A writing protocol with ambiguity resilience in nanoscale systems

Writing is a directional, stochastic process by which an information substrate is constructed. It involves typing characters in a chain one at a time in the presence of memory. Unless editing is possible, plain writing fixes typed characters along the chain, a stepwise equilibration whose history influences successive characters. Here, we present a writing protocol in nanoscale systems that optimizes the interpretation of already generated error-containing information to adequately influence the next character to type. Near the noise level, as characterized by the thermal energy, and in the presence of positive feedback, this resilient protocol presents higher entropy than that in which errors can be corrected but reduces the entropy of plain writing. For negative feedback, it presents lower entropy than both plain writing and writing with editing. When the noise level is low enough, the entropy becomes negative, which reflects an underlying phase transition at low temperatures.