Not yet reviewed by Pith; the record is open.

We present a model of proton geometry where the number and size of gluon density hotspots in the proton's thickness function evolves with the resolution scale of the event given by the Mandelstam $t$ variable in exclusive diffractive $ep$ collisions. We use the impact-parameter dependent saturation dipole model bSat/IPSat, as well as its linearised (non-saturated) version bNonSat. In the latter the proton thickness has a clear interpretation as a thickness and in the former it is directly related to the saturation scale. The resulting phenomenological model for the splitting of hotspots, making full use of earlier experimental and phenomenological studies, is able to describe the entire incoherent $t$-spectrum for $|t|>1.1~$GeV$^2$ with a single phenomenological parameter. We use the previously suggested hotspot model as an initial condition for our evolution. The resulting model is a resolution scale-evolution in the same vein as a parton shower.The incoherent cross section is directly proportional to geometrical fluctuations in the proton's inital state. The hotspot evolution give rise to several kinds of event-by-event fluctuations such as in the hotspot number, width and normalization, and saturation scale fluctuations is a direct effect of these. A natural consequence of our resolution based evolution is that the hotspots obtain an effective repulsion. We use our hotspot evolution model to investigate saturation scale effects in the $t$-spectrum, and found that HERA data is not sensitive to this physics.