I think that is a misconception. Those not-so-realistic games are just derivatives of what is known to most people, but with a twist to some things of your common sense-model of your reality. So if the game makes it plausible that you can shoot portals with a gun "onto" walls and you can walk through them, you still apply your well known model of physics to this exception, even if unconsciously. If those two things (new, non-realistic thing and known rules) fuse well and it is fun, you have a really good game mechanic or -look, like in Portal 1 & 2, or Borderlands. Compared to the shot RvlExile posted above, you find also some dishes in Borderlands, like in this screenshot:



Even though the scene is very strange, you can see a porcellain teacup and a bronze teapot on the table and handwritten place cards on the table. It looks like our super not-so-real freaks with those strange clothings had a nice tea party. Before that crazy girl blew the whole thing off. See? Even if you build a not-so-realistic game you have to stick to real-world things, shapes, dimensions and common sense to make it plausible. If very much of your game is not plausible and highly disagrees with the things the player is used too, he will most certainly reject it. Or won't have fun.

That accounts to all things and relations you know and therefore you can't say it is simpler to make a non-realistic game. Sometimes it is even harder if you introduce very innovative stuff that is in conflict with some fundamental principles people know and are used too. So, yes, making a photo realistic game is very much work for one person, but a non-photorealistic game can be the same amount of work.

Don't be fooled by simplicity. Making simple things really good is sometimes more time-consuming and much more work then throwing in a potpourri of the greatest hits.