I believe I've brought this up in a different thread before, but it's still something that I'd like to see. It seems possible and I believe it could revolutionize digital modeling. It's simple really; just a digital representation of an effects circuit. We already have circuit modelers such as SPICE, so modeling a circuit is doable. You'd just need a A/D converter to change the audio signal into a digital representation of the voltages, currents, etc. and then you could run it through the circuit. SPICE already has a symbol for a wave input with all the necessary values, so why couldn't it be a signal? Think of the things this could do. Instead of modeling the response and behavior of, say, a Big Muff, you'd model its components and circuit and the response would follow. You could have a reasonably close tube tone by modeling an actual tube. And it could be better too, if you'd like. Instead of being limited by the constraints of real components, you could use the ideal. A capacitor with no ESR, an ideal current source, resistors that are exactly in spec. I realize there are some difficulties. Specific active components (tubes, transistors, etc) would require a lot of coding. You'd have to input the device gain, it's input and output resistances, stray capacitances and many more variables. But we already have reasonable models of the real thing in modeling programs, so it might not be too difficult. Say, a table with entry points for device characteristics. You could even use the datasheets for that. Perhaps it would even allow for a range of tolerances that it could randomly select from. I realize there would be limitations; for example it would be tough to model the output of a power amp as the power supply would have to be big enough to put out the actual swing. I'm sure it would likely require a decent amount of processing power as well. Still, wouldn't it be cool to have?