Finally I came across something that can cure the DTD addiction. It's video on how the top scores are made. You can learn a lot, but you may also give up altogether - as I said, the game isn't great after all. So if you want to play your own way and improve at your own pace, skip the videos.
Before the video's and the analysis of the top score strategies, I can disclose some tools that may help you to have more fun. In the challenge mode, there is the 3K Fixed game. Instead of 80 gold at the start of the Hard game mode, you have 3,000, but you can't add anything afterwards. It's a fast way to test out your maze, compare your weapons, and measure the unlisted properties of the weapons, such as how much the freeze towers slow creeps down. The weapons, if available, are the same across the games, but the health of the creeps are not all the same across games.
The other useful game is the 100K Gold in the fun mode. It's the same as 3K except that you can play as in the normal mode after spending the initial gold. Even better, there are 100 levels instead of 50 in the main game, everything else being equal. So you have more gold to spend on the first 50 levels and hence more flexibility to test things out.
Finally, there's The 100 in the challenge mode. It's the same as the hard mode but with 50 more levels. So it's exactly the 100K Gold but you have only 80 gold initially as in The 100 game. The 100 doesn't help you with anything but this is simply a superset of the hard mode with more levels. You may or may not want to start with this game rather than the main game. Because the high score strategies are different - that works for 50 levels doesn't work for another extra 50.
Other modes and games are deviations from the main game. For example, some guys hate juggling, the selling and buying of the towers. It became less of a strategy game than a shoot-them game. But then there's not as much challenge because of the limitations of the game.
I used a spread sheet (Open Office) that contains data of gold earned at each level, the health of creeps at each level, and all the measurable properties of weapons, like firing rate. So you can calculate to some degree how much fire power you need at each point and if you can afford them. I also use the Open Office drawing package to help with designing mazes and acting as blue prints during the games. But with modes such as 100K, the need for these outside tools are less. Also, if you see the videos, precise optimizations are either not necessary nor not possible at the end.
I think there's some sort of collective conspiracy in the DTD forum, which missed a lot of critical information, perhaps in order to keep new players interested in the game. Firstly, as I expected, many top scores are fake. They play some other modes and hack the submission forms to change the name of the game to get impossible high scores. Or they can just hack the flash code downloaded to everybody's computer during play. So the top players started to post their fast forward video on Youtube.
The first video I recommend is a classic strategy for the main game DTD 1.5 Hard, with no juggling, or even no selling at all.
Notice it's so easy using only one Swam (6) tower and all Squirt towers. I have done similar things before but I gave up because it wasn't likely to beat the high score. The strategy is easy. I have calculated that I need the equivalent of 4 swan 6 towers to kill the air bosses. When I saw this, I went back to my spread sheet and calculated if I reduce the sw6 to 2, then I need about 7 squirt 6 towers. So I need only two boost 5 towers and one sw6 to give the equivalent of 2 sw6 towers, cheaper that way. And since building extra sq 6 towers is more economical than adding boost towers, it's pretty easy to place the sq towers without worrying packing them together to touch the boost towers. There are more than enough fire power and enough money left for two frost 5 towers to slow down the air bosses to receive sufficient hits.
As for the maze, I said that the optimum pattern for sq towers is the cross, proportional in size to their firing range. But with two or more sq packed together, the cross doesn't make much sense any more. I used the drawing package to draw range circles on each sq. I think the optimum pattern is like peeling an apple - something like the maze in the video. It's not neat, but it doesn't matter because there are more than enough fire power.
With this setup, a new player and an old player differ only in how fast they hit the send-creeps early-button to add extra scores. Very soon you will hit near the high score of this strategy. Note that the no selling is a self imposed rule. The nearest game is the 3K fixed game if you want to fight for the top spot, but 3K is easier because some bosses are less powerful I think.
The next video is the top score of DTD 1.5 Hard:
Basically the guy almost managed to send the creeps altogether, and kill them gradually long after the last creeps came out into the maze. You need always to build enough sw towers to kill the air creeps, because you can't keep them in the maze. Sq towers aren't useful to air defense at all because they will always be distracted by the land creeps. So only bash towers are useful. Since you have very little money left for it, you don't worrying how to put more of them together with boost towers.
The maze isn't that important as long as you have two alternative paths, gold efficient at the beginning, and space efficient at the end. My staggered DNA pattern works rather well, with three DNA pattern on each path, and utilizing the remaining desktop space as maze. But since you have to share the pellet towers in the two paths to save gold and space, the DNA pattern aren't necessary.
The surprise is the splash towers. Not only that they fire at everybody in range, but they are completely frozen for a long time. I gave them up because of the seemingly weak power for the money. But the freezing allows the other weapons to be many times more effective. So killing the air bosses becomes very easy.
After I saw the true master at play, I decided that this isn't my game. It's all about timing, more of a shoot-them-all game than a war strategy game. Getting the timing right initially is hard but there's not much potential to optimize.
I could have completely cured of the addiction if I hadn't see the 10K fun:
The juggling strategy for the hard mode won't work anymore over 50 levels. But, you can't survive without a little bit of juggling. So it's a balanced game. But at the end, you have to use a lot of splash towers, and reserve a large central area for all the towers. So there aren't much variation to the maze and the use of the splash towers become more of a shoot-them-all game.