Indeed, you are free to. On occasion I've tried to do the same thing. But in practice, I don't think it usually helps. It muddies up the waters of conversation, which depend on everyone having roughly the same definitions of words. At least, you have to be careful, and know that most people won't (immediately) change, and that you're making things more complicated.
It's important to have clear and powerful linguistics. Sticky linguistics make sticky minds. Compare inuits, with their multitude of words for snow - they probably have richer and clearer discussions on snow.
The same thing is true in programming languages - eg people who go back to basics and throw out the cached thoughts that make up a web of assumptions that everyone clings to, and in the process create something weird yet powerful (like Haskell or whatever).
I know, but bringing up alternatives to entrenched meanings doesn't always make things clearer. Probably a lot of the time it's better to introduce a new term, or one that is used infrequently and doesn't have as much baggage, rather then redefining one. Redefining terms when talking to a computer is one thing, but people aren't always capable of throwing out their assumptions, and it almost always takes a long time. In the meantime, where is the discussion? The mess has to be worth it.
Allow me to interject for a due correction: the inuit "having x>>1 words for snow" is a widespread misconception; Eskimo-Aleut languages feature compounding (as German does), thus allowing an arbitrarily high number of variations of the "snow" lexeme, which can be misunderstood by very uninformed and very monolinguistic individuals (such as lousy journalists writing fillers) as being different words.
