PDA

View Full Version : Print-Run Data Entry



syphonhail
06-03-2014, 06:51 PM
I was going to do this privately in a small group, but it makes much more sense to make this a collective project. For us to publicly get information about print runs, we need to collect a decent enough sample size. Print-run information should be public and not left to a select few who benefit.

As such, I have a simple entry form here: https://docs.google.com/forms/d/1Qpsh0ZsernpxeoVXonXm1bGTRjvHXtOhIv3GxVEhHMg/viewform

The results will post here and will be public for anyone to anlayze:
https://docs.google.com/spreadsheet/ccc?key=0AlgAwkjy48cFdFd1SXhjMFZIZVV1V1dtTnVlVDFLR Xc&usp=sharing

Only enter your commons and only enter your commons if you have data for all 11 (do not make up data).

Please enter your username.

Werlix
06-03-2014, 07:08 PM
Good idea, I'll definately be adding some of my draft packs into this

Shivdaddy
06-04-2014, 05:51 AM
If it gets figured out, they will change it.

Xenavire
06-04-2014, 05:54 AM
If it gets figured out, they will change it.

Why would they? It will only ever affect draft, because people will be able to figure out the cards that might have been picked out of a pack they just got passed. It will never be able to be abused to get certain cards out of a pack. Thats true across pretty much every MMO with RNG. You never know the seed, so you can never abuse it. That simple.

hexinggems
06-04-2014, 06:46 AM
Just curious if they mentioned that there was actual the concept of print runs with Hex?

negativeZer0
06-04-2014, 06:52 AM
yes it was just confirmed in another thead minutes ago



Print runs in commons exist, but it's not a consistent print run across multiple packs you open. Print runs just help you read signals in draft (i.e. this common appears next to these two--so when it's missing, I know someone is in that shard to my right).



I'm sure I'm a lone voice in the wilderness here, but I hate that feature so much. Of all the horrible weaknesses of the physical medium, I can't believe that anyone would not only choose to retain this one, but actually make a development effort to implement a flaw that wouldn't naturally exist in a digital game.The fact that you can't tell between print runs and a random distribution of commons in a pack speaks to the fact that this is not really an issue. It's been like this since Alpha.

It's not a flaw. Believe it or not, physical TCGs can both collate using print runs and 100% randomly. Some TCGs have print runs and some are completely random. We chose to have print runs to give more depth to drafts at the top end.




I feel the same way. Last I heard that was only speculation, and I'm sorry to hear that it's been confirmed. I feel the same way about it as I do about the recent discussions on changing the shuffler to avoid resource screw and imitate resource weaving from physical games. Random should mean random, with the only restriction being the rarity distribution and no duplicates in a pack.Let's not conflate randomizing the deck with how we present a subset of cards in a pack. Just because one is random, doesn't mean that the other has to be.




I want to have to make decisions between all the possible arrangements of commons, not just a limited subset.The fact that you don't notice this during the draft right now (feature has been in since Alpha) probably means that there are enough arrangements of commons to make it appear random enough.

Bottom line: it's not like we print off the same run of 11 commons for each pack. You would notice that very quickly. The algorithm is more complicated than that and provides a better play experience.

Also to nip this in the butt right now. Like Xen said Print runs will in no way allow cherry picking packs because of the way pack opening works in hex. If you don't believe me fine but please don't have that discussion in this thread.

negativeZer0
06-04-2014, 07:08 AM
@ syphon / anyone that actually knows what is involved in doing this
forgive my ignorance on this but don't we need consecutive packs to figure this out not just one pack at a time?

TJTaylor
06-04-2014, 07:52 AM
Hey, Zubrin.

I've entered some packs for you and I'll enter a bunch more during the day today.

I wanted to post to help get people started in understanding the print run for commons in set 001.

There are two major tracks for the commons plus two cards that get inserted which do not appear at this time to follow the run at spot 3 and 11. Most of the time, track 1 will be cards 1, 2, 4, 5, and 6 while track 2 will consist of cards 7, 8, 9, and 10. However, there are times when track 2 will bleed over to card position 6 which I will explain momentarily.

Both tracks have cards that are specific to that track, meaning they can't appear in the other track. The tracks also share a pool of cards that can be found in either track. I believe there are at least 3 sheets of commons, though possibly more. One is only found on track 1. One is only found on track 2. And one can be seen anywhere in the pack. The insert cards can be any common in the set. Track 2 will include common 6 only when both tracks contain cards from the sheet that is dual track.

I've inserted a picture of a pack to help illustrate what I'm talking about. In the example, track 1 (red) is being fed by the sheet shared by both tracks while track 2 (blue) is being fed by the sheet of commons that only is found on track 2. Notice that track 1 gets the position 6 card in this scenario.

I hope this isn't too confusing and maybe people can pick up this info and try to figure out if there is a pattern to the insert cards.

1974

TJTaylor
06-04-2014, 09:01 AM
@ syphon / anyone that actually knows what is involved in doing this
forgive my ignorance on this but don't we need consecutive packs to figure this out not just one pack at a time?

For the commons, no. You just need lots of pack examples. But we may never figure out how the insert commons work due to not being able to get consecutive pack results and uncommons are probably impossible as well since each one seems to be on its own track from what I have observed so far.

syphonhail
06-04-2014, 10:16 AM
Thanks TJ, that picture and post is very helpful. Part of the reason to open the project beyond my small group is to get people who know more about this than I do. :)

MisterBurkes
06-04-2014, 11:01 AM
I've been collecting and analyzing print run data for some time and have accumulated hundreds of screenshots (with help from Pentachills and Ebynfel). Here are some of the strings I've discovered:

Pterobot Murder Noble Citizen Ruby Aura
\\\
Ivory Pawn Reversion Shadow Lurker Windborn Acolyte
\\\
Murder Inner Conflict Thunderbird
\\\
Shroomshaw Burn Repel Researcher Adept Briar Legion
\\\
Shield Trainer Mesmerize Bravery Wailing Banshee
\\\
Burn Twisted Fate Wild Child
\\\
Blood Cauldron Ritualist Inner Conflict Bucanneer Savage Raider Runts of the Litter
\\\
Zombie Vulture Skyle Griffin Shards of Fate Ruby Lance Wild Aura
\\\
Thunderbird Wind Whisperer Wild Aura Darkspire Priestess Ruby Pyromancer
\\\
Shroomshaw Noble Citizen Burn Twisted Fate Howling Brave
\\\
Effigy of Nulzen Shield Trainer Gem-Crazed Berserker Countermagic Boulder Brute
\\\
Ivory Pawn Reversion Shadowblade Lurker
\\\
Heavy Welding Bot Countermagic Honeycap
\\\
Blood Aura Turreted Wall Bloodcrazed Zealot Phoenix Guard Trainer
\\\
Captain of the Guard Evolve Suppressive Fire
\\\
Gearsmith Sniper of Gawaine Runts of the Litter

cavench
06-04-2014, 11:12 AM
Thanks TJ, that picture and post is very helpful. Part of the reason to open the project beyond my small group is to get people who know more about this than I do. :)

This isn't as daunting as you may think. I spent about an hour yesterday on this and I already found some patterns. The commons comes in a block of 3~4 cards. Here are what I already found:

Ruby Lance
Wild Aura
Peek
Sapper's Charge

Sniper of Gawaine (?)
Grim Skull Sorcerer
Wild Growth
Gearsmith

Glimmerglen Witch (?)
Devoted Emissary (?)
Giant Corpse Fly
Arena Brawler

Pterobot (?)
Noble Citizenry
Ruby Aura
Twisted Fate

Mystic of the Tranquil Dream
Feral Ogre
Ruby Lance
Darkspire Enforcer

Vine Trap
Axe Bot
Misfortune

Cards with (?) are guesstimates and need further confirmation. At this point I'm not sure if it is worth my effort to continue with this. Like Chark said, it is only for high level players. There is likely hundreds of such blocks and I'm not going to try to memorize them. If you are going that route, then I suggest you memorize them visually with each card side by side, rather than memorizing them by text.

However, for people at my level, it is only worth knowing the blocks that contains the first picks. For example, I always wonder why I see Murder+Inner Conflict+Thunderbird (unconfirmed yet) together so many times before. And this kind of information is best used during 2nd~4th pick for each pack. There are 11 commons in a pack, so if you were handed a pack with 10 commons then it will give you a good idea exactly which card (and shard) that someone on your right is taking. Likewise, its kinda pointless to memorize the blocks that contains the dregs. After 2nd-4th pack, don't try to spot the blocks anymore, just go back to your intuition on the reads.

EntropyBall
06-04-2014, 11:21 AM
Even this small data shows that its more complicated that Chark seemed to be saying. Its not as straightforward as "this card is always between/after/before these 2 cards"

I'm a drafting newbie, but I don't understand how this print run info is supposed to help you know what was taken. From the data, it seems that Gearsmith, Honeycap, Repel, Murder, and Ruby Aura frequently appear in the same pack, but not always. So if you get passed a pack that has 4 of those and no murder, you don't know that Murder was picked.

Shards of Fate, Survival of the Fittest and Scrap Welder, however, are joined at the hip and appear in succession almost every time except when something is between them in slot 3, or on line 28 with Shards just appears in slot 4 by itself.

Edit: and 2 huge posts appeared while I was typing this up.

cavench
06-04-2014, 11:30 AM
Pterobot Murder Noble Citizen Ruby Aura
\\\
Murder Inner Conflict Thunderbird
\\\
Thunderbird Wind Whisperer Wild Aura Darkspire Priestess Ruby Pyromancer
\\\
Effigy of Nulzen Shield Trainer Gem-Crazed Berserker Countermagic Boulder Brute


Thanks MisterBurkes, didn't see your post before I typed mine. The above subsets are the ones that I would deemed worth knowing.

syphonhail
06-04-2014, 11:46 AM
Hey Misterburkes, Pentachills sent me the screenshots last night and my first several entries actual come from that top level folder. I wanted to create a database as there might be some patterns that I (or others) may not simply see and it may be easier to tease those out statistically. We also may observe some more complex dependencies, if they exist.


The biggest thing we need is more data/entries to parse out, but I will keep entering packs. The rest of my data entry will be from my own observations, so if you were inclined to enter your data, we won't over-sample the same data.

Shivdaddy
06-04-2014, 06:24 PM
Why would they? It will only ever affect draft, because people will be able to figure out the cards that might have been picked out of a pack they just got passed. It will never be able to be abused to get certain cards out of a pack. Thats true across pretty much every MMO with RNG. You never know the seed, so you can never abuse it. That simple.


Read bold, draft is what runs this game. They will not let its integrity be ruined. I find it strange you think it would be ok for people to have a separate program up that shows people what the next pack might hold. This happened in MTGO once and once it leaked out they had to change it. It is cheating.

Xenavire
06-04-2014, 06:29 PM
Read bold, draft is what runs this game. They will not let its integrity be ruined. I find it strange you think it would be ok for people to have a separate program up that shows people what the next pack might hold. This happened in MTGO once and once it leaked out they had to change it. It is cheating.

So what, you expect this program to materialise (without any API right now) and be accurate and fast (without API you would have to input the card details manually) and that if all those things aligned, that Hex wouldn't simply change the print runs in a way that would completely invalidate the program?

Seeing patterns in draft is fine. But of course using a program to predict cards is wrong. (Also, you need to remember that we are drafting without knowing who is next to us. You could draft perfectly and still end up against a complete unknown from across the table. And then lose to a bunch of cards that the program couldn't predict.)

MisterBurkes
06-05-2014, 07:55 AM
Read bold, draft is what runs this game. They will not let its integrity be ruined. I find it strange you think it would be ok for people to have a separate program up that shows people what the next pack might hold. This happened in MTGO once and once it leaked out they had to change it. It is cheating.

You can't actually predict the next pack. The existence of a print run algorithm allows us to analyze patterns in a single pack. But from what I've seen, each pack is several random subsets of that print run algorithm. The best evidence of this can be found in some of my own screenshot samples. I took screenshots whenever commons were unpicked (so typically Pack 1-3 per round). There are several instances where I've seen two packs in a row that are either 1. the same exact same subset of 4-5 commons (which is amazing for confirming longer strings), or 2. nearly identical set of commons, but has a 1-2 card offset at the beginning/end (representing a string cut-off).

Imo, print run analysis will primarily remain useful in Picks 1-5 of each round, when important 3-5 card common strings are still identifiable.

Also, I've been unable to discern any correlation of a pack's common strings to its uncommon strings / rare card, except for the fact that judging on screenshots of only Pack 1-2; if a pack contains weaker common strings like: Captain of the Guard Evolve Suppressive Fire or the dreaded Ivory Pawn Reversion Shadowblade Lurker; the entire pack is usually bad, including the uncommons / rare.

I might even go out on a limb and claim this: if a pack has Ivory Pawn Reversion Shadowblade Lurker, you've just rolled the worst pack in the round (tie if there are other packs with that string in the same round). It has interesting implications in regards to the print run algorithm, so my best guess is that the developers intended certain packs to be "weak/bad" to spice up the difficulty in draft.

Last but not least, uncommon string patterns have been notoriously difficult to crack. I think I've found a couple of 2-card uncommon strings, but any 2-card string is basically useless information. Hopefully with enough data, we can finally crack a few uncommon strings.

Peseto
06-05-2014, 08:59 AM
I decided to supply a little bit of data, and am working on R-code to quickly look for patterns. If anyone has a large amount of screenshots and is too lazy to enter them manually just send it to me and i can run a screencapture program on it.

negativeZer0
06-05-2014, 09:59 AM
Read bold, draft is what runs this game. They will not let its integrity be ruined. I find it strange you think it would be ok for people to have a separate program up that shows people what the next pack might hold. This happened in MTGO once and once it leaked out they had to change it. It is cheating.

You have 0 clue what you are talking about. Print runs were a conscious decision to include them so people could use them when drafting. Changing it defeats the entire point of including it to begin with.

Print runs DO NOT allow you to predict the contents in the next pack
They tell you what card is missing from the pack in front of you.
(MTGO had a deeper issue because of the way packs worked in that game that has nothing to do with print runs)

I again ask that if you want to continue this discussion (+/- of print runs) please start another thread (I will be happy to explain in more detail if you want in said other location)
But please allow this thread to stay on topic. Thank you

syphonhail
06-05-2014, 12:06 PM
I decided to supply a little bit of data, and am working on R-code to quickly look for patterns. If anyone has a large amount of screenshots and is too lazy to enter them manually just send it to me and i can run a screencapture program on it.


Hopefully Misterburkes is willing to give you the list of images he has, there are a ton and it would be great if someone can enter them. I would share them, but they are not my files to give out.

MisterBurkes
06-05-2014, 01:43 PM
Hopefully Misterburkes is willing to give you the list of images he has, there are a ton and it would be great if someone can enter them. I would share them, but they are not my files to give out.

I have no problem with sharing them. Feel free to share the Google Drive w/ Peseto.

syphonhail
06-06-2014, 12:57 PM
Just as update, we are up to 137 entries and we are about to get a bunch more that Peseto is working on (I will be manually adding them to the database). Our sample size is becoming, but given that the print runs can come in different combinatorials and may not be simple a list, we need as much data as we can get, so keep opening packs and entering them.

EntropyBall
06-06-2014, 03:09 PM
One thing I've noticed from looking through the data is that there are several spelling errors that will hinder automated analysis of the results. Anyone attempting to do it might want to run unknown strings through a Damerau-Levenshtein distance algorithm (http://mihkeltt.blogspot.com/2009/04/dameraulevenshtein-distance.html) to find the closest match (out of all existing card names)

syphonhail
06-06-2014, 08:10 PM
One thing I've noticed from looking through the data is that there are several spelling errors that will hinder automated analysis of the results. Anyone attempting to do it might want to run unknown strings through a Damerau-Levenshtein distance algorithm (http://mihkeltt.blogspot.com/2009/04/dameraulevenshtein-distance.html) to find the closest match (out of all existing card names)

I had planned on doing manual fixes at some point, if you see any, just let me know (row number and column letter) and I can fix them that way. It's more labor intensive than I like, but that is partly my charge for being the curator.

EntropyBall
06-06-2014, 09:41 PM
i had planned on doing manual fixes at some point, if you see any, just let me know (row number and column letter) and i can fix them that way. It's more labor intensive than i like, but that is partly my charge for being the curator.

2k
45k
120g
118k
122h

syphonhail
06-07-2014, 01:21 PM
Thanks, Peseto is currently cleaning the database as well as adding another 50+ observations to it, so I will copy/paste that into the master data when they are done. Then I will cross check those errors to make sure they are purged as well.

Barkam
06-07-2014, 01:55 PM
Keep up the good work guys.

Peseto
06-07-2014, 06:33 PM
In the new updated version i added about 140 new observations, which should be enough to do some rough Analysis. Gonna post my findings in the next couple of days.

syphonhail
06-08-2014, 12:41 AM
The masterdatabase now contains the 140 additions by Peseto.

Voormas
06-08-2014, 04:09 AM
I started looking at it for a pattern I had noticed previously (Heavy Welding Bot next to Scrap Welder) and there seems to be some cool stuff going on but really running analysis on this kind of thing is way beyond me - i'll just have to keep opening packs as a way to contribute :cool: