A computer is equivalent to a Turing Machine whatever its inner engineering. So, what makes you claim that Entrelacs systems will be somewhat “clever” that current computer systems?
You’ve failed into the “Turing Trap”.
An arrow may be modeled as a node with two edges. An arrow structure is consequently a particular kind of graph. Where is the novelty then?
You’ve failed into the “Graph Theory Trap” (I’ve just invented it :) ).
More seriously, arrows structures may be -and have probably been- studied as a constrained subset of graphs.
But this approach is likely going to miss the key specificity of the Entrelacs paradigm. Arrow constructs are primarily seen as a set of enumerable objects by themselves. The paradigm shift consists in directly handle each of these objects uniquely and individually.
Concretely, an Entrelacs system constantly and incrementally maintains the Transitive ReductionW of the overall graph formed by stored arrows constructs.
The good new is that this task may actually be trivial (reads O(1) complexity) as long as one cautiously avoids graph-oriented technologies.
Try http://www.reddit.com/r/entrelacs