Iteration is an essential approach to the development of any technique, not just software. It's just that code, no matter how high-level the language or kitted-out the infrastructure, is about the slowest medium to iterate with. This isn't to say that it is never appropriate to iterate code, just that we can get a lot of mileage out of instruments as simple as a pencil and paper, or a whiteboard and some frank discussion.

A good indicator of whether or not to use a computer to iterate — or if so, in what capacity — is the volume of information relative to the rigidity of its shape. Computers perform well over large sets of relatively homogenous data, but the performance gains break down when we need to get heterogeneous data all the same shape, or worse, we don't know what shape they should become. Low-cost, low-tech methods enable us to experiment profligately with different shapes, but break down when we add volume. Computers are also useful when we already have many pieces of information and need to experiment moving them around relative to one another. That said, it is a mistake to underestimate the utility of a photocopier, some scissors and some tape.

It's worth noting that this relationship can be periodic too. We can take interim results from a computer into a low-tech world just as easily as we can use them to crisply refine the chaos of yesteryear's communication techniques. For example, most of us can type much faster than we can write by hand, and the results tend to look better. It may make sense, for instance, to assemble microcopy established in a previous sketching iteration into a document, print it out and glue it to further sketches. Potentially more effective would be to simply number the text in the document and put those numbers next to some horizontal squiggles in the sketches.

Conversely, there is little wisdom in designing for aggregated content such as a control surface for a corporate database, or even presentation graphics, without in the very least some kind of feel for the shape of the data. Many an ostensibly good design has been botched when lorem ipsum was supplanted for real content.

As the preeminent digital artist Dr. John Maeda wrote, [t]ap the power of the design machine. Its metamechanical hands do not tire. Just ask yourself first if untiring metamechanical hands are really what you need at your current stage of development.