I agree and there has been gene duplication.
What we have not seen is the generation of a novel enzyme sequence and there are very good mathematical explanations why we would not from cell division alone.
The first step in establishing a mechanism is to model it. Not a toy model but a model that shows a functional sequence can be generated. There are 10 million organized amino acid sequences in human cells. Dawkins tried this with weasel but it required the target sequence to be available. I asked how we would model the generation of 300 english characters. T said gene duplication then random variation. With 300 characters the odds are about 8000 to one that that the first 3 mutations would start to break down the sequence.