A Matrioshka Brain is a hypothetical megastructure, based on the Dyson sphere, of immense computational capacity. It is an example of a Class B stellar engine, employing the entire energy output of a star to drive computer systems.
Such a structure would be composed of a collection of one or more (typically more) Dyson spheres built around a star, and nested one inside another. A significant percentage of the shells would be composed of nanoscale computers (see molecular-scale computronium). These computers would be at least partly powered by the energy exchange between the star and interstellar space. A shell (or component, should a Dyson swarm be the design model used) would absorb energy radiated onto its inner surface, utilize that energy to power its computer systems, and re-radiate the energy outwards. The nanoscale computers of each shell would be designed to run at different temperatures; shells or components at the core could be nearly as hot as the central star, while the outer layer of the Matrioshka Brain could be almost as cool as interstellar space.
The ideal mechanism for extracting usable energy as it passes "through" a shell or component, the number of shells (or orbital levels) that could be supported in such a manner, the ideal size of the shells to be constructed, and other details, are all issues of speculation.
This structure has clear structural analogies to Russian Matrioshka dolls, from which the concept derives its name.
The idea of the Matrioshka Brain violates none of the currently known laws of physics, although the engineering details of building such a structure would be staggering, as such a project would require the "disassembly" of significant portions (if not all) of the planetary system of the star for construction materials. Since it is based on the concept of the Dyson sphere, objections to the feasibility of a Matrioshka Brain are more thoroughly explored in the article of the Dyson sphere as difficulties with the Dyson Sphere itself.
The term Matrioshka Brain was invented by Robert Bradbury as an alternative to the Jupiter Brain - a concept similar to the Matrioshka Brain, but on a smaller planetary scale and optimized for minimum signal propagation delay. A Matrioshka Brain design is concentrated on sheer capacity and the maximum amount of energy extracted from its source the star, a Jupiter Brain more on computational speed.
The possible uses of such an immense computational resource tax the imagination. One idea suggested by Charles Stross, in his novel Accelerando, would be to use it to run perfect simulations or "uploads" of human minds into virtual reality spaces supported by the Matrioshka brain. Stross even went so far as to suggest that a sufficiently godlike species utilizing enough raw processing power could launch attacks upon, and manipulate, the structure of the universe itself.
The theme of computers simulating human life is not new to the world of science fiction, forming central parts of the plots of such works as the Matrix film trilogy and the Star Trek: The Original Series episode "A Taste of Armageddon".
- Cray supercomputers are built in a C shape partly to reduce signal propagation delays.
- On the scale envisaged here, propagation delays would be measured in minutes. This creates an architectural problem that is not found at smaller scales. One can predict two consequences:
- Transmission of signals will also be used as a dform of storage, similar to mercury delay lines used in early computers but with vastly greater capacity.
- There will be significant redundnat computation. In some cases it will be cheaper to recompute a result locally than to retransmit it from elsewhere.
- It is hypothesised that the human brain uses bandwidth/latency trade offs in communications between the two hemispheres in a complex manner. The corpus callosum which connects the two halves carries high and low speed channels. The low speed connections have much higher bandwidth. In combination these can give a virtual connection that is both high bandwidth AND low latency. The 'trick' requires some understanding of coding theory. A customised encoding for a problem domain is transmitted slowly, at high bandwidth. This allows a much lower bandwidth to be used in transmitting a solution, once it is found, provided both sides have access to the encoding. This means that a result can be communicated very rapidly once it is found.