloading the game reloads a bad random number seed, or something. I'm not really sure how the details work, but I think we had to control for reloading before the failure rate was within expected error of what it should be. Basically, reloading constantly is biasing your numbers towards failure. Most likely, Imoen succeeds at about the correct rate on the first try, but reloads after failed attempts are much more likely to lead to failure. Without knowing now the RNG in BG works, I can't really give anything more than speculation as to why that might be.

I know that reloading to try to save Melicamp does not work, presumably because doing so does not generate a new seed.

Yes it does. I've definitely done it.

ZanathKariashi

Yeah, Reloading for Melicamp works fine. But it IS a 50/50 chance per attempt (It literally just rolls d100 and 50 or less he dies, 51-100 he lives. Spell learning is the same way, d100, your learn chance or less succeeds, anything else fails).

kamuizin

Probally someone already told this and i didn't take the trouble of read the previous posts, but just reduce the difficult to the lower, then you have 100% chance of learning spells.