[harry8]

The error here is not understanding the data being transformed.

You won't get it right either way if you don't /know/ how big n is going to be.

If you can't know, why not? Should you even be coding this at all?

Maybe there should be a rule zero.

0) Understand the data on which your data transformation is going to operate.

The extent you don't is the extent to which the endeavour is doomed.