The act of reconstructing something from very fragmentary traces requires us to depict unknown knowns, things that we know existed, but of which we have no actual knowledge. We know that a posthole did once contain a post, but whether that post – at least above ground – was round or square, plain or highly decorated, or how high it was, is something we do not know. At best, we can make rough estimates, but usually those have a wide margin of error. In visualising that uncertainty, applying Occam’s razor – usually a sound scientific principle – is the worst possible choice: if always using the minimal assumptions necessary to reconstruct houses from posts, the outcome will necessarily be the same minimalistic result. And since a picture says more than a thousand words, we will impress a fundamentally false picture of the past on everyone’s mind: on that of the public; but also on our own, who are equally influenced by the illustrations we see in each other’s work.
Thus, in this paper, I will argue that for making our reconstructions more reliable depictions of the past – not in terms of the details we show on each individual one, but in terms of the overall picture of the past we convey through reconstructions in general – we need to be radically creative. We need to produce, not just the reconstruction of how the object of our attempt most likely looked, but several reconstructions which show the range (the ‘standard deviation’) of conceivable possibilities of how it might have looked like – even if, for this purpose, we have to make maximal assumptions.