How does one define a genus? By it's genome? I would call an entirely artificial fruit 'an apple' if it were sufficiently similar to one.

I would call a raven a writing desk if it were sufficiently similar to one :P

But yeah, since we're really talking about human pattern-matching, apple is defined in terms of "common but not strictly necessary traits." Similar violations of any non-fuzzy definition could be constructed for shoes (hand-shoes), wire (a piece of wire shorter than it was wide), green (yellow), politicians (candidates), and hope (simple expression of preference). You might even be able to convince people that if you make a shape out of wood it can be a square even if the sides aren't exactly the same length. However, it is a convenient convention to omit this fuzziness in definitions since it's so common, and instead rely on the judgement of the reader to associate a thing with the closest definition.

Or, if there is no definition that is not different from the observed pattern in some key way ("key" here is subjective and mostly functional - a biologist might find genes of a species key but a non-biologist might find appearance key, and not vice versa), humans might make up a new category for this pattern.