Interesting work and I really liked the LightTable video but I think there's a reason these types of environments haven't taken off.
To understand why programming remains hard it just takes a few minutes of working on a lower-level system, something that does a little I/O or has a couple of concurrent events, maybe an interrupt or two. I cannot envision a live system that would allow me to debug those systems very well, which is not to say current tools couldn't be improved upon.
One thing I've noticed working with embedded ARM systems is that we now have instruction and sometimes data trace debuggers that let us rewind the execution of a buggy program to some extent. The debugger workstations are an order of magnitude more powerful than the observed system so we can do amazing things with our trace probes. However, high-level software would need debugging systems an order of magnitude more powerful than the client they debug as well.
To understand why programming remains hard it just takes a few minutes of working on a lower-level system, something that does a little I/O or has a couple of concurrent events, maybe an interrupt or two. I cannot envision a live system that would allow me to debug those systems very well, which is not to say current tools couldn't be improved upon.
One thing I've noticed working with embedded ARM systems is that we now have instruction and sometimes data trace debuggers that let us rewind the execution of a buggy program to some extent. The debugger workstations are an order of magnitude more powerful than the observed system so we can do amazing things with our trace probes. However, high-level software would need debugging systems an order of magnitude more powerful than the client they debug as well.