Not really, but I’m talking hardware reliability. In hardware defects from bad design is really very different from defects introduced by manufacturing or aging. I’m a software person also, so understand the problems of the design of reliable software, but that is really a totally different topic.
If the bits in a program really did rot, then we’d be on the same page. 
Possible? It is inevitable that you introduce new failure modes when you put in more reliability. A defect in the voter in a TMR system isn’t there without TMR. You just want the new defects you introduce which hurt reliability to be less than the defects you can mask.
Hell, we spent a year on a weird defect whose only effect was to toggle an output parity bit which caused the processor to be marked as bad. If that output wasn’t there, no one would have noticed the intermittent failure. I saw one case where they put the same chip back into service and it didn’t fail for another year.
Evolution is a stochastic optimization process that leaps over local minima. A small nit I’ll pick is that the process is not ‘driven’, it is ‘pulled’ by the fitness function. The fitness function is the environment which is also evolving - prey, predators, climate etc.
There is an elegant pivot structure in the rear wing of dragonflies shown in the lower photo below. The elliptical pivot is located between two driving beams that move in independent sinusoidal patterns. The phase relationship between the two sinusoidal motions determines the wing twist that controls flight.
Googled images of dragonfly fossils show that 100+ million years ago they looked about the same as they do today. But a close up of the wing in the area matching my macro shows a different structure. It appears to be a a vacant area around a fold in the wing. That would work nicely because the vertical folds create stiffness and form the airfoil section. A very flexible pivot that was optimum for the time. Evolving prey, predators and atmosphere changed the fitness function and pulled a different solution. Hence the elegant pivot.
Spanwise stiff, chordwise flexible
When I contemplate the wonder of evolution, the thing that really strikes awe and more than a bit of sadness is all the death that was necessary over the eons to produce this conscious sapient point-of-view I call me. Not only me, but the comfortable life I live mostly free of the savagery of nature, red in tooth and claw.
I believe the nervous system has been evolving for 500 million years. Thats a lot of time to tinker with new designs.
Having said that, most of our brain growth occurred in the last 2-3 million years. Human brains went from 400cc to 1400cc. Our brain/body ratio changed dramatically and thats why we are a technological, global species now. So it took maybe 500 million years to get to a brain the size of a chimp, then another 3 million years to triple our brain size on top of that.
One thing to keep in mind is how effective technology is at augmenting these bodies and brains. Technology is intelligently designed, and look at all the ways medicine can augment the bodies healing process, or all the ways technology can augment our brains (communication tools, the internet, AI, etc).
Most of these technologies are 100 years old or less but they’ve compensated for a lot of failings of biology. Medicine is full of tools designed to compensate for the failures of biology. Biology is not that impressive when you realize how much time, money and technology has to be devoted to compensate for its failings and shortcomings. Especially when you realize intelligently designed technology evolves over decades, not millions of years.
Dogs have evolved rapidly in the last century or two due to selective breeding. Most dog breeds didn’t exist 200 years ago. Intelligent design occurs on a much more rapid timescale than evolution by natural selection. With intelligent design you can see meaningful changes in a species phenotype in around 10 generations of breeding.