|
|
Rainalcar
Promising
Famous Hero
Heroji su zakon
|
posted April 08, 2009 12:52 PM |
|
|
Now I have to simplify the whole process Alex described to include it in the manual, but, one great mistery solved. How about some more of those?
-Is it possible to predict, and what is the system used, based on lower known Mage Guild levels, which spells will appear in the next unpurchased level of the Guild
-What is the system used in determining which hero will be offered in the Tavern next after buying one of two available ones, taking note of hero class non duplicability
If you have any other ideas, share.
|
|
angelito
Honorable
Undefeatable Hero
proud father of a princess
|
posted April 08, 2009 12:57 PM |
|
Edited by angelito at 12:58, 08 Apr 2009.
|
@ Xarfax
As usual, you didn't get a single point of my post.
____________
Better judged by 12 than carried by 6.
|
|
Ecoris
Promising
Supreme Hero
|
posted April 08, 2009 01:21 PM |
|
|
In sense of the analogy I view myself as one of those who claimed the Earth was round.
____________
|
|
SAG
Promising
Supreme Hero
WCL owner
|
posted April 08, 2009 04:05 PM |
|
|
Anders, how do you plan to use the knowledge of level-ups?
I think that soon Alex will post database of trees.
Will you search database for best tree for Crag in game? Will you try to remember key trees?
|
|
dimis
Responsible
Supreme Hero
Digitally signed by FoG
|
posted April 08, 2009 08:22 PM |
|
Edited by dimis at 20:09, 08 Nov 2009.
|
You figure out the title since there are many thoughts
[many thoughts, so unfortunately this is another long post]
__________________________________________________________________
Well, if you want to describe how new skills appear during level-up, there is nothing new. The answer there is 3 years old.
The only change regarding the mechanism appears on the unlikely scenario that you have two (or more) Magic Schools not at Expert level and one of them has to appear due to exception (i.e. on your left); i.e. Wisdom exception does not coincide, or if it does, then you don't have Wisdom, so Wisdom will appear on your right as Basic, or you have Wisdom not at Expert, Magic and Wisdom exceptions coincide and you have no available slot for new skill (so this time Magic appears on your right). Then the choice among the Magic School wants to be uniform (equal weights). Note here that even in some cases the old approach actually predicted uniform (e.g. on Barbarians with Air and Earth not at expert since they had equal weights), and even in cases where this was not uniform, it was still close to uniform (e.g. Barbarians Air (weight 3 ==> 60%) and Fire (weight 2 ==> 40%)). So you realize that this distinction can hardly be observed in practice, because by the very nature of the mechanism it is unlikely to have at least two Magic Skills not at Expert level simultaneously in the first place! And again, the only impact of this is on the "pace" you expect to upgrade Magic Scholls; not on the likelihood of getting new skills.
The other thing is the randomness. Does the guy who plays the game need to know how this thing is defined? My personal opinion is no; this makes it interesting. And for me that's the end of story regarding the impact of the mechanism in gaming. You already know what you need to know in order to predict new skills because eventually you are going to have skills as good as those that you are offered; FLASHNEWS!
Of course, handling randomness has its own interest on giving more accurate results for specific strategies when selecting skills. In other words my belief is that players need to know only about the big picture and hope on a good sequence of "random numbers".
* In terms of mechanism of new skills the model was correct.
* In terms of building up tables like AR and AL, then yes, you need to know how randomness is defined in the game to make the tables accurate to the last decimal digit. But the reason here is very subtle and difficult to understand. The problem is that these 3 equations are "biasing" the result of the general mechanism because clearly they are not general enough. Now, the only true contribution by Xarfax in this thread (Magic @ 4th level is in Heroes Stats and Skills Chances) is that randomness is biased a lot, since we have few sequences of random numbers that express the randomness. Of course this is true but it doesn't give anything; you only have the statement and a period after that. Go figure ...
The first time that we made a prediction (and actually had a solid justification) about a single number on the amount of different sequences that express this sort of randomness was in July 2006, page 10, first paragraph here; compare that number with 255 and tell me about it. The other stuff there are precaution measures for that number and in a sense were supposed to motivate people and do some tests (you don't want me to mention how many additional tests like those I got until today, do you?). One more thing that was not presented in thread and can be verified by our Birthday-Problem expert (explanation of solution), is based only on the first 18 examples of that test, which predicts an expectation of 202-226. Other than that, only Ecoris made a test to figure out how randomness was handled in this game in his bonus post in page 7 and his remarks are apparent in the description by Alex (And really it is a pity that his other post {to my opinion one of the best in the library} was not awarded a QP). But then again randomness was not our primary question throughout the thread.
I guess not everything is a one-liner. You see people have the misconception that they know something if they can teach another person on how to do it (with one-liners, slogans, whatever ...). The truth however is that you actually know something if you can teach a computer how to do it. And unfortunately this causes posts to be big from time to time. You have to deal with many crazy details that people don't really care because they are very unlikely or never thought about them (an example that revived the whole thread was given by maretti who came up with a very interesting question 3 pages ago). Moreover, the problem of understanding stuff depends on the comprehension one has on reading. So yes, I don't expect everybody to understand this post either, but I have to give my viewpoint since some of my earlier bonus posts are attacked.
Finally, it is also my belief that that specific question by maretti and my explanation cleared up some of the thoughts Alex had until that time when he was interpreting code in assembly. In any case though, it is good that we reached an end and I am very grateful to this guy because I would have never spent time to start reversing the code; it is a matter of taste and experience in the approach.
Now, as far as the database is concerned, you guys shouldn't rush it. What do you really want to know? What kind of questions do you want to pose? Because the answers, depend on the questions (not the other way around). And for some of them, we don't even have to generate a database in this post-algorithm era. What I mean here, is that having a huge database with all skill trees for all heroes might be redundant and some people out there might not even have the space to store them. On the other hand, it would be cheap though to generate these trees on-demand for some of the questions (actually you only need the sequence of "random numbers", since that is precisely what characterizes and distinguishes one tree from another - and this seems to be a space-efficient representation for the database, so the space problem is eliminated). So, first think about the questions you want to pose. The answers don't come out of the blue. Magic died in the Renaissance.
EDIT (Nov 8, 2009): Fixed a broken link; document with testing Crag Hack on ANSA.
____________
The empty set
|
|
dimis
Responsible
Supreme Hero
Digitally signed by FoG
|
posted April 08, 2009 09:20 PM |
|
|
Quote: How about some more of those?
Interesting, but in another thread.
Quote: -Is it possible to predict, and what is the system used, based on lower known Mage Guild levels, which spells will appear in the next unpurchased level of the Guild
This has a separate thread on its own.
Quote: -What is the system used in determining which hero will be offered in the Tavern next after buying one of two available ones, taking note of hero class non duplicability
Interesting from a programming point of view but kind of spoiler for the real game. In any case, this belongs in another thread too.
____________
The empty set
|
|
Binabik
Responsible
Legendary Hero
|
posted April 08, 2009 09:38 PM |
|
|
This thread is very long and technical. Can someone please summarize (***in layman's language***) where we are with this? I don't think there are many people would really want to wade through all of this. And could a brief summary of sources and methodology be included?
It would be very helpful for all of us normal peons who are too lazy to "read" through all this. (read = study with a magnifying glass with several mathematic and programming books open to the side)
thanks
|
|
dimis
Responsible
Supreme Hero
Digitally signed by FoG
|
posted April 08, 2009 09:42 PM |
|
Edited by dimis at 21:50, 08 Apr 2009.
|
Yes, Binabik. That has already been done. Read this (the first 8 pages; 2 of which have the weights). Then all you need to add as an "exception" is what I describe in "change 1" in my post in the previous page. End of story. Besides, that was the idea of my post after Alex finished his job in the previous page.
____________
The empty set
|
|
Rainalcar
Promising
Famous Hero
Heroji su zakon
|
posted April 08, 2009 09:49 PM |
|
|
I agree with all Dimis. But, finding out the existence of definitive skill trees via the effort of you people isn't all revealing per se. It's like when I'm playing the game and simply know what will the AI do next in the fight. I know because I've seen it hundreds of times (well, long ago), and although I never really analyzed it, I would rarely err on that part, as I would rarely err which fight to fight.
This in mind, if it's possible to (partially) predict which hero will appear in Tavern next would simply fall into the same category. In other words, if top playing of Heroes requires this stuff, it's a pitty, but is inevitable with top playing in regard.
Btw, the link you posted doesn't really explain my question. It contains the spell probability table by Ecoris, which is already a part of Tribute to Strategists, but my question was different.
|
|
dimis
Responsible
Supreme Hero
Digitally signed by FoG
|
posted April 08, 2009 09:51 PM |
|
Edited by dimis at 22:16, 08 Apr 2009.
|
It wasn't meant to explain your question Rainalcar. I am just saying that your question belongs to that thread; not here - here we talk about skills.
____________
The empty set
|
|
Binabik
Responsible
Legendary Hero
|
posted April 08, 2009 09:53 PM |
|
|
Lol, you call 8 pages a summary? I was thinking of a few paragraphs.
OK, I'll take a look at it.
____________
|
|
dimis
Responsible
Supreme Hero
Digitally signed by FoG
|
posted April 08, 2009 09:55 PM |
|
Edited by dimis at 22:10, 08 Apr 2009.
|
It is 6 pages Binabik if you exclude the pages with the weights and has 2 big examples to stress the important stuff and give all the fundamental information to a guy who knows nothing before they read it. It also has all the sources you need (by the time it was written).
edit: If you are still not satisfied by the amount of information on a first read skip Remarks 2.3, 2.4, 2.5, and 2.11 and Cor. 2.6. You can go through these on a second read.
____________
The empty set
|
|
dimis
Responsible
Supreme Hero
Digitally signed by FoG
|
posted April 08, 2009 10:09 PM |
|
Edited by dimis at 22:27, 08 Apr 2009.
|
Quote: But, finding out the existence of definitive skill trees via the effort of you people isn't all revealing per se. It's like when I'm playing the game and simply know what will the AI do next in the fight.
Here is a fine line now that we cross. Because right now, there is no uncertainty on the outcome. I mean, even in fights you more or less predict what the AI does, but even rarely you make a mistake on your judgment (i.e. why did AI chose to wait?) or even if you predict right, are you 100% sure that you can predict what the AI does in every fight? That's the difference. Of course the outcome of the battle is irrelevant; you are going to win anyway if you are experienced.
Of course I am with you, that eventually such a "calculator" should be made available, since even if I don't do it, others will. It is very simple to implement it once you know the mechanism that causes randomness in this game (the 3 equations given by Alex). So, yes, I will definitely do it if not anybody else does this. But as far as the usage of such a thing, I don't know. I mean, clearly it requires lots of effort to figure out the rules of such an "oracle" (what all of us have done in this thread as well as Alex on his own), but then randomness (for this problem) goes bye-bye. And randomness is a crucial ingredient of the game. That's why I am skeptical on the use of such an "oracle", "calculator", call it whatever you want.
____________
The empty set
|
|
dimis
Responsible
Supreme Hero
Digitally signed by FoG
|
posted April 09, 2009 08:16 AM |
|
|
In the end Rainalcar I think you have convinced me. And the reason is that during the first level ups there is some uncertainty since you don't know which skill-tree (number-sequence) you follow. So you have to base your selections on your knowledge of the exceptions, as well as the probabilities of the various skills among the possible skill-trees that verify your level-ups so far. So in that sense it is sort of random. You don't know the tree right from the start. Of course after some point ...
____________
The empty set
|
|
Rainalcar
Promising
Famous Hero
Heroji su zakon
|
posted April 09, 2009 12:36 PM |
|
|
As stated, I am with you, and it's a pitty that things are determined as much as they are. But, as you said, knowledge will not wait, and the best way is to except it, then to ignore it and hope everyone else does too. Not that it concerns me - I play heroes for fun only these days and will likely not resort to old ways ever again, but if you want to become the best player possible, you must know this stuff.
I will state the guild idea in the other thread, so that it doesn't get forgotten.
edit: grammar errors
|
|
dimis
Responsible
Supreme Hero
Digitally signed by FoG
|
posted April 09, 2009 10:26 PM |
|
Edited by dimis at 22:30, 09 Apr 2009.
|
AlexSpl,
Now that you have the trees for Crag Hack in some form, if it is not a big burden in your schedule, could you post the respective values for AR or/and AL entries? I am really interested in seeing how much these change. (Yes, I am kind of lazy to write even a single line of code for this problem right now).
____________
The empty set
|
|
dimis
Responsible
Supreme Hero
Digitally signed by FoG
|
posted April 13, 2009 07:34 PM |
|
Edited by dimis at 19:34, 13 Apr 2009.
|
revision
Quote: And again, the only impact of this is on the "pace" you expect to upgrade Magic Scholls; not on the likelihood of getting new skills.
Just to criticize my writing. This is not entirely correct, because the pace on which you upgrade magic schools has an impact on the "expected" case in which Magic exceptions will occur again; although this is a small distortion on the output distribution.
However, it is true that both tables AR and AL don't change theoretically for a simple reason. The above problem can happen if "2 or more Magic schools are not at Expert level simultaneously and ...". This assertion already fails in the AL case, because after level 4 you have at most one skill not at expert level. AR on the other hand is computed practically when you reach level 7 or 8 (level 8 for heroes that start with just one skill). Now, the sooner you can get 2 Magic skills with a mighty hero is at level 3, and you can have an exception no sooner than level 7. Still, even then, you don't pick the offer on your left (since you are following the "right path" - AR), so, this change has no impact on the sum that appears in the denominator. Same is true for all the values that are computed on ALTP. The justification is the same as in AL. However, SPOU values computed are slightly incorrect at the moment.
I thought it would be better if I wrote that down and not leave an incorrect statement in the air.
____________
The empty set
|
|
maretti
Responsible
Supreme Hero
|
posted April 14, 2009 12:21 AM |
|
|
Quote: Anders, how do you plan to use the knowledge of level-ups?
I think that soon Alex will post database of trees.
Will you search database for best tree for Crag in game? Will you try to remember key trees?
Im not planning on using it in any way. It would make the game much less interesting if you at some point know what skills you will be offered. Ofcourse there will be an advantage, but it wont be that big. In most cases you will probably pick the same way as you would without this knowledge.
____________
Crag rules, Orrin and Ivor suck
|
|
Xarfax111
Bad-mannered
Supreme Hero
The last hero standing
|
posted April 14, 2009 07:37 PM |
|
|
Quote: @ Xarfax
As usual, you didn't get a single point of my post.
Well of course i "usual"ly only "get singel "-Q"points" by you.
j/k
____________
|
|
Rainalcar
Promising
Famous Hero
Heroji su zakon
|
posted September 10, 2009 12:01 AM |
|
|
Damn, I totally forgot to include this stuff in the manual . Hopefully I'll catch some time soon, still sommer here
____________
|
|
|
|