1) Science is not atheism. Atheism is not science. Although science is atheistic, the evidence is overwhelming that there are no gods behind it. Not the least that the same experiment whether done by a Shinto priest, a Buddhist monk, a Rabbi, an Imam, or an Evangelical Protestant Fundamentalist will get the same results. Nylon clothing is no more special if it's on an atheist, deist, pantheist, panentheist, monotheist, or polytheist. There is one science, that is blind to both the faith of the scientist and the faith of those who use the science. It does not need gods, so is without gods, so is atheistic.
--> Beyond that, there is no other necessary link between science and atheism.
2) Evolution is well-observed.
The primary principles of Evolution:
* Living things have variation
* Some variation allows certain members to produce more than others.
* This beneficial variation is passed to the offspring
If you are instead objecting to "macro-evolution" I have three points for you
* All macro-evolution consists of is many micro-evolutionary events accumulating. There is no barrier between the two
* Macro-evolutionary events in the form of speciation events have been observed in nature and in the lab.
* Beyond 'speciation' we have the growth of new organs as well as the fossil record and DNA sequencing to show the progression and formation of new 'kinds' from a common ancestral species.
3) "That's like believing Google popped into existence and that people did not develop the code for it"
--> As opposed to non-self-replicating code, life is 100% capable of self-replication. In fact that is the sole defining feature separating life from non-life.
--> But excepting that, Google didn't make the changes in one giant swoop. It slowly adapted. It added small features by making minor modifications to its code, adding bit by bit over time to satisfy consumer demand and survive.
---> As to "coding randomly changes for the better without programers changing it", you have never designed a web page. Most of the work for developing a website is literally adding small changes to the code and seeing the effects of these changes. Good effects from the RANDOM yet functional coding go on to stay in the code. Bad effects are removed.
And before you complain, the fact is that the code changes are 'random' because the effects are not known beforehand. See, for DNA, if I were to ask you to guess which exact base-change would be the next to occur you would have only 12 options to choose from. Adding insertion and deletion options for one- and two-base groups, that is still only 16 specific mutations. It's not "random" in the sense of hitting a keyboard with your face and submitting that code. It's "random" in the EXACT same sense of rolling a die and changing the operation to another operation.
And that type of change, while not always OPTIMAL, will often enough give something BETTER than before. Also, most of these changes will be NEUTRAL so they will probably be kept in to experience a change for the better at some later date. And this is exactly what we see in nature: neutral and better. And we rarely see optimal unless it comes from a line of 'betters' upon 'betters'.
So your own example, when the details are applied, disproves your position.