In the original game I dualled at Lv2 and it was really great. You won't spend half the game waiting to regain your abilities. Druid class gives you some cool spelss, that are buffed by your slightly better Thac0 and weapon proficiency.
Hence my advice against what most people say: dual pretty quickly to have a druid with some additional kick.
I see a lot of recommendations for parties to have a druid due to story interactions. I haven't played IWD in years and can't remember what exactly. I am going the Fighter > Druid route. Are there (m)any interactions I will miss out on while I'm still just a Fighter? I was thinking to dual at level 7, maybe level 9.
Random is as random does... it may just be that simple.
Considering that 97% of spells should be learnable, and I've done it on two different occassions at different character levels, I don't think it is random. Repeated it over 40 times and it bears out - it is more like 75%, not 97%.
... 40 iterations is simply not statistically significant.
Random is as random does... it may just be that simple.
Considering that 97% of spells should be learnable, and I've done it on two different occassions at different character levels, I don't think it is random. Repeated it over 40 times and it bears out - it is more like 75%, not 97%.
... 40 iterations is simply not statistically significant.
More importantly, don't specialists have a penalty to learning non-school spells or something? Also, 40 is enough that you should be pretty damn close to the right rate of success/failure. Perfect? No, but you'll be close.
I learned those spells in two separate occasions, and had to reload a minimum of 10 times per go to learn them all. Remember, my mage is not a specialist mage, so he has no penalty. He should have failed to learn one or two of those spells, not nearly half of them.
I learned those spells in two separate occasions, and had to reload a minimum of 10 times per go to learn them all. Remember, my mage is not a specialist mage, so he has no penalty. He should have failed to learn one or two of those spells, not nearly half of them.
There has been much discussion about this, most people are pretty confident its not working right, but from whst I've read of large samples, the ratio holds up. Mostly I would just reload if I feel cheesed by the game. I wouldnt tolerate getting robbed of a really rare scroll unless I was playing a very strict game.
Random is as random does... it may just be that simple.
Considering that 97% of spells should be learnable, and I've done it on two different occassions at different character levels, I don't think it is random. Repeated it over 40 times and it bears out - it is more like 75%, not 97%.
... 40 iterations is simply not statistically significant.
More importantly, don't specialists have a penalty to learning non-school spells or something? Also, 40 is enough that you should be pretty damn close to the right rate of success/failure. Perfect? No, but you'll be close.
Correct: Specialists do have a non-school scribing penalty; however, he claims to be a general mage (dual classed) so that would not apply.
No it is not. First, because it was not a straight up 40 and second he admits to reloading throughout the multiple scribing sessions. Further, confirmation bias runs rampant in the RNG discussions. We have not seen any actual attempt to formally test the RNG; mostly it is emotionally charged anecdotal evidence along the lines of presented here in this thread: I have an 85% chance to scribe and I lost two scrolls in a row! I reloaded and I lost another one! The RNG is clearly not working because this would NEVER happen!
In addition we are also not seeing people coming forward and saying that I scribed all my scrolls at 9 INT (35%) without and failures... People tend not to complain about randomness that favours them.
It would be very easy to setup a proper crowd sourced RNG test, but we simply do not see that happening when it is so much easier to merely whine about the evil RNG. Sad fact of life
If a RNG is worth having, previous outcomes are irrelevant, nor should loading. 40 attempts will almost certainly get you pretty close to yhe correct ratio.
Note, iirc, flubbing 3 attempts in a row at 85% success is gonna happen ~0.34% of the time, as in, REALLY unlikely to run into more than once in a while. I have done some basic statistics, and I would say 40 is a reasonable test, as there aren't variables. It isn't like a medical study, where things like age, weight, genetic background, etc can and do affect the results, thus requiring very large samples.
There have been posts made of people running extremely large samples, similar to the auto-roller. I can't verify these attempts, but they reported the RNG was not out of line.
Yeah, as DreadKhan says, reloading should not matter. In fact, reloading helps with our tests, since some scrolls would have to be learned multiple times by reloading until they stick. I would say that it took me 50-60 scrolls reads to learn those 40 scrolls. So, actually the number of events is more than 40.
I'm not sure exactly how many scrolls I learned, but put it this way. I emptied my inventory except for a scroll case. Then I unpacked that scroll case into my inventory until no more scrolls could fit. Then I learned every scroll in my inventory, and unpacked from the scroll case. After that, I unpacked from the scroll case again, and repeated the process.
Then, a few levels later, I did exactly the same unpacking process another two times.
So two full inventories worth of scrolls were learned, and about 1 in 4 scrolls had to be learned more than once because of failures.
If a RNG is worth having, previous outcomes are irrelevant, nor should loading. 40 attempts will almost certainly get you pretty close to yhe correct ratio.
Note, iirc, flubbing 3 attempts in a row at 85% success is gonna happen ~0.34% of the time, as in, REALLY unlikely to run into more than once in a while. I have done some basic statistics, and I would say 40 is a reasonable test, as there aren't variables. It isn't like a medical study, where things like age, weight, genetic background, etc can and do affect the results, thus requiring very large samples.
There have been posts made of people running extremely large samples, similar to the auto-roller. I can't verify these attempts, but they reported the RNG was not out of line.
REALLY unlikely is the definition of random event. However, there are other factors to consider: Class, Race, INT, scroll/spell, level, place/location, game - those are all variables that would need to be accounted for, among MANY others!
The reloading (as he admits in his post) impact the results not by influencing the RNG, but by confusing the actual number of successes & failures. He was not tracking the numbers closely and has resorted to whining about something because it did not go his way. Classic conformation bias in action: "I think" "was about 75%" "about 3 out of 4" - no empirical data whatsoever.
That someone apparently tested it and found the RNG to be function within expectations pretty much puts the issue to rest. Obviously a properly conducted in game test would be preferable.
Yeah, as DreadKhan says, reloading should not matter. In fact, reloading helps with our tests, since some scrolls would have to be learned multiple times by reloading until they stick. I would say that it took me 50-60 scrolls reads to learn those 40 scrolls. So, actually the number of events is more than 40.
Here we again have nothing empirical upon which to base your claim. 'I would say' '50-60'
I'm not sure exactly how many scrolls I learned, but put it this way. I emptied my inventory except for a scroll case. Then I unpacked that scroll case into my inventory until no more scrolls could fit. Then I learned every scroll in my inventory, and unpacked from the scroll case. After that, I unpacked from the scroll case again, and repeated the process.
Then, a few levels later, I did exactly the same unpacking process another two times. So two full inventories worth of scrolls were learned, and about 1 in 4 scrolls had to be learned more than once because of failures.
Uncertain number of attempts and uncertain number of failures... 'about'
RNGs can be streaky - flip a coin and see how many heads you can get in a row - that does not mean that they are broken, merely that they are random. Ultimately the RNG _may_ have issues - but you are from proving that it actually does.
No matter how many people hack the games, in some instances watching the assembly code during run time and are able to explain without a doubt how the RNG works there's always going to be people claiming the RNG is skewed (usually against their favor)
Reloading increases the successes and failures. Put it this way - reloading would not occur if there were no failures. If there was only one failure, it would only occur once. I remember reloading at least 10 times per learning session.
I would say 50-60 events is statistically significant for an event like this. Since it is a binary outcome, it can be compared to the flipping of a coin. What are the chances of flipping a coin 60 times and having tails come up on 45 of those items? Pretty low, right? You expect a 50/50 distribution and instead you get a 75/25 distribution. It is the same with this - I expected 95/5 and I got 75/25.
It sounds like I am not the only one experiencing issues. I'll see if I can record a video tonight.
Reloading increases the successes and failures. Put it this way - reloading would not occur if there were no failures. If there was only one failure, it would only occur once. I remember reloading at least 10 times per learning session.
Ignoring that reloading occurs for any number of reasons, all reloading proves is that reloading occurred. Your memory is not an empirical source and "at least 10 times per session" is not even in the realm of the scientific precision necessary for the baseless statements you are making.
I would say 50-60 events is statistically significant for an event like this. Since it is a binary outcome, it can be compared to the flipping of a coin. What are the chances of flipping a coin 60 times and having tails come up on 45 of those items? Pretty low, right? You expect a 50/50 distribution and instead you get a 75/25 distribution. It is the same with this - I expected 95/5 and I got 75/25.
As this thread proves, you can say whatever you like, but the fact of the matter is that you are wrong: - ~60 odd events is not significant (and let us not forget that this number keeps changing) - scribing may not be a binary outcome - coins are not binary
Now we get into why you are mistaken. Yes it is unlikely that 45 tails would occur in 60 flips, but it is not so unlikely to be impossible; however, on so few flips 60 tails would also be within a possible probability. Distribution does tend towards 50/50 over a _LARGE_ sample size: 60 is _NOT_ large. A 95% chance of success does not mean you only fail one scribe every twenty. If you had a single scroll and had failed to successfully scribe it, would you be here complaining that your distribution was 0/100%? Certainly seems like it... Ultimately your expectations were fundamentally flawed.
It sounds like I am not the only one experiencing issues.
No, it merely sounds like you do not understand statistics & probability. In the end you may yet turn out to be correct, but it will not be in any way related to your self proclaimed 'proof'. Luck is like that after all.
We must apologize to the OP for participating in this ridiculous tangent and will cease continuing to do so.
We stand by our original claim that Fighter9 is the optimal level to dual to Druid, although Ftr7 is an acceptable compromise; however, a Ftr/Dru multi is worth considering as well.
Comments
Hence my advice against what most people say: dual pretty quickly to have a druid with some additional kick.
No it is not. First, because it was not a straight up 40 and second he admits to reloading throughout the multiple scribing sessions. Further, confirmation bias runs rampant in the RNG discussions. We have not seen any actual attempt to formally test the RNG; mostly it is emotionally charged anecdotal evidence along the lines of presented here in this thread: I have an 85% chance to scribe and I lost two scrolls in a row! I reloaded and I lost another one! The RNG is clearly not working because this would NEVER happen!
In addition we are also not seeing people coming forward and saying that I scribed all my scrolls at 9 INT (35%) without and failures... People tend not to complain about randomness that favours them.
It would be very easy to setup a proper crowd sourced RNG test, but we simply do not see that happening when it is so much easier to merely whine about the evil RNG. Sad fact of life
Note, iirc, flubbing 3 attempts in a row at 85% success is gonna happen ~0.34% of the time, as in, REALLY unlikely to run into more than once in a while. I have done some basic statistics, and I would say 40 is a reasonable test, as there aren't variables. It isn't like a medical study, where things like age, weight, genetic background, etc can and do affect the results, thus requiring very large samples.
There have been posts made of people running extremely large samples, similar to the auto-roller. I can't verify these attempts, but they reported the RNG was not out of line.
I'm not sure exactly how many scrolls I learned, but put it this way. I emptied my inventory except for a scroll case. Then I unpacked that scroll case into my inventory until no more scrolls could fit. Then I learned every scroll in my inventory, and unpacked from the scroll case. After that, I unpacked from the scroll case again, and repeated the process.
Then, a few levels later, I did exactly the same unpacking process another two times.
So two full inventories worth of scrolls were learned, and about 1 in 4 scrolls had to be learned more than once because of failures.
The reloading (as he admits in his post) impact the results not by influencing the RNG, but by confusing the actual number of successes & failures. He was not tracking the numbers closely and has resorted to whining about something because it did not go his way. Classic conformation bias in action: "I think" "was about 75%" "about 3 out of 4" - no empirical data whatsoever.
That someone apparently tested it and found the RNG to be function within expectations pretty much puts the issue to rest. Obviously a properly conducted in game test would be preferable. Here we again have nothing empirical upon which to base your claim.
'I would say'
'50-60' 'not sure exactly how many' Uncertain number of attempts and uncertain number of failures...
'about'
RNGs can be streaky - flip a coin and see how many heads you can get in a row - that does not mean that they are broken, merely that they are random. Ultimately the RNG _may_ have issues - but you are from proving that it actually does.
No matter how many people hack the games, in some instances watching the assembly code during run time and are able to explain without a doubt how the RNG works there's always going to be people claiming the RNG is skewed (usually against their favor)
I would say 50-60 events is statistically significant for an event like this. Since it is a binary outcome, it can be compared to the flipping of a coin. What are the chances of flipping a coin 60 times and having tails come up on 45 of those items? Pretty low, right? You expect a 50/50 distribution and instead you get a 75/25 distribution. It is the same with this - I expected 95/5 and I got 75/25.
It sounds like I am not the only one experiencing issues. I'll see if I can record a video tonight.
Ignoring that reloading occurs for any number of reasons, all reloading proves is that reloading occurred. Your memory is not an empirical source and "at least 10 times per session" is not even in the realm of the scientific precision necessary for the baseless statements you are making. As this thread proves, you can say whatever you like, but the fact of the matter is that you are wrong:
- ~60 odd events is not significant (and let us not forget that this number keeps changing)
- scribing may not be a binary outcome
- coins are not binary
Now we get into why you are mistaken. Yes it is unlikely that 45 tails would occur in 60 flips, but it is not so unlikely to be impossible; however, on so few flips 60 tails would also be within a possible probability. Distribution does tend towards 50/50 over a _LARGE_ sample size: 60 is _NOT_ large. A 95% chance of success does not mean you only fail one scribe every twenty. If you had a single scroll and had failed to successfully scribe it, would you be here complaining that your distribution was 0/100%? Certainly seems like it... Ultimately your expectations were fundamentally flawed. No, it merely sounds like you do not understand statistics & probability. In the end you may yet turn out to be correct, but it will not be in any way related to your self proclaimed 'proof'. Luck is like that after all.
We must apologize to the OP for participating in this ridiculous tangent and will cease continuing to do so.
We stand by our original claim that Fighter9 is the optimal level to dual to Druid, although Ftr7 is an acceptable compromise; however, a Ftr/Dru multi is worth considering as well.