Now I get it, flooding the bugs probably works better. You can do that with plain water, no need for insecticide. Might cause some problems with short circuits and such though.
Strange, how this bug-wash gave me a good idea.
Bugs do have different behaviour as working software. In a 3d presentation of the working memory one could see where the flow is focused.
If one was to construct a computer that shuts down in a frequency each pulse starting up with nothing in stead of continue to work with the already present data in memory; wipe it and start all over.
Such computer would be unable to crash or to put it more correctly, it would continue to crash permanently while working. Using a heat/flow map could identify unfinished processes. Bugs just don't stick with programing rules. As a result, any pile of data could be ran as if it was software the PC would not crash.
This brings me to a really old and odd idea I had 22 years ago. What if a CPU would try to run some random data as code, make good documentation of what went wrong and adapt the random data to be more valid code and run it again.
Depending on routines written by real humans the apparatus will remove code that has no clear function or preforms a duplicate task. Both with respect to the amount of chaos available in the system.
We can teach it how to build user interfaces so that we can always make sense of the end result. We already have some technology based on this concept. For example maximum 1 pop-up per 10 seconds plus one extra for each user interaction. More as that would be an error marked for deletion. We can just fire off random pop-up data at the pop-up blocker and it will allow some and block the rest. All it needs is a closed loop where the detected error is directly enforcing change of the software.
We could build that mega brain, it will take trillions of cycles to make a simple application and we shall have no idea what it will chose to show us untill we do.