Natural selection may not select for simplicity. It may not try to reduce dependencies. A gene can depend on thousands of other genes. And a gene can carry out multiple functions
This one's pretty straightforward; natural selection has two very common approaches to dependencies:
1. If the presence of something is not a reliable feature of the environment, dependence on that something will be ruthlessly weeded out.
2. If the presence of something is a reliable feature of the environment, dependence on that something will be created and grow to permeate most existing systems.
The first effect has a lot of popular awareness; the second doesn't. They're both important.
Edit: Also - https://en.wikipedia.org/wiki/Radiotrophic_fungus
The fact that it is regulated to a margin of 0.3% shows quite how important it is. Increase or decrease the body temperature by 5% (on an absolute scale, so Kelvin), and you will be a dead human...
Cancer or diabetes, no. Diabetes is not even a "disease" in the sense of an organism that can be alive or dead.
This is an excellent point. It can be applied to a huge number of biological phenomena, but it almost never is.
An example where this idea actually has made it into the mainstream is body weight. I read an article that pointed out that, yes, we're fatter than we used to be, but regulating body weight by measuring energy intake and expenditure is doomed to fail. Articles making this claim are a very rich genre, but I found the particular argument fascinating:
People routinely maintain the same body weight, within say 5 pounds, for years at a time. This is a level of accuracy that we do not have the technology to achieve by using measurements of energy inflow and outflow -- we can't measure dietary energy that precisely! (And modern technology is, by most standards, very precise.)
Thus, there is a very strong suggestion, based on this incredible stability, that your body is regulating itself to a certain weight. If you adjust energy intake, or outflow, other things will adjust accordingly; those are unlikely to be primary determinants of the system's behavior.
If a reliable feature is evolutionary advantageous, then new systems will come about to take advantage of that (and may out-compete old systems, indeed, to the point of extinction), I agree.
But, if a reliable feature isn't evolutionary advantageous, it can be ignored or even removed.
Negative features may need to be managed, rather than taken advantage of. Examples: ionizing radiation, landfill.
I'm not necessarily arguing it's untrue, just not quite the hard-and-fast rule that 1 is.
This is pretty subjective. If you grow up with sufficient dietary iodine, that's worth 10-15 IQ points (a gigantic amount) compared to the counterfactual. Iodine supplementation is highly effective in populations that have presumably been going without for, at the very least, many hundreds of years. This is the reason we iodize salt.
With that in mind, how much of a hard-and-fast rule is #1?
The other side of this argument is that dependencies can be surprisingly subtle. This is pure speculation, but the thing that leaps out at me about atmospheric nitrogen is that oxygen is extremely corrosive. If the atmosphere were more oxygen and less nitrogen, lungs of today might experience a lot more wear and tear than they are currently designed for. (Or not! I really have no idea. But I think the hypothetical has some thought-experiment value anyway.)
Something is only weeded out if it is sufficiently disadvantageous. There are lots of examples in biology where mutations are neutral and allowed to accumulate. There is also a lot of junk DNA, and even though we've found that some junk DNA was misclassified, there still is a lot of junk DNA in each genome
For example cobalt and nickel were both essential for the first living beings. However, on the continents they are much less accessible than in the oceans.
Because of that, most terrestrial plants have lost the dependency on cobalt (which is why they cannot provide vitamin B12 to those who eat them) and most have also lost the dependency on nickel (except that for many terrestrial plants nickel, while no longer strictly required, can still be useful by enabling them to use urea as a source of nitrogen, when other sources of nitrogen are scarce).