Hi,

I'm wondering if the algorithm and/or production happens in some sort of order as in if it will produce together lots of decks with the same houses. I bought the starter box and the two decks were brobnar, untamed and dis. Then a few weeks later I bought another deck couple to see if I could get to see any mars and I did but both new decks where brobnar, mars and dis. So out of four decks all four have brobnar and dis and then two with untamed and two with mars.

If the maths in this post are correct then the probs of same houses three in two decks are 4.49% so the probs of that happening twice are even lower, thus not impossible, probabilities are evil like that. Anybody knows if within a box of decks is any diversity guaranteed?

https://www.scientificamerican.com/article/why-our-brains-do-not-intuitively-grasp-probabilities/

I've purchased 10 decks now, 5 of which had overlapping houses

2x logos dis sanctum

3x sanctum shadows Untamed

I just traded or gifted my overlapping decks to friends

This was between 2 different stores and 3 different boxes. I have yet to buy a full box to see if a single display contains any amount of house overlap

After opening one deck, the odds of opening the next deck with the same set of house is 1 in 35. Your two decks matched, but 34 other people that opened 2 decks likely didn’t match. It’s the exceptional cases we notice though. I haven’t seen any posts about “I opened two decks and they had different house combinations” yet.

As for variability in booster boxes. Since decks are randomly generated digitally, I’d say any patters found in a booster box is purely coincidental. It’s not like they are printing all the Mars/Sanctum/Dis decks at once and then another set and another set and then putting them in a big bin to shake up. Random numbers just work in fun, random, ways. Heck, there could be an entire booster box of Brobnar, Mars and Dis. Statistically improbably, but not impossible.

With only 3 decks, I got two of the same house combination myself (even in the same order: Dis, Logos, Brobnar), yet they both play very differently.

Part of the metagame is that "what if" of getting houses you've got before: keep, trade, sell? I embrace that and intend to keep both decks.

That's fine. I am willing to bet that those two decks play slightly different and have different combos

The decks are still different, so what's the problem?

3 hours ago, Lyraeus said:

That's fine. I am willing to bet that those two decks play slightly different and have different combos

Yes of course

2 hours ago, Robin Graves said:

The decks are still different, so what's the problem?

No problem at all, only I'd like to see more of the other houses. I'm just curious, I'm in Australia, which is a pretty small market compared to... well almost any other country and were wondering if we've ended up wiht lots of the same round here.

I ended up with 8 decks before I had each house represented. 2 of my first 3 had the exact same factions. I thought this would be the one thing I didn’t want but I absolutely love how different they play and how fun they both are.

I can't be quite sure, but sometimes it seems a box will have one house appearing in most of the decks. At the pre-release a lot of players had dis, while the box I bought seemed to have more Shadows and Mars, but that can easily be a false impression.

My girlfriend surprised me with an early Christmas present of four decks so we would have something new to play while we spent a week at her parents. All four have Mars as one of their houses.

I had to buy a sixth deck before I got one without Mars in it. That's how random works.

People have this idea that random should be smooth. It isn't. Ask someone to make twenty dots on a piece of paper randomly and they'll almost always make a very smooth, even arrangement where the dots aren't clustered at all. Ask a computer to put twenty random dots on a piece of paper and clustering is something that will happen a lot. This phenomena is a real challenge to scientists who sometimes have a difficult time determining if something that looks like an unusual cluster IS unusual, or just random stuff randomly clustering. There are entire branches of mathematics dealing with this problem.

The human brain is just terrible at statistics. It wants to find patterns in everything, even where they don't exist.

Hi

I think I am lucky. I have 14 decks. Once I bought 2 decks at my LGS and both were the same houses. 4/14 decks are the same houses. My 7th deck I bought was a horseman deck and my 8th deck had a Maverick. Archon colours: 2 silver, 4 yellow, 1 blue, 1 green, 1 purple, 5 bronze. My friend has 5 decks and 3 of them have purple archons. Out of 14 decks: 6 Brobnar, 4 Mars, 6 Dis, 5 Logos, 4 Shadow, 8 Sanctum, 9 Untamed. 2 of my decks have ''Director'' in the name: X. Steele Director of the Cackling School and L. Yassafrass, Director of the Public Pub, I opened both of those decks one after the other LOL. My 3 favorite names so far: Hungrily Subordinate Xandre, Overlord Pinkthrop the Poor Companion, L. Yassafrass, Director of the Public Pub (Yassafrass, what a great name). No rude names yet:(

I'm from Australia too, and when I got my first 3 decks, I was also surprised with the same houses appearing. The 4th deck is fine though.

1st: Sanctum Shadow Untamed

2nd: Brobnar Dis Untamed

3rd: Dis Logos Untamed

4th: Brobnar Logos Mars

All those untamed, and mars came in late. But even then, they each have their own styles of untamed, so not that bad.

12 hours ago, Suttkus said:

I had to buy a sixth deck before I got one without Mars in it. That's how random works.

People have this idea that random should be smooth. It isn't. Ask someone to make twenty dots on a piece of paper randomly and they'll almost always make a very smooth, even arrangement where the dots aren't clustered at all. Ask a computer to put twenty random dots on a piece of paper and clustering is something that will happen a lot. This phenomena is a real challenge to scientists who sometimes have a difficult time determining if something that looks like an unusual cluster IS unusual, or just random stuff randomly clustering. There are entire branches of mathematics dealing with this problem.

The human brain is just terrible at statistics. It wants to find patterns in everything, even where they don't exist.

Ohh yes, the glorious "it's not a pattern if you keep adding zeroes to the size of data" defense. 😏

2 minutes ago, 10Ten said:

Ohh yes, the glorious "it's not a pattern if you keep adding zeroes to the size of data" defense. 😏

More like the '2 is not a valid sample size for anything, neither is 100' defence or rather attack/critique. Humans are wired to recognize patterns, that is basic psychology. This means that when looking for patterns, we have a high false positive and low false negative rate. There is a reason statistical research needs to be codified so much and is constantly evolving to be more restrictive, if we want to see something we probably will find a way to see it.

24 minutes ago, Admiral Deathrain said:

More like the '2 is not a valid sample size for anything, neither is 100' defence or rather attack/critique. Humans are wired to recognize patterns, that is basic psychology. This means that when looking for patterns, we have a high false positive and low false negative rate. There is a reason statistical research needs to be codified so much and is constantly evolving to be more restrictive, if we want to see something we probably will find a way to see it.

And sometimes, there just might be a pattern.

6 minutes ago, 10Ten said:

And sometimes, there just might be a pattern.

I agree, I'm just saying that here it is very far from likely.

OK.

I open a deck and get any three houses, then.

My second unopened deck has a pretty high chance of drawing a duplicate house with my first, because at that point any of the three will suffice. In my third deck the chance of drawing a duplicate house will diminish somewhat, my gut would say about a 60% if I received 2 duplicates in deck 2 and about 45% with just 1.

I think with a sample size of just 4 decks you will see a common house in 3 decks about 66-75% of the time. Even though I may not have any overlap in my third deck there is still a 60/45% change my fourth will share a common house. Plus if my third shares a different overlap with deck 2 that also becomes a possible 3 house overlap.

Five of my seven decks had Mars.

At first I was very happy because I liked the theme. Unfortunately they ended up being my least liked house to play.

Conveniently, the Numberphile YouTube series posted a video today on how bad humans are at doing random:

https://www.youtube.com/watch?v=tP-Ipsat90c

On 12/18/2018 at 9:24 AM, Admiral Deathrain said:

I agree, I'm just saying that here it is very far from likely.

We have no idea on how the algorithm works for Keyforge.

With the 3 separate factions per deck and only 7 factions, there is going to be patterns/runs emerging out of the decks.

5 hours ago, 10Ten said:

We have no idea on how the algorithm works for Keyforge.

With the 3 separate factions per deck and only 7 factions, there is going to be patterns/runs emerging out of the decks.

Judging from the data that is available even through just the app we have no reason to assume anything but a random house selection on a per deck level. Occasional occurrences of emergent patterns do not change that, quite contrary, they are hints that there is actual randomness (or at least an adequate aproximation of it) utilised, rather than some form of equalised randomness.

We simply have no good reason to theorise about patterns in deck distribution. The manufacturing process of completely digital printing sure doesn't benefit from it. Yes we can't be sure how the algorithm specifically works, but there is nothing supporting the theory that it doesn't work as advertised and a good amount of data suggesting that it does.

Sort of the inverse to the OP. I bought my first (and so far only) 2 decks from B&N and have zero overlap. So with two decks I've got 6 of the seven houses covered. The math says that should happen around 15% (off the top of my head) of the time but it feels like good luck to me.

On 12/19/2018 at 5:35 PM, Admiral Deathrain said:

Judging from the data that is available even through just the app we have no reason to assume anything but a random house selection on a per deck level. Occasional occurrences of emergent patterns do not change that, quite contrary, they are hints that there is actual randomness (or at least an adequate aproximation of it) utilised, rather than some form of equalised randomness.

We simply have no good reason to theorise about patterns in deck distribution. The manufacturing process of completely digital printing sure doesn't benefit from it. Yes we can't be sure how the algorithm specifically works, but there is nothing supporting the theory that it doesn't work as advertised and a good amount of data suggesting that it does.

Actually no.

See the issues with Statistics is we don't live in that world.

Sure there may be 11 billion gizillion decks capable, but you know.... There's only what 12decks per box? 10-15 boxes per case? Having 35 cases with the exact same faction distribution isn't even a blip on the 11 billion gizillion possible, cuz statics says so. Yet how would you feel when someone says "cause statistics", you still feel cheated, because we mere humans don't live in the statistical world