Why is it that games tend to rarely vary their form? You start with a set of activities-- running, jumping, shooting-- and while they may deepen and expand, they almost never change significantly through the entirety of the experience. By (very flawed) contrast, consider how much more malleable as a medium books and movies can be: They can switch genres,  alter perspective and even change subject matter entirely. A movie like Good Morning, Vietnam, for example, starts out as a comedy but ranges into romance, drama, suspenseful action and even tragedy (Hancock, From Dusk Til Dawn, Vanilla Sky, District 9 and Click are movies that could also fit this example).

What is it about games that so limit their form? Is it the maturity of the medium? The strict genre expectations of the audience? Or is it possible that one of the medium's greatest strengths-- interactivity and the process of engaging with it, which is basically learning-- is simultaneously a weakness of sorts? I lean toward this. Maybe the process of learning and mastering mechanics sets a kind of upper limit to what a game can depict and the sum of what experiences it can convey. Card games and board games seem to share this limit-- chess does not morph into poker, for instance, and it would be hard to see it do so (I wouldn't count playing both at the same time).

Or maybe the whole question is flawed and the comparison to books and movies essentially apples to oranges. You might argue the bulk of movies & books fall into well defined genre categories, for instance, varying similar plotlines maybe much like a shooter or platform jumper or racing game varies levels.

But I can't shake the idea of how most games can't really even switch genres let alone their overall form. When you start most, you know you'll be doing basically at the end what you started at the beginning, just maybe with different permutations and contexts. Imagine the howls of going from a shooter to a match-3 game, or from an RTS to a racing game, even if it was a smooth transition. Games that have attempted this, like Spore (or the little known Gordon Alliance decades before it), often run afoul of the problem of sacrificing depth for breadth or suffering from mechanics that just don't cohere well together.

What do you think?

• Changes within a game need to be slow and simple enough for players to play effectively while they take place, without misunderstandings. On the other hand books, and to a lesser extent film, can disorient the audience until the changes are accepted and/or understood.
• Games have a formal structure of rules, which cannot be ignored, while narrative media have much more elastic internal coherence standards. If an airplane pilot is shot down and rescued by locals, in a story it is a reasonable and realistic development of what happens of a character, while in a flight simulator game it's the end (at least until the pilot returns to service and the rules of the flight simulator game apply again).
• Games containing multiple "forms" need to embed the respective rules in a rules superstructure (for example, trading simulations and action games can be combined meaningfully by acquiring equipment for use in the action parts and by looting trade goods in the action parts). This is far more difficult than simply making things happen in a story.

The thing is, when you start playing a game, you probably like it because... it's that game. If you're playing a side-scrolling shooter, you don't really want to start playing a top-down strategy game instead. Unwelcome changes like this are a real turn-off.

I can think of examples I've personally experienced, even, and I'm not even that hardcore of a gamer:

• Power Rangers S.P.D. on the GBA was normally a side-scrolling beat-em-up/platforming sort of game, but every once in a while it had a crappy Missile Command-esque stage. I dreaded those stages at the time and there was one point where I almost quit because of one of them, IIRC.
• In Mario & Luigi Bowser's Inside Story, I was constantly annoyed by gameplay changes to the point where I just never completed it (I got stuck on one of these stupid mini-games it forces on you, turned the game off, and never turned it on again). It's like the developers couldn't decide on a gameplay experience and just shoved as much crap in there as possible.

Think of it this way: if I don't like the game enough to play it to the end, then I'm not your target audience. If I do like the game, I don't want the experience hijacked in the middle of it. So really, making drastic gameplay changes like this is just bad for everyone involved. Plus, as has already been mentioned, it's extra cost.

Well, when I'm sitting with a tea in one hand playing a turn based strategy I would really hate if there were parts of the game where I would be forced to jump, shoot and do other arcadish activities like that

There used to be games like that in 80s and early 90s but fortunatelly those extincted by now. And it makes me quite happy I would say

BTW, Sword of the Samurai is one of the few of that type that sounded quite reasonable http://www.homeoftheunderdogs.net/game.php?id=1112 (at the beginning you are samurai - arcade and then a daimyo - strategy).

I don't believe anybody has approached this from the development side: the added complexity of putting in what is essentially an entirely different game. That's a lot of extra development time and hassle to do it well, and if you DON'T do it well the "less limited" form of the game will be a hindrance and not a benefit. Players will look at your RTS sections as half-assed and unfun, thus pulling down the entire game if you don't pull them off.

Many games do attempt to play with genre within their particular rules. For instance, Starcraft has levels without basebuilding and with very limited resources, making them akin to some sort of dungeon crawl/adventure game. These are built to vary the games experiences without actually requiring the player to learn new skills.