Found a new way to play. Seems great! http://www.geekswithjuniors.com/blog/2014/2/10/threes-the-best-single-swipe-puzzle-game-on-ios.html
Here come the sequences Data is in, sanitized and nicely formatted. The generous donors are: CzechCongo LeopardGecko Ryan Clark Scorpion008 If you wish to participate (it is not too late) then please follow the instructions in this post to record your card draws during a game and send me a PM. To circumvent the character limit of these posts, I have formatted the sequences in a separate html document for easier viewing and later expansion. I also provide the raw, sanitized sequences if you wish to crunch them yourself in the program of your choice. Here is the Python script I use to format them, which also contains an example of a simple statistic (see below). Threes! Sample Sequences formatted | raw One thing I wondered was this: how many 1s (or 2s) can you draw before a card of the other color comes up? In other words, I was trying to find the number of colored cards in the undrawn stack (see this post). The highest number I found in the sequences so far is 6 which would suggest that there are 3 red and blue cards in a freshly generated undrawn stack. See sequence #4 directly after h:48 there are six 1s without a 2 in between. Looking at sequence #4 more closely however reveals that 3 red/blue cards in the undrawn stack are too few. Figuring this out is left as an exercise to the interested reader. I would therefore suggest that there are 4 red/blue cards in the undrawn stack.
I'm pretty sure it isn't. Not sure, but I've friended him on GC and you'd think someone that went through that amount of honest work would be chomping at the bit to at least tell a single soul, let alone a developer. Also, our internal data doesn't really point to it. Not saying it's 100% BS, but yea. If he got to 6144, I'd love to watch him play another round.
Neat. Is there anything else to be learned here? Also, I think I have found some interesting things in the save game (plist) about the randomization. The file is under Threes!/library/preferences. Use iFunbox or ifile. If you can't see it I will try and find a way to upload mine.
Just got my best score ever,just over 21k,previously my highest was around 7,000 so as you can imagine I'm grinning like the cat that got the cream. Grr screenshot not working.
I said before that every few times I launch the game on my iPod Touch 5 there is no sound, well now it's every time. Is this a known issue?
Lucky you! I am topping out just under 10k currently. (for the plist file) Note some interesting things: 1) future spawn numbers 2) games since tutorial 3) grid size (4) 4) highest rank (1, perhaps it means only 3's at this point?) 5) pseudo spawn coordinates item 0-3 (again, pseudo randomness) 6) pseudo spawn coords list 0-11 7) pseudo spawn numbers item 0-2 8) pseudo spawn numbers list 0-11 9) pseudo spawn special item 0-20 10) pseudo spawn special list 0-20 11) several "randomness" numbers to choose how random of a number to spawn I guess 12) reported beyond number 13) all the highscores have a grid included to help check if there was cheating
Played with my wife taking turns (she's a newbie) and we scored my 2nd highest (over 22,000) AND more importantly, we're still on speaking terms!
Well, it had to happen. I hacked the game, and without that packet sniffing stuff. For the dev and those competing, my scores are not legit, (actual score is 10k). I have learned a lot about how the saving process works and how to bypass the implemented security. Not to mention I have seen the 768, 1536, 3072, and 6144 cards. I can show a screenshot if you guys want. If the devs want to pm me I will let them know exactly what I did. I still want to figure the algorithms out and play normally though guys. If the devs want to give me a clean slate, Gamecenter wise, feel free to do so, and I can then reinstall the game afterwards and wipe my local scores. If anyone has questions, feel free to ask, but be aware that I cannot advocate cheating and may not help you. Edit: To the devs, did you already know I hacked? If so, how?
Yikes! I guess that is one way to find out the mechanics of the tile allocations. Here is my current game: Like a few others, I am at 768 but am not at all sure if this is a way to make it to 1536, despite being in a relatively good position at the moment with quite a few open spaces. I am very disappointed that the creators of this wonderful game are not more forthcoming with how the tiles are allocated. 1. For what I observed, I am not convinced at all that the tiles are randomly allocated. Instead, there appears to be some fixed element with some pseudo-random elements added in when calculating what tile will appear next. 2. Even though Scorpion008 was able to locate the graphical assets (I presumed) that suggest the existence of 1536, 3072, and 6144 tile cards, this still does not prove that these tiles can assembled with the current mechanics of the game. Perhaps these are just extra assets that the developer has created. 3. Further, despite the wonderful work done by other posters in trying to elicit some pattern to the tile allocation, I am more concerned on whether or not that the game is subtly working AGAINST the player, deliberately and artificially increasing the chances that a WANTED tile will appear in an UNWANTED position. This is based on the fact that I observed many times whereby, despite a new tile may appear in multiple positions, the tile will appear in the LEAST FAVORABLE position. It is too much of a coincidence that this happens so frequently by chance alone. If this is the case, I would be very sad if such is the reason why the developer is silent about how the tiles are allocated. The reason for doing so is obvious. Once the number of empty spaces drops to 3 or below, you are far more likely to be at the mercy of the "random" tile allocation, since there is little room to recover from an unwanted tile position. Assembling a high numbered tile is near impossible if you do not have a lot empty spaces used to move the tiles around.
Nice of you to be open about it. Thinking about how I would manage savegames in a plist, I can imagine what you did. However, when it comes to figuring out the algorithm for new cards I feel that the fields in the plist file are more confusing than useful. Why should they be? It is not meant to be divined, we are just ambitious (and clever?) enough to attempt it. If they just told us, it would be boring, to be honest. I'm not quite clear what you mean with your second sentence here but I (and some others) share your opinion that cards are not chosen completely at random since that would lead to you losing a game due to an imbalance of red and blue cards most of the time. As far as I have given this thought, there should be nothing preventing you from achieving these cards honestly. The established theory of building a highway to match cards of increasing face value does break down at some point because these highways fill more and more of the board and take longer and longer to set up. I'm sorry to bust your conspiracy theory but the reason for you feeling this way is most likely a cognitive bias. "Losses" (i. e. negative experiences when positive outcomes were desired) weigh more than twice as heavy as "gains" (positive outcomes when you desired them and there was a possibility for a negative outcome). I recommend reading Daniel Kahneman's "Thinking, Fast and Slow" [Wikipedia] if you want to learn more about these and other mechanisms that go on in your mind without you realizing In other news, I have worked out a way to empirically determine the size and card distribution of the undrawn stack with the sequences I gathered. I have run some errands first but you can expect interesting results within the next 12 hours.
This is just selective memory at work. When you get a bad tile you notice it and remember it, when you get a good one you just don't notice and forget. Frankly, my impression was the exact opposite: many times, the tiles I got were just "too good to be true" and I thought the game might be silently attempting to help me. It's probably simply truly random and we are just noticing different things.
Why would the Developer tell us the tile allocation algorithm or anything like that? I am not quite sure why some people are digging so darn deep to try and find this information. Just play the game... Also, please do not post any secrets you divine on this post as there are some of us that don't want to know (but I like to follow the thread). Who knew game devs were such evil SOBs.
First results are in! Because we can. ;-) To me, knowing the algorithm doesn't ruin the game and I have the skill set, the time and motivation to find it. I respect your decision to not learn anything about the inside works of Threes!. I will therefore post any results in Spoiler tags and I would welcome it if everybody who posts findings or comments about other people's findings did the same. Speaking of which, I have some interesting results to share with the group. Spoiler I extended my Python script to now produce more statistics about the sequences. I also went ahead and did some tests. Let's recap: we assume there is a Tetris-style "grab bag" algorithm at work here. Since we are dealing with cards I like to call it the "undrawn stack". We assume that this undrawn stack contains an equal amount of blue and red cards to prevent games from ending due to flooding of one color. My first experiment went through all sequences and tried to find the lowest number of blue/red cards needed in the stack to make the sequence plausible. My results are as follows: Lowest number of 1s/2s needed in sequence #1: 38 Lowest number of 1s/2s needed in sequence #2: 4 Lowest number of 1s/2s needed in sequence #3: 28 Lowest number of 1s/2s needed in sequence #4: 4 Lowest number of 1s/2s needed in sequence #5: 19 I strongly suspect sequences #1,#3 and #5 to have typos as these numbers of blue/red cards seem very unlikely given the even spread of 1s and 2s in the sequences and the fact that the longest chain of one color is 6, as mentioned previously. Following that, I went ahead and assumed 4 blue/red cards in the undrawn stack. The question now is: how many white cards are in there? To figure this out, I created another experiment which goes through all sequences with a stack of 4 blue and 4 red cards and counts how many white cards it encounters until it runs out of both blue and red cards at which point it will log the number of white cards found and reset the stack. Here are the results of this experiment: Number of white cards needed in sequence #1: {'max': 9, 'min': 3} Number of white cards needed in sequence #2: {'max': 6, 'min': 2} Number of white cards needed in sequence #3: {'max': 6, 'min': 2} Number of white cards needed in sequence #4: {'max': 7, 'min': 3} Number of white cards needed in sequence #5: {'max': 5, 'min': 3} I then tried a similar experiment where I also fixed the number of white cards to 4 or 5 but then all sequences failed. This leads me to the hypothesis that the number of white cards in the undrawn stack is not fixed. Asher Vollmer (the coder of this algorithm) must be sitting at home grinning deviously while reading this thread. Sorry for the big black box.
Usually when you delete a game from your device you get asked if you want to remove the gamecentre data with it, then you start a fresh when you reinstall.
