The best way of improving your skills at something is to actually do it. The old saying that practice makes perfect is mostly true. I am one of those people who don't believe that perfection is possible as there is always room for improvement. The second best way of learning is by learning through others. That is at least the excuse that I use when playing games. Just playing a game is not enough, though. To learn from a game you need to think about why you are enjoying the game. Think about why the elements work. Often even the annoying things serve a purpose. I have found that it is easier to learn from a bad game than a good game as the mistakes that make the game un-enjoyable tend to stand out far more than things that make a game good.
My sister got me Torchlight II as an early birthday present (yes, she told me I could install and play it before my birthday this Sunday) so I decided to take a look at why this game is getting such good reviews. I am only a few hours into the game but have enjoyed the game so far. The biggest lesson to be learned is that small cute dogs can be extremely vicious.
The game supposedly randomly generates the maps to increase the replay-ability. The levels I have played so far are pretty good but I have noticed that a lever puzzle was repeated making my second attempt at getting the treasure exceedingly easy as I knew exactly what steps were needed to solve the puzzle. What I suspect is going on is that the random dungeon uses a mix of random dungeon generation and canned chunks of dungeon.
This is actually a good approach to take, as it gives the replay-ability of randomly generated dungeons while also giving the much more streamlined and thought out designs that human created maps have. For me, and probably a large number of Torchlight players, this random generation of maps is lost on me as I have such a backlog of games that it is unlikely I will be replaying the game multiple times.
One advantage of hand-crafting the maps, at least in theory, is that it gives the designer more control of the game difficulty. Torchlight II appears to have solved this problem by using the player's current level when generating the quest map. So far every map I have come across has had the levels exactly mach my current level. I am not sure if this is actually the case or my exploration has just coincidently matched the level suggestions. I am hoping the later as one advantage that canned levels have is that it is possible to grind if the game gets too difficult. If all the levels are based on the level I am when I reach the quest then if the game is getting too hard, I am going to have problems.