The computer is a part of you, and you should feel all the pain for trying to access out of bound memory

Thanks to cheap copying and the internet, developers can now ship that pain to any number of users instead of keeping it to themselves.

Why would it? A computer has no nervous system or self preservation biology that would make use of pain signals.

We are talking about pain, not a pain signal. Decrementing register can be a pain signal, but it can't produce feeling of pain (qualia). If humans can be simulated on a Turing machine, we can't feel that definition of pain too.

If we make a robot sufficiently similar to us, it would indeed feel pain (qualia). The real question is: Where is the complexity/similarity treshold?

They obviously can't feel pain. Imagine you simulate that robot with a turing-complete cellular automaton on paper. At what moment in time will it feel pain as in qualia? When you scribe the next tick on a paper?

Isn't the most painful instruction a conditional jump? Especially when the CPU had preloaded the following instructions and realizes it has to abruptly branch somewhere else. If there is a mechanism of pain in the CPU it should be triggered by this unpredicted waste of power and time.

If the computer is running a complex algorithm that tries to keep the value of %eax as high a possible, then I think it would feel pain if you decremented that value.

Is reinforcement learning just hurting the computer until it does what we want it to do?