A few quick issues that I encounter with this (in trying to build a basic waveform generator):
* At a slant angle view a lot of components are blocked and cannot be clicked, one must change the view in order to target them.
* There is no feedback provided to disambiguate between when the mouse will engage in camera/view navigation vs component translation.
* No feedback is given in the hover state to let the user know when component they are targeting. (I frequently moved the whole breadboard by mistake).
* Component placement is very hard. The holes are small and all look the same. Offering some feedback about which holes would be connected would be much easier to use.
* Connections are hard to ascertain. Sometimes this requires a change of views, but also the bread board get cluttered. Offering some visual affordance for the connections could help make this superior to a physical equivalent and also remove the requirement for tacit knowledge (most people don't know how the rows/columns of a breadboard are connected.
* Wires get tangled. They currently take a linear path (in XY) between the two connections. Some research has shown that using curved paths is easier for the eye to follow (so something like a smooth step). Alternative offering some ability for the user to push/pull them apart or edit a bezier path might help her).
* Components have no flexibility. One must use wires to get the correct circuit topology which complicates things visually/ interactively.
Again I want to be clear that I think that this technology is really interesting and it has a lot of promise - especially since it is a pure web stack. I also think that it is not yet at the stage where it can arrogantly mock its rivals for using 15 year old UI component libraries.
"here's an interactive simulation to illustrate my point <embed src="diode-url" />"
instead of:
"to see my point please download these files, download this version of ltspice, and then give it a run"