> We use a similarly flawed model every day: fruit and vegetables. What's in one group or the other is based on vague similarities and is mostly arbitrary, but

> We use a similarly flawed model every day: fruit and vegetables. What's in one group or the other is based on vague similarities and is mostly arbitrary, but it's still useful to split the huge category of "edible plant parts" into more manageable chunks. This is completely wrong. Those categories aren't any smaller, in a practical sense, than the combined category "plants". The gain in manageability is zero. The reason we divide the categories that way is that we put them to different purposes, not that they'd be too large if combined.