Sometimes, game-like features in an application can make work fun. For example, Stack Overflow uses badges and points to coerce its users into doing work.
I think the main issue is that in most applications, they are used for a purpose. They don't need to incentivise the user by making it more "fun" and it's generally a distraction. Imagine what would happen if Visual Studio (or whatever your favorite IDE is) gave you badges... Just like here, many people would concentrate on acquiring those badger instead of writing good code.
Another thing is that, at least in the case of badges/achievements, they're fairly meaningless for offline applications.
Nowadays, games are synonymous with community.
Most line of business applications don't include a wide variety of multiplayer or community aspects to them.
It's definitely true that 'game' features in an application might be distracting and detracting from the effectiveness of a lot of applications.
The idea of adding game features to a product is to impose some sort of economy to productivity--a reason for working. For example, the badges here are kinda neat, but what really drives people to do well on SO is the reputation. It enables them to make a larger difference and more impact, and then also ties them to a feel of responsibility for the site. I think SO really strikes a good balance here.
Although, game features in other apps can be insulting imagine this:
> gcc -c main.c -o main.o
Compiling... while your waiting, what's your favorite color?
Edit The question you might want to answer very specifically is "What behaviour are you rewarding, why are you rewarding it, and what is the reward?" If all those have to do with productivity and nothing to do with some orthogonal happiness (ie social standing) I'm not sure its going to work. End Edit
On a completely different note, you must watch this talk on "Human Computation". Wow.
http://video.google.com/videoplay?docid=-8246463980976635143
It talks about using games to categorize images for Google. A little off topic, but you might appreciate it.
Games are really educational applications. True, what they generally teach is how to play the game, but they're still educational.
By the time you finish a typical game, you're an expert in a dozen different mechanics, know how to handle complex scenarios, and can recognize multiple different foes and their patterns.
While game mechanics themselves ("jump!") may not be applicable to typical applications, a look at how games approach teaching certainly could be.
Brilliant question Evan! And now for my definitive answer:
I think that any work can become fun if you break it into attainable challenges. An application becomes a game when it provides these challenges, explains them, and gauges success or failure.
The difficulties in building challenges into applications are...
For these reasons, building specific challenges into an application has very limited value. Social games may be an exception because other users partially define challenges and gauge progress appropriately on a case-by-case basis.
Flashing lights and other shiny stuff. Good games are loaded with colour and give the same pleasant stimulation as watching fireworks.
One of the places where you can see this principle being effectively applied is for applications that use people to generate or index content. In these cases, the game-like aspects are a way to encourage self-moderation. For example, on SO, the rep and badges aim to encourage constructive behaviour like higher-quality answers, peer review etc. Similar systems exist on many generic forums, as well as sites like boardgamegeek and wikipedia.
I could imagine this kind of thing working well for things like community/company wikis, software documentation, or adherence to coding standards or test coverage. The problem, as ever, is to stop the game becoming the main focus. For example, if you could get rep for tidying up your intranet wiki, I can guarantee there would be some people who would do that all day, when their main job was something quite different!