Siankan
Posts: 1048
4/27/2017
|
Optimatum wrote:
Your evidence is also incomplete. For example, when Mood cards were reenabled, I too drew one a couple days later. But that isn't evidence of weighted draw RNG because the pattern hasn't changed. My deck is pretty trimmed and I regularly draw one or two mood cards a week. What you mentioned isn't clear evidence because you've only given your impressions of card appearances at those times, but not during the earlier "sparse" period or the period after the frequency is no longer "behind".
It is also to be noted that Mood cards, particularly, are affected by deck size. When I was a new player, it was not uncommon for me to draw three or four a week, largely on the back of my restricted deck size. Now that I am much further into the game and have unlocked far more cards in my deck, I am lucky to get a tithe of that. Since most people's available deck pool gets larger over time (unless they are deliberately trimming, and even then there's only so much you can do), it is logical to assume that for the majority of players, mood cards will get rarer over time.
Also, when Mood cards reappeared, they were a new and exciting event. You are much more likely to remember them, and the memories have greater impact on your perception. Six months later, even if you've been drawing them with the same regularity, they are not likely to stick in your mind the same way, creating an illusion of scarcity. Differences in playing intensity and number of cards flipped can also affect drawing rates while being entirely unrelated to the card itself. In short, impressions are not reliable, and without hard numbers to crunch, a conclusion cannot be drawn.
--
Prof. Sian Kan, at your service.
|
|
|
+2
link
|
DrMoriviri
Posts: 67
4/27/2017
|
About the reddit thread: I’ve been thinking about it and I think there’s something wrong with the “experiment”. In particular, with this phrase ” For the scope of this experiment, I'm only comparing cards labeled "Standard" frequency that weren't affected (added to or removed from the deck) by any choices I made during the period in question. This left me with 78 relevant cards”. This means that there were far more “standard” cards drawn during the experiment, but omnilyx only considered the ones “not affected by any choices he made during the period in question”.
Point is, adding or removing a card in your deck affects the probabilities by which other cards are drawn, and if during the experiment card were added or removed this could have skewed the results.
Let’s say my qualities allow only 5 cards to be drawn (obviously impossible case): 3 standard, 1 infrequent (80%) and 1 unusual (20%). The odds of drawing a particular standard card are 100/400 = 25%; the odds for the infrequent are 80/400 = 20%; and the odds for the unusual are 20/400 = 5%.
Now, if I add another standard card (let’s say I buy a 3 card lodging), my new odds, for a particular standard card, lower to 20%; for infrequent 16% and for unusual 4%.
So, if omnilyx in the period of the experiment added or removed much cards from his deck, the result shown could be wrong.
Sorry for my English, not my first language
edited by DrMoriviri on 4/27/2017
|
|
|
+1
link
|
Kaijyuu
Posts: 1047
4/27/2017
|
Naw it still works.
Say you're playing a game that requires dice with 4 sides, but all you have are 6 sided dice. You can play fairly by simply rerolling any time you get a 5 or a 6. You simply throw out and don't count the extra data.
A similar principle works here, where they were just comparing standard frequency cards. Drawing a non-standard frequency card did not affect their results at all; it was simply thrown out. Of course, their experiment told them nothing about the non-standard frequency cards, and only told them things about the standard frequency ones, but that's fine; we're just checking if the displayed frequency for standard cards is accurate to a statistically significant level, and it appears pretty convincingly that they are not.
edited by Kaijyuu on 4/27/2017
--
Be of good cheer. Our contacts have assured us that your sins are forgiven.
|
|
|
0
link
|
DrMoriviri
Posts: 67
4/27/2017
|
Kaijyuu wrote:
Naw it still works.
Say you're playing a game that requires dice with 4 sides, but all you have are 6 sided dice. You can play fairly by simply rerolling any time you get a 5 or a 6. You simply throw out and don't count the extra data.
I'm not sure this applies to the experiment
Let's say i want to know if a dice (d6) is rigged, so i throw the dice an arbitrarly large amount of times and keep track of how many 1s show up. Now imagine the first day of my experiment i use in fact a d6, but the second a d20, the third a d4, the fourth again a d6, etc. The times a 1 shows up doesn't indicate if my original d6 is in fact rigged.
I think this is what happened in the experiment, since the setup suggests that several standard cards were added to (or removed from) the deck, affecting the odds of other cards showing up.
edited by DrMoriviri on 4/27/2017
edited by DrMoriviri on 4/27/2017
|
|
|
0
link
|
Rostygold
Posts: 346
4/30/2017
|
Siankan wrote:
This can be taken advantage of, to some degree. If I really need to draw multiples of a certain card, then in addition to deck-thinning I will arrange things so that I'm only drawing one card at a time, thus giving me the best opportunity to draw it twice in a row. It's not a huge difference, but every bit counts.
I second this. It gives better control over the drawing of opportunity cards, especially in places which have their own decks.
edited by Rostygold on 4/30/2017
|
|
|
0
link
|
Alysian
Posts: 57
5/1/2017
|
Rostygold wrote:
Siankan wrote:
This can be taken advantage of, to some degree. If I really need to draw multiples of a certain card, then in addition to deck-thinning I will arrange things so that I'm only drawing one card at a time, thus giving me the best opportunity to draw it twice in a row. It's not a huge difference, but every bit counts.
I second this. It gives better control over the drawing of opportunity cards, especially in places which have their own decks.
edited by Rostygold on 4/30/2017
I do as well, and can I just say I love the mathematical enthusiasm this community displays? I don't have the chops to meaningfully contribute to it, but I really appreciate being able to read through all the work others have put into thinking about these things. Carry on 
--
Alysian, gone North, grieved, gone.
Alybye, A Midnighter available for children's parties. No appointment necessary.
|
|
|
0
link
|
Rostygold
Posts: 346
5/1/2017
|
Thanks for everyone else's posts here. I have another thing to raise.
I do know that cards already in the hand will not appear in the hand again, but is there anything from Failbetter officials that clarify this further?
(The following assumes that suincide's theory is true.)
Specifically, there is the ambiguity about why these already on-hand cards are not drawn. The first possible reason is that their contribution to the range of the RNG roll of drawing a card is already excluded, i.e. the RNG roll simply cannot land results which lead to the cards.
The next possible reason is that their weightage is not excluded, meaning that the RNG roll can land on their segment in the range. However, the RNG roll is re-rolled.
My guess is that the first possibility is likelier, since this is smarter programming which leads to less computing overhead from not having to make re-rolls over and over in the case of the RNG roll somehow landing on the sub-ranges of the cards over and over.
edited by Rostygold on 5/1/2017
|
|
|
0
link
|
Optimatum
Posts: 3666
5/1/2017
|
Would it really matter? The second possibility would be less efficient, but from a player POV there wouldn't be a difference, aside from slight potential lag.
--
Optimatum, a ruthless and merciful gentleman. No plant battles, Affluent Photographer requests, or healing offers; all other social actions welcome.
 Want a sip of Cider? Just say hi!
PM me for information enigmatic or Fated. Though the forum please, not FL itself.
|
|
|
0
link
|
xKiv
Posts: 846
5/1/2017
|
In fact, in principle the set of drawable cards can change every action (or when you change equipment, buy something, sell something), so it makes more sense to me to just recalculate everything (which includes excluding already held cards).
--
https://www.fallenlondon.storynexus.com/Profile/xKiv - a witchful, percussive, dangermous and shadowry scholar of coexplodence, hopsidirean, and walker of fallen kitties.
|
|
|
0
link
|
Siankan
Posts: 1048
5/2/2017
|
xKiv wrote:
In fact, in principle the set of drawable cards can change every action (or when you change equipment, buy something, sell something), so it makes more sense to me to just recalculate everything (which includes excluding already held cards).
This, I think, is likeliest to be on-target. We know from multiplied experience that you can draw a card you were made eligible to draw five seconds earlier, so the simplest solution is that card options and probabilities are calculated each time a card is drawn.
--
Prof. Sian Kan, at your service.
|
|
|
0
link
|
Saklad
Posts: 528
5/2/2017
|
If we’re going to get into it this much, we might as well try decompiling Sunless Sea to see if Failbetter left Opportunity Cards in with the rest of the StoryNexus code.
edited by Saklad5 on 5/2/2017
--
Saklad5, a man of many talents
|
|
|
0
link
|
Anchovies
Posts: 421
5/2/2017
|
Saklad wrote:
decompiling Sunless Sea
I'm pretty sure this would end up with fingerkings coming out of your disk drive.
--
Perhaps our role on this planet is not to worship God — but to create Him.
—Sir Arthur C Clarke
Lionel Anchovies. Character on indefinite hiatus.
|
|
|
0
link
|
Kaijyuu
Posts: 1047
5/2/2017
|
Guess I'm safe then, not having any optical drives nor floppy drives, and can reasonably switch to a solid state hard drive. 
--
Be of good cheer. Our contacts have assured us that your sins are forgiven.
|
|
|
0
link
|
Rostygold
Posts: 346
5/2/2017
|
Anchovies wrote:
Saklad wrote:
decompiling Sunless Sea
I'm pretty sure this would end up with fingerkings coming out of your disk drive.
It's easier than you would think. Failbetter writes the code for Sunless Sea and packages most of them in plain text.
Finger-kings would write code in more sophisticated ways, and they don't need hands for that.
|
|
|
0
link
|