There was a thread a few days ago on Asher's Twitter account about what could be done to better balance the difficulty of the game post-768 or post-1536. Some of the ideas were to restrict the range of the '+' cards (i.e. raise the lower limit above 6), or to give stronger hints about the bonus cards; e.g. '+' for one digit, '++' for two digits, '+++' for three digits. (or '+' for 6/12, '++' for 24/48, and '+++' for 96 and above.) Another idea that just occurred to me would be to reduce the 12-card "deck" to 9 cards (three 1's, three 2's, three 3's) once the 1536 tile is reached. This would preclude the current worst-case scenarios where the "color count" goes to +- 4, which are basically unsurvivable if they occur at critical moments. Any other ideas how to make the 6144 card slightly more achievable, without bending the rules too much? I've achieved 1536 a handful of times now, but not yet 1536/768, and I'm nowhere close to Whalend yet. (Congrats solrelami, nicely done!)
So close!! (Final score, 88,683, currently #223 in the Game Center rankings.) I'm getting my Threes strategy down to a science! Here's my strategy: I line up the largest cards along the right side of the board, with the highest at the top. This card never moves. Then I make sure the right column is full, which gives me three degrees of swiping freedom: down, up, right. From there, it's a matter of keeping the larger cards migrating to the right, which I'm slowly getting better at, and merging them into the rightmost column when they get there. When the rightmost column can merge upward, I quickly fill in the lower-right gap (ideally with a large card) to restore the three degrees of freedom. It's important to keep a nice "convex" healthy board, avoiding small cards sandwiched between large ones. This game was easily on track for the 3072 card until the pesky 192 bonus showed up! A hint that that it was huge (e.g. '+++') would have been extremely useful here. Must get Whalend! I'm addicted! Halp!
After being thoroughly addicted to 2048 for a couple of weeks I've finally caved and got Threes. That's at least one sale they owe to the clones. I'd never have bought this without 2048 showing me how addictive a game like this can be.
Åaand yes it is rather addictive. I got over 7000 on my first go, but still haven't made it to 8000 - and haven't managed to get a tile higher than 384.
Thanks! Just got Whalend again, second game in a row! Final score 187,971, good for #97 on the high-score list. I was sure that finally getting Whalend would cure my addiction, but alas... If you're still playing with your AI, I'd recommend assigning a very high weight to keeping the highest card in the corner, and keeping the next-highest cards adjacent along an edge. In my entire game I think I only moved the high card once, and was lucky to be able to move it back quickly. This gave me the entire rest of the board to work on making the next-highest card. My general strategy is typically to look for the longest nearly-complete "snake" on the board, and work towards completing and collapsing it before focusing on the next-longest snake. Also, if I have three large cards in a row out of order, like { 96, 48, 192 }, my first priority is to match the 48 and collapsing the group to a 384. You had mentioned your AI penalizes these out-of-order sequences; I presume it's smart enough not to penalize e.g. { 3, 1, 2 }, but probably it should more strongly penalize e.g. { 96, 3, 96 } than { 6, 3, 6 }, because {6, 3, 6 } is easily fixed whereas { 96, 3, 96 } can stay problematic for a long time. In fact, {96, 48, 192 } is a pretty good sequence to have if there's a 24 next to the 48. Another idea: your AI currently has two phases of lookahead, where you first exhaustively look e.g. 4 moves ahead trying all possibilities, then look an additional 3 moves ahead inserting blank spaces. In between these, or instead of the blank spaces, could you try a move or two inserting "rocks" that can't be merged with anything? this could keep the exponential game tree more manageable, while better simulating what actually happens in the game.
Does anyone else have a problem with this game disconnecting from Game Centre so it doesn't upload the score? My two highest scores (25k and 27k) are not showing on Game Centre because of this and it's really annoying.
Unfortunately I'm too busy with other things at the moment. One thing I did try was to encourage the AI to keep cards in increasing order on each line; but that gave worse results. Simply encouraging it to keep the highest card in a corner might be interesting, then the rest should follow automatically from the rule that encourages keeping consecutinve cards next to each other. Good point, the AI does indeed know that 3,1,2 is a good sequence but makes no difference between the other sequences you mention. I had tried that, but it didn't work as well as simply leaving the spaces empty. I don't fully understand why it works at all, since the extra space could encourage the AI to put the board in risky positions, but the simulations say it does.
I tried 2048 to make sure I was gonna enjoy that one, and after two games, I deleted 2048 and bought threes! to support the dev. Good game
I saw a friend of mine playing 2048 one day and asked him why he don't play Threes. He said that it was much harder to get through. So maybe some people think Threes is harder to progress through. I like Threes but I can see why they try other games.
Yeah. I can't convince my wife to give Threes a proper try. She likes just playing 2048 over and over again (not reaching the 2048 tile more often than not) because it requires so much less brainpower than Threes.
Off the topic of clones... If anyone has earned the final one-point achievement, can I see screen proof that it needs a 12288 tile?
New high score, 208,554! Good for #55 on the all-time list. Keeping the high card in the corner for most of the game made all the difference; the bonus 384 sliding in on the right was the killing move at the end. In hindsight it would have been safest to slide the bonus card in on the left, and hope for a 6 or 192. (28% chance, but better than the 14% chance of matching the 96 on the right.)
Blimey! I can't imagine getting a score like that. I've only managed to get a 768 tile three or four times.
Getting Whalend every few games now. This board is currently #26 in the rankings. I'm reassessing the need to adjust the game's difficulty; with the strategy I'm using Volleo seems quite achievable, though in the endgame it does take some luck with the bonus tiles. Btw Greg if you're reading this, nice touch giving Terrence an extra pair of dead eyes! haha Note: 259,194 in base 3 is 111011112210, which tells you what's on the board: one 3072, one 1536, one 768, zero 384's, one 192, one 96, one 48, one 24, two 12's, two 6's, and one 3
