So if another arm or leg is ridiculous and not part of evolution, then why do humans have 2 arms verses just 1? Or most animals have 4 legs versus just 3? Why "just" 2 and not 3 arms or "just" 4 and not 5 legs? Why do some species have 2 eyes while others have more? Couldn't an extra 3rd arm in a human be selected for if it proved advantageous within a certain environment? I think you need to analyze your story a little closer.
Good questions, Seth, but the answers are not simple. It is to do with developmental biology and the way that bodies form. Vertebrates are constrained by their evolutionary past. I have written the following to help me understand the concept but as it was written purely for my own purposes, the explanation is rather clunky:
Identity gene produces master protein, which determines the type of protein being expressed. If the master protein is present then the interpreting gene gets switched on as the master protein binds to the regulatory portion of the interpreting gene.
The identity genes are the homeobox genes.
To make a protein a piece of DNA is required which we call a gene. The gene trancribes a protein, but what switches this piece of DNA 'on' to start it transcribing? There are two parts of the gene, the regulatory part and the transcription part. If a protein attaches itself to the regulatory section then that can act like a switch. Some attached proteins switch on the transcription, and other times they switch it off, depending on the nature of the genes involved. The switch protein attaches to a 'binding site' which it has a certain specificity for. The level of specificity varies, so that some switch proteins bind strongly to the binding site, whilst others bind weakly.
NOW, Take an embryo of a fly. It starts as a single cell, and then it starts changing. The nucleus divides and produces two. Both of these now divide, producing four, and this continues, so that at the end of this process there are many nuclei in a single cell. Of course, in each of these nuclei there are all of the genes for the whole organism.
The cytoplasm that the nuclei exist in came from the mother, and at one end is a source of a protein, this is transcribed from RNA passed to the cell from the mother. Let us call this protein 'A'. This means that there is a gradient of this protein through the embryo, with less of the protein appearing furthest from its source, as would be expected.
What does this protein do? It switches on the production of other proteins. We can imagine a gene which has a high affinity for this protein being switched on everywhere, and so a second protein appears, which we can call 'Z' and is found everywhere in the cell. We now have two proteins, a gradient of the first and a general diffusion of the second.
Let us now introduce a third protein, 'Y'. This one has a lower affinity to 'A', but will still get transcribed everywhere, and so there will be a gradient of proteins, protein 'A', 'Z' and 'Y' will be found near the source of 'A' whilst mainly 'Z' & 'Y' at the further end.
Now, if 'Z' & 'Y' compete with each other, such that if 'Z' is present it attaches itself to the regularotory region of 'Y' and switches it off and vice versa an interesting thing occurs, especially if 'Y' has a higher affinity to 'Z' than 'Z' has to 'Y'. Where there is a lot of 'Z', it will switch 'Y' off, and where there is a lot of 'Y' it will switch 'Z' off, but if all other things are equal 'Y' outcompetes 'Z'. There will now be a band furthest away from the source of 'A' which has 'Z' in it but elsewhere there will be 'Y'.
We can take this to the next stage, where another protein is considered, 'X'. If the same criteria are applied to 'X' as were applied to 'Y' so that these two compete like 'Y' and 'Z' did and if 'X' has less of an affinity to 'A' then there will be a section of 'Y' and a section of 'Z'.
Repeat this several times, and there will be a series of bands of proteins, with edges that are slightly blurred between each other which are ultimately based on their position in relation to the source of 'A', from 'S' to 'Z'. If each of these bind to another gene then in each of these bands a discrete protein will be found, and this will be found in definite areas, with none of the blurring. We can call these proteins 'L' through to 'R'.