>Theoretically, with a powerful enough computer (many, many years from now), you could accurately predict any future event.

Uh...no.

There are formal mathematical proofs demonstrating that perfect solutions to systems of non-linear differential equations (like the weather) can only be calculated if one possesses perfect knowledge of boundary conditions.

It doesn't matter how powerful your computers are, imperfect knowledge of every conceivable (and/or unimagined) relevant condition will eventually overwhelm your model, no matter how accurate the model is. To simplify things, a butterfly flapping its wings in Angola will eventually lead to an unexpected countercyclone over Idaho. And there are millions of buttefly's flapping their wings every moment. Increased computational power will not help avoid this.

And that's before we start considering human errors during the programming and data entry, which will be considerable.