First, this approach is oversimplified. Each action requires parameters, but may proceed with semantically various sets of objects. Depending on a situation, what is available to us may limit our options. Also we may vary the volume (we not only press the gas pedal, but decide how hard to press).
Second, using matrix multiplication may be inefficient in real time. Consider hierarchical fall through with logarithmic complexity.
Overall, I welcome this approach. Indeed, we need to store not only facts but also causal relations.
Thanks for reading the post. I think we’re talking at different levels of abstraction. The goal here isn’t to propose a full decision system or optimize runtime, but to build intuition: how a graph maps to a matrix, and what basic linear algebra operations mean once you do that. Even very simple mathematics already captures rich dynamics like propagation and convergence. The simplicity is intentional.
“Science is what we understand well enough to explain to a computer.” - Donald Knuth
First, this approach is oversimplified. Each action requires parameters, but may proceed with semantically various sets of objects. Depending on a situation, what is available to us may limit our options. Also we may vary the volume (we not only press the gas pedal, but decide how hard to press).
Second, using matrix multiplication may be inefficient in real time. Consider hierarchical fall through with logarithmic complexity.
Overall, I welcome this approach. Indeed, we need to store not only facts but also causal relations.
Thanks for reading the post. I think we’re talking at different levels of abstraction. The goal here isn’t to propose a full decision system or optimize runtime, but to build intuition: how a graph maps to a matrix, and what basic linear algebra operations mean once you do that. Even very simple mathematics already captures rich dynamics like propagation and convergence. The simplicity is intentional.
“Science is what we understand well enough to explain to a computer.” - Donald Knuth