I agree, but this would give you a starting point in that you could see exactly what steps were used to create the figures in a paper. It might be that their h

I agree, but this would give you a starting point in that you could see exactly what steps were used to create the figures in a paper. It might be that their high level description (usually in papers anyway) is correct but their implementation is flawed in some subtle way that peer review doesn't pick up. Assuming a correct implementation this would be useful for anyone wanting to use the methods of the paper. For example, I've spent the last 3 weeks coding a numerical solver for some equations in a fluid mechanics paper. Having the code for this available would have saved me figuring out all the quirks of the solution.