From a game design standpoint, Dota 2 seems like an objectively bad game. There are several design decisions (that I suspect are mostly due to the series’ origins in the Warcraft 3 engine and its restrictions) that appear from the outside to be pretty darned stupid. Somehow it all comes together for a clearly very successful game, though.
A good game is an intuitive game but Dota 2 frequently is anything but. To prevent the enemy team from gaining experience and gold, a good player will kill friendly units (even player controlled ones in some cases!) when they have a sliver of health left. ‘Last-hitting’ enemy units is, of course, a good idea too - unless you are playing a ‘support’ character and your lane partner is a “carry’ - arbitrary categories defined by the community, and only vaguely by the game. Even upon seeing an enemy at the start of the game it’s usually a very stupid idea to immediately begin attacking him, going against pretty much everything players have learned in every other video game.
The Idle Thumbs crew mentioned something like this in a recent podcast, though on the topic of Starcraft 2. The question was asked ‘what should I do next to get better at Starcraft?’ There’s no clear answer to this, and I feel this is even more true about Dota 2; pretty much the only response is ‘just keep playing.’ And in a game that is clearly non-intuitive at times, telling players to just figure it out is maybe not a good thing.
Valve, usually fantastic at making easily accessible games, seems to have realised that there’s not much chance in doing so with Dota 2. Instead of a gentle step-by-step introduction to mechanics (which as Idle Thumbs pointed out, there can be no clear path to do so) they’re planning to introduce a coaching feature, allowing an experienced player to coach a new player, live. Probably a very valuable tool for newbies but it will be interesting to see how interested the coaches are at devoting an hour or so to it.
Overall, Dota 2 seems to be a game greater than the sum of its parts. Or perhaps it’s an example of how obtuse mechanics can build great dynamics. Regardless, its success is a clear indicator that something has gone right, and I think game designers who work on refining games as much as possible could do well to examine how Dota 2 (and similar games) stay successful while retaining complexity.