I like the concept of the game and it indeed requires analysis to play, but I think the algorithm for the starting setup of the board needs to be improved:
I looked at 2 examples and on both you cold not use 8 tiles as they were next to ones which were not in sequence, so if you select the highest of the tiles that join then the maximum possible score regardless of how you join the tiles was 80 in the first board and 88 in the second board. The average of the tiles is 5, so if you can't use 8 of the availale 24 - this leave a maximum possible score on average of score of 16x5=80. You have a choice on the tiles that are joined so you can slightly improve on this, but it is usually impossible to use all tiles and you won't always be able to use the highest value tiles for the tiles which are joined and not in sequence. So a high score of 101, must have meant a quite a favourable starting board as a score of 101 is impossible in most boards. Having a fairer starting board would means you would know if you are improving, or doing better than someone else, but a fairer starting board is not trivial to implement. One possible way would be to keep track of the sum of the usable tiles as they are generated and if you are above or below the average, then bias the generation of the next tile so it is of lower or higher value.

Rated 4 out of 5 stars by Mike on April 7, 2014