Join 3,514 readers in helping fund MetaFilter (Hide)


The History of Threes
March 27, 2014 10:05 PM   Subscribe

The designers of the hit iOS game "Threes" reveal the complete design process behind it and their thoughts on clones.
posted by empath (118 comments total) 35 users marked this as a favorite

 
Oh my god this damn game. I hate the developers.

Now I will go read the link. (After I tell languagehat about this because he's the one who told be about this game and I hate him for that.)
posted by rtha at 10:09 PM on March 27 [4 favorites]


The 2048 stuff keeps showing up around me and I've been "... but Threes" every time. Nice to see it laid out. Now I'm glad I bought Threes and gave the original developers some scratch.
posted by immlass at 10:15 PM on March 27 [7 favorites]


I'm usually not won over by this sort of thing because everything comes from somewhere and a fun game is a fun game.

But... I've been playing 2048 for a while, and I've gotten up to 1024. But I noticed I wasn't enjoying it much, and it's because I still don't know what I'm doing. I'm getting better but I can't figure out *why.* When my friend sent me this article, it shed some light on why I feel that way. It's a nice confirmation that, in spite of superficially similar clones, hard work or lack thereof really does show itself eventually.
posted by drjimmy11 at 10:21 PM on March 27 [2 favorites]


I enjoy Threes more than 2048, mostly due to the thoughtful game design aspects that add an element of challenge that is lacking in all the clones. It's good to see that the developers made a little money off the original, despite the best efforts of the imitators.
posted by Blazecock Pileon at 10:22 PM on March 27 [8 favorites]


I didn't know they shared skin with Ridiculous Fishing, neato!

It's a bummer that these fellas felt so interrupted by 2048. The work and love game designers like them put into the mobile platform is something I strongly admire, and take pride in supporting.

Buy indie cell phone games y'all. Otherwise the Zynga chodes win.
posted by oceanjesse at 10:26 PM on March 27 [10 favorites]


I bought my mom a copy of Threes because I loved it and thought she'd really like it but she didn't get into it very much and then a couple of weeks later she sent me a link to 2048 and I nearly cried.
posted by jacquilynne at 10:34 PM on March 27 [12 favorites]


Add me to the list of people who paid for THREES! after getting addicted to 2048, and enjoys both games in very different ways. I find it fascinating that 2048 took an already simple game and managed to simplify it even further, while still ending up with a pleasing and viable puzzle.

There's really only one concept 2048 has that THREES! doesn't: an ending. A goal. Asher treats this like a bug ("We wanted players to be able to play Threes over many months, if not years") but for me it was both the hook that motivated me to keep trying, and a promise that I could eventually stop. It made it both more addictive in the short term and less life-consuming in the long term.
posted by mbrubeck at 10:34 PM on March 27 [7 favorites]


2048 is cool in its own right because of how the github community latched onto it, releasing all of those cool AI hacks and multiplayer mods. I love it when things like that happen.
posted by oceanjesse at 10:43 PM on March 27 [5 favorites]


I really enjoyed playing Threes and wrote an e-mail to Asher shortly after their launch asking if they'd be interested in working together to create a Windows Phone port of the game. He was polite in his refusal, and I left it at that. So yeah, it was really interesting reading this post.
posted by Gomez_in_the_South at 10:48 PM on March 27 [1 favorite]


That was very thoughtfully laid out. This is something that's going to keep happening. Hell, for Greg and Ridiculous Fishing it's happened before with a troublesome clone from GameNauts. It's something that understandably causes a huge amount of frustration, especially when your labor of love is cloned in 6 days. These are people who love games and want to make great games and art, and put nuance and craft into it, and it's not so much about money. They're not going to get litigious like some big corporation like Mattel would, and probably can't afford to anyway. But clones are a real problem within the app economy and they always leave a sour taste in my mouth. I don't see any easy solutions, but maybe within gaming culture we need to actually start criticizing this kind of thing and not being so blase about it?
posted by naju at 11:30 PM on March 27 [1 favorite]


The author goes to the Threes took 14 months well repeatedly even stating "We know Threes is a better game, we spent over a year on it." without seeming to realize what a logical fallacy that is. I was yearning to try Threes but I don't have a smart phone and now having read that it is Nintendo hard I won't bother borrowing my wife's phone to try it out.
posted by Mitheral at 11:38 PM on March 27 [1 favorite]


Mitheral, since you can do so freely in the most literal sense of the word, I encourage you to discover what the author is talking about without using "logic."
posted by oceanjesse at 11:45 PM on March 27


Oh yikes, rereading your comment and then mine, I assumed your wife had already paid for and downloaded Threes. Sorry for my tone.
posted by oceanjesse at 11:49 PM on March 27


Somewhat related, Greg Wohlwend's post about marketing Threes is great.
posted by kmz at 12:24 AM on March 28 [1 favorite]


A lot of their year spent on it was paring away bad ideas, not adding good ones. And the end result is amazingly elegant. I'm particularly impressed with how the difficulty progression over the course of a game naturally emerges from their simple rules. The more different values of tiles you have taking up space, the harder it is to make a match.

But that elegant simplicity does make it really easy to clone.

That said, Threes! is a much better game than 2048:

The fact that you are given bigger pieces later in the game keeps it from slowing down or getting too repetitive (to get a 2048 you have to make something like 700 moves)
The fact that pieces come in from the side means you can control where new pieces go. That is essential to deeper level of play in Threes!
The fact that you only move one space per swipe gives you more control, but also introduces more tiles while you're exercising that control.
The characters on the tiles are completely adorable


OMG I LOVE THIS GAME SO MUCH
posted by aubilenon at 1:05 AM on March 28 [1 favorite]


I like 2048 because I can play it faster -- it would be a much better game, difficulty wise, with a clock, like if new tiles came on every second, whether you made a move or not.
posted by empath at 1:15 AM on March 28 [2 favorites]


Related: Clones. Clones everywhere.
posted by effbot at 1:26 AM on March 28


I'm assuming it's one of the clones he's talking about, but I do prefer Fives on Android.
posted by MattWPBS at 2:48 AM on March 28


Based on my own experience with 2048 as well as that of about a dozen people around me, I find the claim that they both beat 2048 on their first tries a bit difficult to believe. Is Threes really such excellent preparation for the other game? Also, isn't there an element of chance involved in 2048, especially after you have a 1024 and a 512 on the board and all those powers of two to keep around?
posted by holist at 2:52 AM on March 28 [1 favorite]


Can someone explain why it seems like it's easier to get big number on 2048 than it is on threes? Is there some different algorithm trickery going on? does no one know?

Because it definitely plays differently. I can crush at 2048 pretty easily, but i struggle to get higher than around 2000-ish as my score on threes.

On another note, is the app store ever going to step up and ban dumb infringey cash grab clones like this? It makes it feel like a cheap knockoff mall kiosk.
posted by emptythought at 3:01 AM on March 28 [1 favorite]


I'm assuming it's one of the clones he's talking about, but I do prefer Fives on Android.

Isn't Fives exactly Threes but starting with 2+3 rather than 1+2 ?

I feel (with Fives at least) that, like Tetris, it's dependent on which numbers you get given, but unlike Tetris, these days I don't trust developers to use a real random number generator to decide this, and not use evil (like making sure you get easy games every so often to keep your hopes up)
posted by doiheartwentyone at 3:19 AM on March 28 [1 favorite]


So I bought Threes based on this claim of being better than 2048, and so far it's quite disappointing in the details. Less smooth, takes a minute to start up, "sign" your game, etc.
posted by dhoe at 3:22 AM on March 28 [3 favorites]


I started with 2048, and am still addicted to it, even though I can only get to 1024, and at that point it gets hard. Threes I bought because I heard all the cries of ripoff. It's good, but I find it much more challenging. My board fills up and it's game over far too quickly. I haven't played it enough, I suppose, to "get" it.
posted by zardoz at 3:24 AM on March 28


The problem with 2048 is that there is a ridiculously easy strategy for beating it: completely avoid using one of the direction keys, and use one direction on the other axis only when you have to. Say, skip up completely, and only press right if there is no down or left move. Larger values will automatically collect around the bottom left corner, eventually leading to 2048.

Once I noticed the trend, it took me about three tries to get to 2048: mostly because I was absentmindedly pressing the up key late in the game, or pressing right when not strictly necessary and allowing a new 2 to block the bottom left corner.
posted by Dr Dracator at 3:36 AM on March 28 [1 favorite]


2048's strategy is basically trivial. I can beat it 60% of the time, and I could probably do it more often if i was a bit more thoughtful with it.

Threes is different because there are a greater variety of tiles being added, and you don't collapse all the way over-- in 2048 if you collapse tiles a few rows apart, only one tile gets added-- not the case for threes.

I actually found that playing 2048, because the strategy is so obvious, made me better at threes. My top score at Threes went up from about 7,000 to 25,000 after a couple of days spent figuring out 2048.
posted by empath at 3:41 AM on March 28


Once I noticed the trend, it took me about three tries to get to 2048: mostly because I was absentmindedly pressing the up key late in the game, or pressing right when not strictly necessary and allowing a new 2 to block the bottom left corner.

You can do better if you lock the bottom row (or even the bottom two rows) instead of not pressing right. Gives you a lot more freedom.
posted by empath at 3:42 AM on March 28


There is a game design problem with threes though -- it does sometimes give you basically unwinnable boards because of the tiles it gives you. I've had games where I get 4-5 blue tiles in a row and no red tiles, which basically is an auto-loss. They should have some kind of algorithm to keep the blue and red tiles on the board roughly even.
posted by empath at 3:45 AM on March 28 [1 favorite]


Yeah, I'm a Threes fan, but I can see where it could turn some people off. This is going to sound weird but in some ways it's *too* good. By that I mean if you play fast you're not going to get those high scores, my best game on Threes probably took place over a span of 5 or 6 hours and involved me putting down my phone several times, doing other things and then returning to it.

That is not what a lot of people look for in their mobile games, sometimes you just want that quick 5-6 minute gameplay hit while on the subway, but it turns out that Threes starts off that way and then changes its skin as you get better at it.

That can be frustrating, I know some people who loved the game for the first few weeks but then hit a wall and have moved on to other things. I've come close to that point a few times, I'm not much of a "math guy" but I do enjoy discovering and analyzing patterns and that's what keeps me playing.

BTW, noone has said much about the emails section of this article. It's pretty cool to see a breakdown of the creative process that goes into developing a game.
posted by jeremias at 3:47 AM on March 28 [4 favorites]


Threes, meet Betamax. It came first, too.
posted by obiwanwasabi at 4:06 AM on March 28 [3 favorites]


I haven't actually played Threes - I gather I'd need to buy it for a smartphone to do so (and the google play store is claiming it's incompatible with my Xperia SP now that I look - struggling to understand why that would be...)? I certainly can't find it online anywhere, which would pretty much be the only way I would normally be convinced to spend money on a mobile game - some kind of try-before-you-buy. I guess I'm not the target consumer for them - but I have played 2048.

"All tolled, we sent 570 emails in this email thread."

Oh god, can't tell if intentional or not, this is making me twitch so badly.
posted by Dysk at 4:07 AM on March 28 [1 favorite]


Browser version of threes.
posted by empath at 4:23 AM on March 28 [4 favorites]


is the app store ever going to step up and ban dumb infringey cash grab clones like this? It makes it feel like a cheap knockoff mall kiosk

Seems like it would be impossible to create a working definition of what a clone is without screwing over someone or other. What if it's exactly the same except for one critical difference? Or two? When does it stop being a clone and start to become something else? What if the clone is better than the original? Should that not be allowed?
posted by ultraviolet catastrophe at 4:24 AM on March 28 [5 favorites]


"All tolled, we sent 570 emails in this email thread."

Oh god, can't tell if intentional or not, this is making me twitch so badly.


Same with "That old chess-nut…" Do they mean an eggcorn?
posted by ultraviolet catastrophe at 4:28 AM on March 28 [4 favorites]


Thanks empath! I do notice that is itself a (unauthorised?) clone, which I find slightly amusing...
posted by Dysk at 4:36 AM on March 28


"That old chess-nut…"

This, especially with the hyphen and everything, struck me as more of a pun than an unintentional eggcorn. The "all tolled" and "pouring over" not so much, but the presence of "chess-nut" is what's making me so twitchily unsure.
posted by Dysk at 4:40 AM on March 28 [2 favorites]


Threes is also available on Android.

I honestly find I prefer 2048, though. Or Hexagonal 2048. It's not just about reaching 2048; you can keep going. I got an 8192 once.
posted by Foosnark at 4:56 AM on March 28


There is a game design problem with threes though -- it does sometimes give you basically unwinnable boards because of the tiles it gives you. I've had games where I get 4-5 blue tiles in a row and no red tiles, which basically is an auto-loss. They should have some kind of algorithm to keep the blue and red tiles on the board roughly even.

There's a surprisingly in-depth toucharcade thread with a whole lot of details that people figured out about how the draw algorithm works, which is not entirely as random as you think. I do agree that it can give you unwinnable boards, but there is some logic in how it's distributing tiles.
posted by zempf at 5:47 AM on March 28


The "all tolled" and "pouring over" not so much, but the presence of "chess-nut" is what's making me so twitchily unsure.

Could we drop this derail? Damn pre-Madonnas.
posted by yerfatma at 5:49 AM on March 28 [20 favorites]


I am not a developer, and I do not work directly with the Apple Store, but I do work with people in a company that has submitted apps to the Apple Store, and I'm having a hard time with the "six days to clone" figure. It takes about a week for the Apple Store to review and approve a submission for the store, so to have a clone of a game out in less than a week, the clone would have had to have been conceived, designed, programmed and submitted within, at most, 48 hours of Threes hitting the store.

Which isn't impossible, I guess -- and I don't want to blame the victim here -- but was Threes really cloned and submitted that quickly?
posted by Shepherd at 5:53 AM on March 28


I've been working on my own "deceptively simple" abstract puzzle game for the last year. I'm worried that I might be walking into something similar: the carefully balanced difficulties that tickle my design sense will get filed off in a lookalike clone with a larger win button. What's disturbing for me in this case is that the "less interesting" clones may in some cases be meeting the needs of the audience better. And as a striving game designer, this is kind of depressing.
posted by GameDesignerBen at 5:54 AM on March 28 [1 favorite]


"less interesting" clones may in some cases be meeting the needs of the audience better.

In the case of Threes vs 2048, Threes will often sit there loading for about 30 seconds on my otherwise quite speedy Nexus 5. The version of 2048 I run is always instantaneous.

Also, when I first ran Threes it played music and made noise that I could not turn off because the developers so thoughtfully disabled the settings menu during the tutorial. On a mobile game, the VERY FIRST THING one often wants to do is kill the sound.

As far as actual design goes, 2048 is never frustrating and sometimes that's better than being challenged -- more gratifying, even though less "interesting".
posted by Foosnark at 6:04 AM on March 28 [2 favorites]


The best thing about the Threes discussion is finding the most perfect game ever invented: 2048 Numberwang.
posted by viggorlijah at 6:06 AM on March 28 [11 favorites]


Threes is just poorly implemented, sorry. I sympathize with its creators, but the Android version at least certainly doesn't feel like polished work.
posted by dhoe at 6:08 AM on March 28 [1 favorite]


finding the most perfect game ever invented: 2048 Numberwang.

That is just the best thing ever.
posted by dabug at 6:21 AM on March 28


I was tempted to make a post this morning about the wonderful Doctor Who version of 2048, but I'll just put it here instead.
posted by jbickers at 6:24 AM on March 28 [2 favorites]


This is why I stopped buying iOS and Android games and bought a damned PS Vita. It costs more, sure, but when I go in the Vita store thing I know that every game there is different, and not a crap clone of something else. It also doesn't support any pay-to-play shenanigans.

For all the complaining about Apple App Store's complex rules for entry, all you basically need to do is not hack into any libraries you're not supposed to. Your app can be complete garbage, and still get in.

If you want to get into Sony's Vita store, you need to have some meetings with very important people and convince them that your game is actually fun and not a money-grab or rip-off.

Sometimes having a giant meddling corporation ruthless controlling content delivery can help both developers and consumers. Neato.
posted by sixohsix at 6:28 AM on March 28 [2 favorites]


Shepherd, the six-day clone was "in a different market", which I take to mean not the App Store.
posted by nicwolff at 6:35 AM on March 28 [1 favorite]


I can't speak to the Android port, but I play Threes on an iPhone 5c and it is a delight.

I mean, a merciless destroyer of productivity, but delightfully so.
posted by Narrative Priorities at 6:39 AM on March 28


Threes is fantastic—just the thing for making subway trips vanish. There's something deeply rewarding about bringing a board back from the edge of ruin and into viability with a few careful moves. "Holy shit!" I'll find myself saying after narrowly avoiding Game Over.

The fact that these moments are fairly common is a testament to the game's deceptively careful design, I should think.

And of course, it's adorable.
posted by Sokka shot first at 6:51 AM on March 28 [3 favorites]


Isn't Fives exactly Threes but starting with 2+3 rather than 1+2 ?

Yup, but also without the cutesy "this is Mr Bobblethwaite, he lives fudge!" characterisation of the different numbers. That just rubbed me up wrong in Threes.
posted by MattWPBS at 6:52 AM on March 28


Threes on Android is just not at all a smooth experience. Maybe the iPhone version is better implemented. I was just trying it again now and it managed to forget I had gone through the damned tutorial.

I feel like the Android version was an "oh yeah, that is a thing that exists which people use too!" afterthought and it was rushed because of clones and whatnot.

2048 is like popping bubble wrap. It's just such a pure experience of combining and distilling stuff. There are no obstacles, it's just the open road. Threes does not feel like that to me, with 1s and/or 2s always getting in the way (leaving aside the cutesy stuff and the poor performance and bugs).
posted by Foosnark at 7:02 AM on March 28 [1 favorite]


I will admit my bias up front - I'm lucky to know Greg Wohlwend. But I looooove Threes. I really don't play a lot of phone games normally but Threes (and Ridiculous Fishing, and to a lesser extent Hundreds) I keep coming back to. I love how after a solid 30 minute session of threes on the train commute home, I can look at 4 lanes of car traffic and see 2 white cars in a row and want to smoosh them together. I always love how puzzle games can get in your head like that.

I also like that you can play Threes the way you play 2048 - fast and instinctive, swiping quickly without planning more than a couple moves ahead, and you'll regularly get 2,000-4,000 points. But you can also play it as a deep strategy game and plan your moves way in advance to get the higher scores. My highest score is about 9,000, which isn't that great but I haven't been able to replicate it yet - mostly because I tend towards the fast and instinctive style of play.

I like the sound effects - when I take a while thinking about a move and it says "boooo-ring" I almost always say "shut up!" back at it, and then laugh at myself.

Threes is slow to load for me on my iPhone 4S and can be a little laggy sometimes. But I blame that on my 2 year old phone plus the requirement that games on the iPhones integrate with that STUPID gamecenter.
posted by misskaz at 7:09 AM on March 28 [1 favorite]


The music in Threes has some kind of magic ear-worm quality. I downloaded the game last Friday and the tune has been playing in my head ever since. Normally it's pretty easy for me to sing Mary Had a Little Lamb or the ABCs and get rid of something like this but I am not having luck. Love the game, but I turned off the tunes.
posted by morganannie at 7:23 AM on March 28


Amazing article, so much fun to read someone's development process. Thanks!

I think it's funny that the article talks so much about clones, but never once mentions how Threes has a UI heavily inspired by Loren Brichter's Letterpress. The simple flat graphics, the feel of the interaction with the pieces, even the flow of the app are all borrowed. Nothing at all wrong with that, both are beautiful iOS apps. (Sorry Android people.) But we all stand on the shoulders of giants, as they say, and I'm surprised they didn't acknowledge Loren's.

The way 2048 has eclipsed Threes is interesting. I think 2048 is a success largely because it's easy, because it's not a very good puzzle. People want to feel smart, they want to beat the game, and 2048 certainly is generous to casual play. It helps that it pretends to be a math game (it's not really), making people feel smart at math is a sure-fire winner. It's also important that 2048 is free and browser-based, way easier to share and try quickly. It's also beautifully implemented, much like Threes is, and so it's approachable. As the Threes guys say, 2048 is not a game one can play for months. But that doesn't mean it's not a success in its own little viral creampuff way.
posted by Nelson at 7:25 AM on March 28 [3 favorites]


A little more context on "chess-nut":

You know, “simple to learn, impossible to master”. That old chess-nut…

Given the context, I'm pretty sure "chess-nut" was intentional.
posted by DevilsAdvocate at 7:42 AM on March 28 [2 favorites]


As a massive fan of Threes, that was a great read. I had no idea it took so long to develop, but what really amazed me is how complex and awful the colors/graphics in the early prototypes are when you're really familiar with the final simplistic game. I mean, I know in my head that simplicity takes lots of time to refine and take away all the unnecessary cruft, but it's amazing to go back through their process and see how they got to the final point.

I loved Threes since its debut, and I thought the 2048 clones are junk because of the automated mechanics. Playing a 2048 game really feels like just pressing up and left into corners until you win. Sometimes I can extend a Threes game for ten minutes making last-ditch efforts to build up connections.

The straight-up clones of it are disappointing, I accidentally found a crappy version based on 2's instead of the number three within days of the release of Threes (which I tweeted at Asher, who was disappointed as well, not sure if I was The Guy That Found The First Threes iOS Clone or not though).

After a month or so of playing it for many hours (high score is 27k something, still trying to get that elusive 1536 card), I think the only change I'd like to see is it having a "dark mode" that was inversed, with black background and cards, so I can play it before bed for 10-15min without the blinding white light illuminating my bedroom. White in the day, dark at night would make this game even more perfect.
posted by mathowie at 8:29 AM on March 28 [4 favorites]


Browser version of threes.

This probably proves how much forethought went into Threes in that I find the browser version roughly animated, seemingly arbitrary in mechanics, and needlessly cruel on how next cards fall (in five quick games I can barely muster 1,000+ points where on iOS I can get 3k almost every time).

It's amazing how it can be exactly the same, but feel completely different just based on a few tiny twists of random variables.
posted by mathowie at 8:37 AM on March 28


I find it interesting to see how much diversity in opinion there is regarding which game is superior. I had heard of 2048, but never did shit with it until I saw the doge2048 post here, then got hooked, then tried the original 2048 and downloaded one of the versions on my Nexus. I had seen Threes in the store, but it didn't look appealing (nor did Five). After reading this post, I do want to give it a try, but I have a feeling I might be in the 2048 camp. That said, I love the style of Threes, and I love reading the emails.

Reading the emails fascinates me because of the casualness of it. I got to test Tic-Tactics and I had tried to be very detailed in my reports but I know at times I was somewhat casual, and I was worried my casualness in parts might be bad feedback, and then to see these guys here have such a casual approach to discussing things...

Well I'm making a game myself (in fits and starts and really just terrible at programming as it is, but...), my friend gives me feedback and it is a LOT like this, and it's nice to see that there isn't some magical barrier in terms of how "the pros" work vs lil' old me. I'm also testing a game for a friend of mine that I can't wait to see it come to fruition, but I can't say anything else right now as it's very early in dev.

I wish programming games were easier. I wish I had a more natural skill at it. I wish I had a drive. I get so inspired when I see these sorts of things.

If you ever want to make a game and you feel defeated, instead of inspired, at seeing something so amazing, just look at all the shit out on the app store, the clones, the dumb fucking games that suck so hard, and say to yourself "I CAN DO BETTER THAN THIS BULLSHIT! FUCK YOU, I WILL!" and make yourself RAGEDEV at the utter tripe that's out there to prove you're better than that.
posted by symbioid at 8:58 AM on March 28 [1 favorite]


Developers love the app store because of the low barriers of entry. You can get your game on the store quickly and cheaply.

Developers hate the app store because of the low barriers of entry. Others can get their games on the store quickly and cheaply.
posted by 2bucksplus at 9:01 AM on March 28 [8 favorites]


I don't understand people who don't like Threes. I hardly ever play such games and I'm addicted to Threes. That's all I have time for now, I have to go play Threes.

*gives rtha a noogie*
posted by languagehat at 9:03 AM on March 28 [1 favorite]


There's really only one concept 2048 has that THREES! doesn't: an ending. A goal.

I've been thinking the same thing lately -- I kind of miss games that I can win instead of just endlessly reaching for a higher score or completely an ever-expanding number of levels. I'm sure they're still out there, but in the casual gaming world I don't seem to find them. I have a few two-player games, which allow to play against the computer, that are winnable, but they're a) mostly based on physical games and b) plagued by AIs that are too easily beaten. In order for winning a game to seem fun, I also have to sometimes lose.

Based on my own experience with 2048 as well as that of about a dozen people around me, I find the claim that they both beat 2048 on their first tries a bit difficult to believe.

I don't find it all that hard to believe. My first try was completely half-assed, mostly random button pressing, and I got a 1024 and then I easily got a 2048 on my second try as soon as I started thinking even a little bit about how to manage the board. I can fully imagine that anybody who stopped and thought about the game for a minute before they totally mangled their first board (by, say, the point where they had a 256 or so) could easily have beat it first try.
posted by jacquilynne at 9:23 AM on March 28


2048 succeeded where Threes "failed" because the mobile environment is all about free to play and short play sessions. People don't want to pay for content, and they want to be able to consume their free content in short bursts. Words With Friends, Candy Crush, Draw Something (for bad drawers like me), Dots, Flappy Bird. They all fit this profile. I don't know of any paid game that went viral to the extent of any of those games. I haven't played Threes, but if it actually achieves its goal of easy to play, hard to master, then it's done itself a disservice in terms of adoption. You're not going to see Go go viral, either. The market does not want that.

If you're going to make a game that you are proud of hitting all your game design principles, do it. But don't complain when it doesn't take over the world, because in all likelihood, you didn't make something the market wanted.
posted by shen1138 at 9:30 AM on March 28 [1 favorite]


Is Threes a better game? We think so. To this day, only about 6 people in the world have ever seen a 6144 and nobody in the world has yet to “beat” Threes. But that’s what’s better to us as game designers.

I suppose this means that what's good for them as game designers is directly at odds with what's good for game players. (The only way to win is not to play...)
posted by Shmuel510 at 9:32 AM on March 28


Do puzzle games typically have a "win" state? I don't mind that there is no way to truly beat Threes, because I just play it as a little distraction and brain-teaser while I'm on a commute or need a break from thinking about work stuff. Not every gamer in the world is goal-oriented, especially for mobile games I would think. I get frustrated sometimes when I see myself screwing up a board, and that I haven't broken the 5-figure score barrier yet, but it's still enjoyable while I'm playing it.

To each his/her own though.
posted by misskaz at 9:36 AM on March 28


I have noticed that people who *really* like games tend to like hard games — because they play a lot and because they tend to get into strategy. More casual folks, though, don't. It reminds me of an ex I could never play games with (other than d&d) because I prefer to save my hard thinking for other things in life and look to games for the more mindless side of recreation. And he wanted strategy. And thus there is Threes on one hand and doge2048 on the other.
posted by dame at 9:48 AM on March 28 [2 favorites]


You can talk all you want about difficulty and the puzzle, but from where I'm standing Threes is a wildly better game for one reason alone: the music, sound, and personality from the tiles. It gives the whole thing a sense of polish and depth that I find 2048 lacks.

Also, yeah, 2048 is waaaaay too easy.
posted by Itaxpica at 9:56 AM on March 28


I suppose this means that what's good for them as game designers is directly at odds with what's good for game players. (The only way to win is not to play...)

I don't know. The fact that you can't win Tetris or Bewjeweled doesn't make them bad for game players. Though I think what they really mean by "nobody has beat it" is that nobody has advanced beyond the levels for which they made art.
posted by aubilenon at 10:05 AM on March 28


There's really only one concept 2048 has that THREES! doesn't: an ending. A goal.

This is the exact reason I found Nanobot Builder to be so much more enjoyable than Cookie Clicker or any of the other incremental games - you can beat it.
posted by jbickers at 10:08 AM on March 28


I would loooooooooooooove to see a link to the automated script that wins 1% of the time.
posted by zscore at 10:14 AM on March 28


Oh geeze, a terrible article from the LA Times about 2048. And now I have a pretty bad opinion of Cirulli too.
posted by kmz at 10:15 AM on March 28


This one has 'Get Hint' and 'Auto-Run' buttons.
posted by box at 10:16 AM on March 28


I have noticed that people who *really* like games tend to like hard games — because they play a lot and because they tend to get into strategy. More casual folks, though, don't. It reminds me of an ex I could never play games with (other than d&d) because I prefer to save my hard thinking for other things in life and look to games for the more mindless side of recreation. And he wanted strategy. And thus there is Threes on one hand and doge2048 on the other.

Weirdly, Threes seems to serve both things for me. Sometimes I just want to kind of mindlessly speed-play and see how far I get, and it's good for that (for me), and sometimes I want something that makes me think and plan and plot, and it does that, too. That's why I think it's evil!
posted by rtha at 10:16 AM on March 28 [3 favorites]


Hey box, I meant a link to a script that iterates the sequence they outlined in the article and wins 1% of the time. Because this suggests otherwise. I think they're misleading/lying in the original article.
posted by zscore at 10:20 AM on March 28


Hey box, I meant a link to a script that iterates the sequence they outlined in the article and wins 1% of the time. Because this suggests otherwise. I think they're misleading/lying in the original article.

You're parsing the original article wrong, the every hundredth time is referring to moving left. They only say that the very simple algorithm can beat 2048, but in fact I'd bet it would be a much higher winning percentage than 1%.

And your link is irrelevant because it's talking about random moves. Automated != random.
posted by kmz at 10:31 AM on March 28


I'd probably say something like 'exaggerating,' but, yeah, I agree. The strategy they outline works okay until the board starts filling up, but it falls short at that point.
posted by box at 10:31 AM on March 28


Sometimes I just want to kind of mindlessly speed-play and see how far I get, and it's good for that (for me), and sometimes I want something that makes me think and plan and plot, and it does that, too. That's why I think it's evil!

The worst is when I'm playing fast and loose then suddenly have some big numbers on the board and think O Crap I Shouldn't Waste This Opportunity and then I'm taking 2 minutes to plan each move and I REALLY should be asleep. Argh!
posted by wemayfreeze at 10:32 AM on March 28 [1 favorite]


(I just tried the very simple algorithm by hand--it was good for a score of 2600, with a 256 as the biggest tile. I do not plan to try it 9,999 more times.)
posted by box at 10:33 AM on March 28


('Okay, once more,' I told myself. Score of 892, biggest tile a 64.)

I knew about Threes before I knew about 2048. I'm playing 2048 now, and it's nice to know that, when I get tired of 2048, Threes will still be there.
posted by box at 10:34 AM on March 28


Can we talk Threes strategy? Corner generally seems best but you can get really screwed when you run out of swipes in your two directions. It seems a system where you keep your highest tiles in the middle of one wall could mitigate that but I haven't found much success with that yet ... Thoughts?
posted by wemayfreeze at 10:35 AM on March 28


kmz, gotcha, I understand what they're saying now. At the bottom of the github there's a little thing where they talk about a simple cycling algorithm that shifts the maximum tile distribution one to the right. (which is what I thought they were talking about).

But I do think it's pretty important to talk about expected win percentage. Maybe the average player will invest 2 hours to realize what a decent (but dumb) strategy is, and then it will take them four hours on average to win. That's a pretty reasonable time course for a game nowadays.
posted by zscore at 10:40 AM on March 28


I think it's interesting that, while I have only just glanced over the web version of Threes, which may have differences, actually in my experience the two definitely have different strategies, and 2048 is not necessarily easier. Yes, you can win games using a simple rule, but you'll still lose a lot of them that way. Starting from that point and adopting more complicated plans of consolidating larger blocks and trying to build chains to be able to clear larger parts of the board at once, I'm able to win much more regularly. But then there's the temptation to click that "keep going"... I have not yet managed a 4096, but I've gotten as far as a bottom row with a 2048 and a 1024, which felt pretty good. At that point, it is definitely not a too-easy game.

It is however one that I really need to put down.
posted by Sequence at 10:42 AM on March 28


Can we talk Threes strategy?

My high score is around 30k and I don't really use either a "corner" or "middle" strategy, I try to use a "don't get stuck" strategy. That really boils down to thinking of the game from two angles:
  1. Hyperfocus in *only* on setting up your 1s and 2s for combo
  2. Hyperfocus on finding a thread of 3 or 4 tiles to get a big number
The trick is knowing when to alternate between the two.
posted by jeremias at 10:47 AM on March 28 [1 favorite]


There's some interesting discussion going on in my twitter feed between a bunch of indie developers w/r/t the impact of Free-and-Open-Source-Software and this whole 2048 vs Threes thing. It's interesting that 2048 first blew up on HackerNews.
posted by GameDesignerBen at 11:06 AM on March 28


My high score is around 30k

*vibrates with envy and hatred*

and I don't really use either a "corner" or "middle" strategy, I try to use a "don't get stuck" strategy.

Okay FINE I'll try that.
posted by rtha at 11:11 AM on March 28


I'll give that a shot jeremias!

FWIW my high score is just shy of 30K and I've been a corner man my whole life.
posted by wemayfreeze at 11:18 AM on March 28



Can someone explain why it seems like it's easier to get big number on 2048 than it is on threes? Is there some different algorithm trickery going on? does no one know?


It's because 2048 always gives you a 2 or 4 card, but Threes gives you a 1, 2, 3, or (occasionally, but more importantly) a higher card. Since 2048 always gives you low cards, you can just keep the board sorted by ONLY moving in two directions. If you do that, when you make a 256 it will be very close to your existing 256. In late game Threes!, you will occasionally be given a 96, which is unlikely to be close to its match even if that's on the board. So you have to do some work to get them together. I am not sure that would even be viable with 2048's movement mechanism!

Can we talk Threes strategy? Corner generally seems best but you can get really screwed when you run out of swipes in your two directions. It seems a system where you keep your highest tiles in the middle of one wall could mitigate that but I haven't found much success with that yet ... Thoughts?

If you have a 768 on the middle of an edge your danger is that you'll get a 384 on either side of it, and then it will be difficult to get them together or to make a third 384 (because your board is so full of 786's and 384's). It may be possible to manage this, but it is not trivial. It seems like it might be viable to put the 786 one in, and work on the first 384 in the corner, and then work on the next one above that, but then you're basically sacrificing a whole row of workspace, which space you will need to buffer temporary imbalances of 1's and 2's.

Early game, the corner strategy works well, because the new cards are limited by your highest value (they can be something like 3 ranks less or lower). But once you start getting 48's and 96's you have to do some work and take some risks to keep them near other big tiles. If I can do so without making too bad a mess, I'll try to bring those in on one of my high value walls. If you can make your second row match your edge row, sliding over and merging them is great in terms of opening up space on the board, but that has the downside of putting all your big numbers in one row, which is tough. But the corner strategy in general involves having a couple copies of several numbers, which once there's a big range of cards in play seriously cramps the work area, which means you can go from doing great to game-over very quickly.

I haven't ever really tried it yet, but I kind of wonder if it makes sense to shift over into a boustrophedon arrangement at some point. Deciding when to make that sort of transition is always really tricky, and I'm not even sure it's a great idea. To get a 6k tile you'd need to manage 17 different values on a 16 space board, so if I ever want to get there, I suspect I'll have to switch to something radically different from what I'm already doing. And have a run of good luck.

My record is ~70K.

My other record's 30. Having read that post on the new card algorithm I suppose the lowest one could go would be 12, but that would require some very improbable draws.
posted by aubilenon at 11:23 AM on March 28 [1 favorite]


I have noticed that people who *really* like games tend to like hard games — because they play a lot and because they tend to get into strategy.

I really like games, but this is only the case up to a certain point and only with certain kinds of games. I will plan out characters in MMOs but I don't necessarily want the difficulty turned up as much as some hardcore types do. Racing and action games need to achieve just the right balance of hard and winnable.


Anyway, I think it's worth considering that 2048 isn't necessarily "easy" -- if you ignore 2048 as a goal and instead concentrate on beating your previous best (or friends' best) scores. Instead, it seems to me that Threes is typically shorter, slower, and more subject to random punishment and random reward than 2048.


from where I'm standing Threes is a wildly better game for one reason alone: the music, sound, and personality from the tiles. It gives the whole thing a sense of polish and depth that I find 2048 lacks.

I don't want my phone chirping at me while I'm sneaking in a game on the toilet or in public or whatever. Music and sound get turned off immediately for all mobile games. The little animated personality things were cute the first time but really just strikes me as completely unnecessary and irrelevant to the game. And given that Threes runs significantly less smoothly than 2048 on my phone, those extra touches have made themselves unwelcome.

That's not depth any more than the aluminum spoiler bolted to the neighbor kid's Civic is depth.
posted by Foosnark at 11:58 AM on March 28 [1 favorite]


I have noticed that people who *really* like games tend to like hard games — because they play a lot and because they tend to get into strategy.

I have a friend who does really like games, but he actively dislikes games being difficult. For some games (e.g., super meat boy) that means there's no point, but there's a lot of games (e.g., Rogue Legacy) he manages to fully enjoy by editing saves or patching the binary or whatever.

As someone who does enjoy challenges in games, I have surprisingly complicated feelings about this. What's wrong with cheating at solitaire? My gut says that something is, but my brain sure can't think of what it could possibly be.

(He doesn't do this in games with leaderboards, with the possible exception of ProgressQuest.)
posted by aubilenon at 12:35 PM on March 28


I really like doge2048. I'll have to check out this "threes" game
posted by rebent at 1:11 PM on March 28


My other record's 30. Having read that post on the new card algorithm I suppose the lowest one could go would be 12, but that would require some very improbable draws.

I'm pretty sure someone on my Twitter stream finished with a zero or a 3. Let me find it.
posted by mathowie at 1:25 PM on March 28


Here it is, I think this is a score of 21 given the combos.
posted by mathowie at 1:29 PM on March 28


Wow, those low scores are really hard... I can't seem to avoid combining a couple of 3s.

And now I have that tune stuck in my head again. Agh.
posted by RedOrGreen at 2:12 PM on March 28


mathowie: Ones and twos are worth nothing and every other card is worth 3 ^ (log2(value / 3) + 1) points. So each 3 is worth 3 and that screenshot ought to be worth 24. Unless it doesn't count the card that ended it?

No I just checked, it does count the card that ends it. Not sure why it's not red, though.

(Edit: oops, wrote the wrong formula. fixed.)
posted by aubilenon at 2:41 PM on March 28


"The FOSS Community has grown complacent about the dangers of irresponsible distribution. When you give something away for free, you need to assume some responsibility for what people do with it. Just because you are a nice person doesn’t mean everyone is, and closing your eyes to that fact doesn’t absolve you of it’s consequences."

SpellTower and Ridiculous Fishing developer Zach Gage on the problems of Open Source and the success of 2048.
posted by GameDesignerBen at 3:07 PM on March 28


I had played 2048 before, but not Threes. So on my android phone I went to the Play Store, found the official version was $1.99 (not even a trial version) and downloaded one of the free clones instead.

Didn't see what the big deal was...not that much different from 2048, I thought. Until I read the article here in depth, and realized the clone was not an exact clone. The clone was more like 2048, in that it spawned its new tiles in random spaces (not just on the side swiped from) and moved tiles all the way over (rather than just one space).

OK, I thought, the game appears to be well-recommended, and it's not like I can't afford two bucks. Bought the real thing. And discovered what Foosnark had discovered earlier:

Also, when I first ran Threes it played music and made noise that I could not turn off because the developers so thoughtfully disabled the settings menu during the tutorial. On a mobile game, the VERY FIRST THING one often wants to do is kill the sound.

Except I'm even less tolerant than Foosnark, given that I keep my phone on "vibrate" 99% of the time anyway. And if the phone is on vibrate, the only apps that get to make noise are those whose main purpose is to make sound and which I would not even open unless I wanted sound—the music player and the podcast player. IF SOUND IS NOT FUNDAMENTAL TO WHAT THE APP DOES, IT DAMN WELL BETTER NOT BE MAKING NOISE WHEN THE PHONE IS ON VIBRATE OR SILENT. I shouldn't even have to actively change the settings on a game, even the very first time, to disable sound if the phone is already on vibrate.

So I start it up, it goes into its tutorial, with sound. I double-check to make sure I didn't accidentally take the phone off of vibrate. I didn't. Play Store. Refund. Get the "Are you sure? The app will be deleted" message, and, dissatisfied with the lack of an "OH HELL YES GET THIS OFF MY PHONE NOW" button, I must content myself with a mere "Yes."

That said, the browser version empath linked to above seems to be much closer to the original, if not exactly the same, and playing that a bit I can see how it's a much deeper and more interesting puzzle than 2048. Might play with that a bit more over the weekend.
posted by DevilsAdvocate at 3:44 PM on March 28 [1 favorite]


THis was a timely post for me because I finally beat doge2048 yesterday, after about a week (which I am glad I don't know how many hours it actually took) of trying, and the only reason I was playing it in the first place was that it was pretty easy and I knew you could win (because of the people in the original thread who beat it within the first few posts.) So now hopefully I can never play it again, because I am hellishly bad at computer games but really susceptible to addictive behavior with them and so now I know enough to NEVER TRY PLAYING THREES, so, woohoo!

I realize that's not very interesting but I feel that - for very good reasons, really - people who are INCREDIBLY TERRIBLE at games are under-represented in these discussions, so, data point or something.
posted by hap_hazard at 3:51 PM on March 28


DevilsAdvocate: Huh. The iOS version does mute the sound when the ringer switch is off. I wonder why they don't also do that on Android.

(As a result I didn't even realize the cards talk until I'd played quite a few hours.)
posted by aubilenon at 3:57 PM on March 28


Thank god there is a minor Intolerable issue with the $2 version so people can feel justified playing the free clones instead.
posted by kiltedtaco at 5:35 PM on March 28 [3 favorites]


Sorry Threes, 2048 didn't just copy but actually improved on you, and in the hardest way possible — by simplifying. You may be a better puzzle, but 2048 is a better game — it plays faster, almost like an action game, and is beatable.
posted by Tom-B at 5:38 PM on March 28 [2 favorites]


Seems like it would be impossible to create a working definition of what a clone is without screwing over someone or other. What if it's exactly the same except for one critical difference? Or two? When does it stop being a clone and start to become something else? What if the clone is better than the original? Should that not be allowed?

know it when i see it.

i would much rather 10 people with good ideas that were "too close" to existing ideas got shut out than 10,000 bullshit clones and stuff made the app store unusable, and made it really hard to recommend games to friends because they'll just see a ton of stupid clones that even get to use similar or the same names with other words added or punctuation or whatever.

I would be perfectly fine with them erring on the side of "no". If they're going to have as many silly and honestly kinda assholey rules for the app store as they do, can they just buckle down and actually curate this shit?

At the very least they need a better search that puts this kind of shit way at the bottom.
posted by emptythought at 5:45 PM on March 28 [1 favorite]


Here's the post mortem writeup on the music in threes.
posted by garlic at 6:51 PM on March 28 [1 favorite]


Threes and the dark side of "open"
posted by garlic at 6:54 PM on March 28


I love Threes, and I was all ready to jump to the dev's defense. But you know what, I think the story played out exactly as it should have really: The 2048 devs should have acknowledged their inspiration of course, but I'm not finding myself that angry with them. The end result is that we're all richer. I appreciate that the devs express similar nuanced feelings in the FPP. It was also really neat to see the design process!

Fives on the other hand should be shut down. That's just a straight rip off.
posted by Popular Ethics at 7:31 PM on March 28 [2 favorites]


people who are INCREDIBLY TERRIBLE at games are under-represented in these discussions

I'm totally not a gamer, I hate first person shooters, I'm terrible at most xbox games, not into tower defense things either. Plants vs. Zombies and Angry Birds seemed pretty difficult to me and I gave up on both within a day. But I like simple casual games that can hold my attention for more than a week of binge playing in my free time.

For the almost seven years I've used an iPhone, I've played three games incessantly, constantly, and throughout. One of them is a paid version of Solitaire that is close to perfect for giving me a game I can finish in 3-5min and play anywhere I have time to kill. It's similar to the classic Windows 3.0 version but has a bunch of touch shortcuts to make gameplay faster (double-click to add to decks). The other is Tiny Wings, the jumping bird/timing game that I played for several months straight but once I finished 90% of the achievements I kind of gave up on it and revisit it once every six months or so for a few weeks at a time. The last one is Threes. I suspect five years from now, if you ask me what I played for ten minutes before bed last night I'll tell you I played Threes. It's simple but something I will return to for a very long time and I doubt I ever get tired of it because it's basically impossible to master it and there's no end to the game.
posted by mathowie at 7:54 PM on March 28 [5 favorites]


If your idea is so easily clone-able, perhaps it's not such a great idea to begin with?

We know Threes is a better game, we spent over a year on it.

This sounds dumb and smacks of sour grapes.
posted by xmutex at 11:08 PM on March 28


If your idea is so easily clone-able, perhaps it's not such a great idea to begin with?

How long does it take to paint a copy of the Mona Lisa or Xerox a copy of Lord of the rings?
posted by empath at 5:02 AM on March 29 [3 favorites]


Space Invaders begat Galaga, etc.
posted by Tom-B at 12:55 PM on March 29


Relevant article (somewhere between opinion and analysis).
posted by nobody at 9:27 AM on April 1 [1 favorite]


Klepek just wrote a pretty good article about the clones with interviews with the 1024 and 2048 guys. And wow, I didn't realize the 2048 in the App Store is a clone of the web 2048 with no connections to Cirulli.
posted by kmz at 11:38 AM on April 3 [1 favorite]


Both of these last two links are worth reading. The gamasutra one is especially good and has some great discussion in the comments.
posted by wemayfreeze at 8:11 PM on April 3


Started playing Threes a couple of days ago. It's interesting, don't know if I'll have a good opinion on it before the thread closes though.
posted by JHarris at 2:05 AM on April 26


I just had a very good session, and I have a couple of suspicions about the tile generation, things that are not obvious on first play. I hesitate to mention them here because they'd be spoilers....
posted by JHarris at 3:18 AM on April 26


I believe articles linked in this thread talk about tile generation.
posted by garlic at 6:02 AM on April 26


Hm, I don't remember that, but it's been a month since I read them. Ah well.
posted by JHarris at 6:10 AM on April 26


Yeah, learning all the rules for where new tiles can appear definitely opens up better planning and deeper gameplay.
posted by mbrubeck at 5:17 PM on April 26


Yes, but I'm conflicted about it, or rather, I see how others can be conflicted.

I'm an old school Nethacker, and I don't necessarily see a problem with hidden rules and mechanisms that must be discovered through play. But there's a prominent thread of thought in game design currently that thinks these things are always bad. The fact that generated tile locations are not entirely random counts as a hidden rule, and some would say takes away a little of the purity of the game. I know, when I had my little revelation about tile generation, I soon after had my best game ever.

There are other random elements there too, that if it turns out are not actually random, could make for much better play. I will have to examine them.
posted by JHarris at 6:01 PM on April 26


I take the opposite view: The purest form of gameplay is reverse engineering. I consider a game won if I understand it well enough to design and implement an algorithm for winning it, even if I suck at it in "real life." Conversely, I don't feel any victory when I finish a game if I can't explain how I did it.
posted by mbrubeck at 8:17 AM on April 27


(That is, I take the opposite view from the “prominent thread of thought” that JHarris mentions. JHarris’s own view seems closer to mine.)
posted by mbrubeck at 8:19 AM on April 27


« Older Mike Tyson highlights with Street Fighter sound ef...  |  "Debate has surrounded the leg... Newer »


This thread has been archived and is closed to new comments