Read the above post carefully:
"Life simply evolved without fixed boundaries to its layers of abstraction."
Biological life absolutely does have many layers of abstraction. Many of layers of abstraction can appear to be fairly reliable for most people. This indeed is necessary for life to function. The problem is that there alway a chance of a given boundary not respected for some people and some processes. We have a brain and we have an immune system. They intersect in a number of broad ways - they're both information processing systems. Skin, organs and fascia are distinct to an extent but there's odd intersections. Some animals can generate vitamin D internally. Human beings must generate from sunlight on their skin for whatever quirky reason of evolution. Some people actually have their Thyroid gland on the bottom of their tongue but for most people it's at the base of the neck.
Similarly genetic coding is definitely not random coding. It's very tight, highly optimized coding but coding that doesn't necessarily have reliable boundaries and abstractions. Functions? Objects? Variables? Why should those exist in these circumstances.
I'm not saying it's total chaos. Indeed, A biological system can seem simple enough. The problem is these system are just chaotic enough on enough distinct logical level that engineering principles break down as they are applied to create complex systems for diagnosing and curing people. Suppose you have ten generalizations. Each is true for 90% of the population. Suppose you have a system that depends on all of the generalization being true. The system work for only about 30% of the population. The problem is raising reliability past 90% on any one generalization becomes very hard because you outliers on every single system. Suppose in a different population, some of the generalization are true 80% of the time rather than the 90% of your original population - your system becomes less reliable without it being terribly obvious why. Biology involves more paradoxical statistical effects than anywhere else - that's only the part why you can reliable statistics.
Once you look this stuff, you can go from thinking you'll get a complete handle on a given process to realizing it can only be partly dealt with by ordinary engineering - it certainly can be partly dealt with. Adding ten more assumptions to our original system would make it completely unreliable. Given that each step is uncertain and failure is costly, there's an incentive to make each step manual. The amount of human done, manual labor in health and biology makes every step from research to implementation literally a thousand times more costly than something like chip design. Of course progress is being made but its within the context of these paradoxical effect. I suspect that synthetic life, nanotechnology and better AI will be needed before things get better.
There's video on synthetic life that discusses the genetic-code-as-assembly-language stuff.