Distinction networks are a fine-grained parallel model of logical deduction and of combinational circuits. Losp is the implementation. The general idea is to convert a logic problem, expert system database, gate-level specification of a silicon circuit, or any other form based on logic into a boundary logic form, and then to convert the boundary logic form into a network of homogeneous distinctions.

Technically, distinctions are NOR (or dually, NAND) gates with any number of inputs. Their behavior is substantively different that silicon NOR gates, since distinction networks are based on boundary rather than conventional logic.

The first three pieces say the same thing in different words.

Distinction networks have been implemented on a sixteen-node hypercube with nodes configured to operate asynchronously in parallel. The fourth piece describes the demo for this work. This is the first implementation of an entity-based architecture, which later formed the basis of the VR system VEOS.

The final piece describes entity-based computation, and includes asynchronous communication algorithms that achieve logical deduction and circuit optimization.