I think there is some problem with randomisation in the playtester
TappedOut forum
Posted on April 5, 2018, 1:47 p.m. by Argy
I do a lot of practice using both the play tester on this site, and cards irl.
There is some sort of problem with the randomisation process in the playtester, and it seems to directly relate to number of lands.
If you have 24 lands you either get consistently flooded or screwed. Much more than you do irl.
By comparison, if you use 25 lands you rarely get flooded, and you almost never get screwed if you use 23 lands or less. Even if you use a number of lands as small as 20, the playester usually gives you two in your opening hand.
That never happens with the same frequency irl.
I don't know if anything can be done about this, but I thought I'd bring it up.
I might see it more than other people because I use the tester at least 20 times per day.
I hate testing on TO. I always export the deck into Cockatrice and goldfish there. It's also nice to run a local game with 2 decks instead of 2 windows on your browser.
April 5, 2018 3:08 p.m.
I usually just test on TappedOut to see if a deck works.
It's also easier because I can test where ever I can have a phone.
The play tester never used to be this ... predictable.
April 5, 2018 5:15 p.m.
I used to build my decks and play mock games with myself until i figured out how to use the play tester on this site (never used a different site play tester)....But i use this one A LOT... like sometimes i playest my deck when im board and in the car/potty... truthfully... i dont really have any issues with the site playtester.... i get mana screwed and mana flooded just about the same.... and even janky combo decks designed to go off by a certain turn tend to work out the same here as in real life.... but thats just for me anyway.
April 5, 2018 5:41 p.m.
Icbrgr I hope you're not playing Magic while you're actually DRIVING your car.
April 5, 2018 8:07 p.m.
Femme_Fatale says... #8
Before you start making claims, make sure to gather your data and present it. Because from the data I've gathered in the past, T/O's playtester has one of the best, if not the best, shufflers you'll ever find.
I say this, because I too exclusively use the T/O's playtester, and haven't found any significant problems with it that isn't also something that can easily occur IRL.
Here's data I've gathered for various reasons in the past. Nothing is over 1,000 tests so it is all pretty small sample sizes, but I'll do a run of how many turns it takes to get 5 lands, 1 of each coloured source in a three colour deck with 3 colourless lands in both 24 land deck and 25 land deck.
To read the mind funeral tests, read like this:
-Number of cards milled until 4 lands | number of times this occurred | number of mulligans
-# # #M
First Mind Funeral test I did, before recording mulligans:
Second Mind Funeral test I did:
April 6, 2018 6:46 p.m.
I didn't say I exclusively use the olaytester.
I said I use it AND I playtest irl.
If you say it's working, and you've tested it, that's good enough for me.
I will point out, though, that it's erroneous to suggest that just because something has worked in the past that means it will never have problems.
My topic was to try to find out if the play tester is working at the moment. Nothing more, nothing less.
April 6, 2018 7:31 p.m.
are yall talking about the 1.0 or 2.0 playtester?
what browser?
April 6, 2018 7:35 p.m.
If Femme_Fatale says it's fine, it must be fine.
I play using Safari, 1.0
April 6, 2018 7:44 p.m.
I seem to recall that the random used by 1.0 reveals a bug in the random implementation of some browsers
April 6, 2018 7:50 p.m.
SeekerofSecrets says... #13
I use chrome but im not sure if i use 1.0 or 2.0
April 6, 2018 7:53 p.m.
Femme_Fatale says... #14
Argy, in the moment it should be fine for 2.0. The second half of the coco data was made in the past 2 weeks. But since you use 1.0 this data doesn't have any relevance ... I'd do V1.0 testing but it goes down a lot so I can't get a lot of data.
yeaGO: are we thinking of the same bug yeago? Where sets of cards will cycle in and out of appearing in your deck? Example: getting Fiend Hunter in about 10 games in a row, and then not getting it in the next 10 games, and then have it reappear etc.
I've been exclusively using 2.0 playtester on chrome now and I haven't noticed that quirk, but it used to be common place in 1.0 and I feel like I experienced it on chrome the most? It's been a while. Could have also been firefox. However, that was back before I learned to gather data on my playtests so I don't know if it was a thing or if it was me just looking into something that didn't actually exist.
SeekerofSecrets V2.0 has the box of colourful buttons to the right, V1.0 has no colourful buttons.
April 8, 2018 6:09 p.m.
Femme_Fatale says... #15
Sorry it took me awhile Argy, but here it is. I didn't know what data to gather nor how to read it, so I went through a couple of iterations. Also Dominaria stuff. Thus, I only have a little over 200 data points for the 24-25 land comparison on V1.0 playtester.
The general gist of the data however, presents that there is no statistical significant difference between the two data sets. The mulligans are statistically the same, and the variances between the number of occurrences for each land count and average number of the strings of nonland cards are within acceptable parameters.
However I don't know if yeaGO did any changes to the playtester within this time.
I'll be doing comparisons with V2.0 playtester later, and that will contain some more different data types. Uncertain how long that will take.
SeekerofSecrets says... #2
I'll second this, I seem to always see my cards in multiples no matter how many copies are in deck. I don't know if this if from my limited tested done on this site (I normally use untap.in) or if there really is an issue.
April 5, 2018 2:14 p.m.