You don't understand the power of selection. Take a few billion bacteria descended from a single genetically defined progenitor and treat them with a toxin or withold a vital nutrient and you will likely obtain colonies descended from single mutant individuals that can deal with the toxin or synthesize their own previously required nutrient. The genetic change occurred randomly in the large population but the selection pressure (i.e., toxin or lack of nutrient) caused the mutant strains to replace entirely the ancestral type.
You also need to think differently about your computer program example. Computer codes like DOS and Windows are abstract instruction sets and, unlike the genetic code where every sequence of three bases has potential meaning, this will not be true of computer codes. Additionally, nonsense sequences in the genome are effectively ignored or serve a merely structural function. The first time a computer command mutates it comes to a dead stop and has to be fixed by the programmer. The programmer may fix it by restoring the original command or he may replace that command with a new one he took from a different place in that program or from a different program entirely.
The programmer is merely the agent of selection. He is presumably motivated by a desire to make the program more robust or to do more things faster and more easily but he must always keep the program working. He is not an omnipotent designer but must work with what he has learned and may use trial and error approaches extensively at least until the code has evolved to the point where accummulated conventions prohibit innovations to the basic program. At that point modular programming predominates with whole sections of code being adapted for use with existing programs (i.e., add-ins). As the constraints of speed, memory, user interface, and needs change the programs will change too. Not to mention the fact that poorly written codes are abandoned in favor of better ones due to selection in the marketplace. Randomness and selection exist in the process and it matters not at all that humans are making the selections. All that is necessary for you to do is recognize that selection occurs in nature.
Your claim above is entirely false. Computer instructions are always broken down to their most basic instruction sets (machine code in Binary / Base-2), either by other software or by the CPU itself internally. The only abstract part of computer codes is that programmers usually prefer to view and work with abstract instructions rather than in Binary code. At the Binary (Base-2) level, every possible machine code has a specific meaning. This is identical to DNA codons, except that DNA is broken down to a Base-4 level.
" Additionally, nonsense sequences in the genome are effectively ignored or serve a merely structural function. The first time a computer command mutates it comes to a dead stop and has to be fixed by the programmer."
Again, your claim about computer code is completely erroneous. A computer command can mutate into another computer command and cause unintended behavior, but that does NOT mean that the program will immediately stop all execution.
Nonsense. The programmer is the intelligent designer, not an unaided, random, "natural" selector. The programmer is to programs what God is to Life.
You obviously know very little about programming. All computer languages eventually convert their instructions into machine code. These are a very small set of instructions each of which has a specific purpose such as to move a number from one place to another, increase the number by one or decrease it by one. It is indeed very similar to the DNA instruction set.