I think Steve Klabnik explains it best:

If you include no new features, then there’s no incentive to use your language. If you include too many, not enough people may be willing to invest the time to give it a try. Language designers should give careful thought to how strange their language is, and choose the right amount to accomplish what they’re trying to accomplish.

While this is specific to programming language design, it applies in general.

See:

• You need a novelty budget
• Choose Boring Technology
• The language strangeness budget