A science project is a feature that is really cool and challenging from a technological standpoint but is way overkill for the end-user scenario at hand.

Back in the late 1990's, a bunch of us cooked up this idea for a networked screen saver that ran at night after most people had gone home from work. You told it the physical location and compass orientation of everybody's monitor. The networked screen saver created a virtual red bouncing ball that traveled around the building in three dimensions. The screen saver showed a viewport into the three-dimensional virtual world that contained the bouncing ball.

This is a clear example of a science project: Nobody's going to sit there and type the positions and orientations of every computer monitor in the building. Even if we wrote this screen saver, nobody would actually use it. Most of the enjoyment is in actually writing the screen saver than in actually running it.

One type of science project has high set-up costs for low benefit, like our bouncing ball screen saver.

Another type of science project requires hardware that very few people have right now. For example, "If you have a tablet connected to at least two touch-enabled external monitors, then you can..."

A third type of science project is simply trying to solve a problem that nobody really considers to be a problem. You're doing it just for the Gee Whiz factor. For example, "If you have a pair of Bluetooth headphones, and you walk back into range of your computer, the computer can automatically unpause your music." Yeah, I guess you could do that, but it also means that while you are away from your computer, you're walking around looking like an idiot because you're wearing headphones.

Now, there may be an actual useful feature hiding inside a science project, but until you find that feature and bring it to the surface, what you basically have is a science project.