Replace Mechanics Broken


#1

I can’t tell you how many times I have replaced 2 or 3 cards at the beginning of the game just to have the same exact 2 or 3 cards to come back out. Maybe the algorithm needs tweaking?


#2

Lots of people feel the same way

Mathematical breakdown.

Original post here.


#3

I get the math to an extent, but what are the odds that 3 of the same card come out on a statistically suspicious basis? For example in one of my decks I have 1 crackthorn, almost any time i draw it and replace it, i draw it again, even if it is the only card I replace.


#4

[quote=“Chillyo, post:3, topic:7141”]
almost any time i draw it and replace it, i draw it again
[/quote]How often? It has to be something totally crazy like 9/10 times you get it back before it becomes suspicious. There’s so many players here that really unlikely things will happen to somebody every now and then.

Just note, I did my own test of mulligans here and Abrakam did too, and neither of us found any problem.


If you want to test it you need to do something like this.
Keep a table of how many times you rerolled your crackthorn, and how many times you got it back.

To avoid confirmation bias, the important thing is that you decide and physically mark that you’ve decided you’re going to reroll it before you actually redraw cards. The before is really important, to stop your own subconscious from messing with things (like forgetting to write the result if it doesn’t happen). You should put a mark in the table that you’re about to reroll, and only then do a reroll - beside that mark put a yes or no if you got it back or not. If you forget to put a mark then ignore the roll that time - this is kinda pedantic but it’s amazing how many errors are done if you’re not pedantic. For this you probably need at least 30 tests before you’re getting useful information. And every time you put a mark in your table you need to write the result - if you forget you ruin the test.

That’s a lot of work, so only do it if you really want to test it yourself.


#5

There are two types of people in this world - ones that suffer from confirmation bias and ones that run the numbers because they suffer from confirmation bias.

Yup, we all crunched the numbers because we felt something is wrong. People tend to think people who ran the numbers did something wrong, or didn’t include some cases and in general wanted to confirm a result. That last part is right - except they wanted to confirm something is fishy, it just turned out there is nothing such.

People underestimate both how confirmation bias and how math analysis work in practice. You see 10% number on a whiteboard and it feels fine, you live that 10% and it feels off. Too often. Not often enough. Worse yet, you ask a mathematician if 2+2=5 and he will answer: “impossible”. You ask the same question to statistics guy and he will answer: “extremely unlikely”. Silly thing, how this works.

And when saying “people” I do mean “all people”. Including the ones doing the number crunching. It’s just a fact of nature, that we put weight of emotions behind what we live through, and it means there is a huge dissonance between what we experience and what the numbers actually are.


#6

I will track all of my rerolls today and post my results here. I don’t think it is confirmation bias because I literally get the same cards back so often that I question every reroll now because sometimes I end up with 3 buffs and no playable cards. I will start with 10 practice games and reroll every game and if I don’t notice anything odd I’ll stop there. If things get weird I will continue in casual rerolling certain cards that show a pattern.


#7

I can’t say. They said they tested it and that it was fine, so you must just be getting