Maynooth's DNA computer reaches its answers by settling into its most stable state

An ordinary computer runs far from thermal equilibrium. It spends energy holding encoded bits in place and switching between them quickly and reliably as it executes an algorithm ^[2]. In the 1970s Charles Bennett showed that any computation can, in principle, be modified to run with reversible dynamics, and a machine built that way could operate close to equilibrium and use far less energy ^[3]. Later proposals took the idea further. The answer itself would be the system's equilibrium state, and the computer would drift toward it as it settled ^[4].

Building that is hard. "The idea presents multiple challenges that include finding a physical implementation that is computationally expressive and programmable, has easily prepared initial states and has a controllable energy landscape for rapid navigation to target outputs with high probability," said Damien Woods, who led the Maynooth University team ^[5][1].

Most earlier DNA computers ran out of equilibrium. They were either driven by molecular fuel or depended on carefully prepared starting states and tightly controlled conditions ^[6]. DNA strands pair by simple, predictable rules, so the team could engineer the energy landscape directly. Badly bound strands detach and better-fitting ones replace them until the mixture reaches its most stable state ^[7]. The layout borrows from DNA origami, in which Paul Rothemund showed in 2006 that a long single-stranded scaffold could bind a chosen set of short strands ^[8]. Here the scaffold carries an engineered sequence of binding domains, and each short strand, called a tile, stands for one possible value at one step of the calculation ^[9].

"Eventually, the system settles down into its energetically preferred state which encodes the answer to the computation: a sequence of tiles each bound to the scaffold and to neighbouring tiles on its left and right," Woods said. "The competitive process of binding executes the computation." ^[11]

The same mixture can be used again. "The same molecular computer can be reused to perform new calculations, running the same program up to 25 times on different inputs," said Abeer Eshra of Maynooth ^[15]. Spread across the 10 programs, the 700-plus computations average more than 70 runs per program ^[12][1].

"Our system uses just a handful of different kinds of molecules, never really following an organized process of steps, never making irreversible steps, and yet ending up with the right answer," said Constantine Evans, also at Maynooth. "When thinking about computation at a molecular level, reliably making even those seemingly simple computations is very hard." ^[17]

The thing this doesn't tell you is how much energy a calculation used. The Physics World report does not include a measured figure, and it does not say what speed the team had expected before calling its results fast ^[1]. I think reliability is the firmer finding for now. The energy saving remains Bennett's theoretical case for building a computer this way, and this device has not yet shown it ^[3].