Atheism, the non-existence of God (Gods), has a wide variety of meanings. Here I use the word “Atheism” to refer to “strong atheism”, or “methodological naturalism”. It asserts the non-existence of supernatural beings and supports scientific methodology as the only effective way to investigate reality[1].
I’m not in a position to judge whether it is right or wrong. I just hope to raise some of its fundamental assumptions, benefits, and limitations.
Naturalism relies entirely on scientific method to investigate reality, which includes the making of a hypothesis, the prediction of a possible outcome, the test of such a prediction, and repetition of the experiment[2]. To me, the most interesting method is “repetition”. This makes science a universal language. Everyone believes in science. If you don’t believe in a theory, go ahead and do the experiment yourself. You can either disprove it or support it from the experiment. Here I use “support” because scientific theories can never be “proved” correct. Maybe someday a better theory will replace the current one.
However, “repetition” has several limitations. First, it requires the experiment to be repeatable. That is, whatever the “value” the experiment intends to test, it must either remain unchanged, or follow a predictable pattern over time and space. Some of the physical values fall into this category and can be tested repeatedly, but some may not.
A simple illustration of this limitation is as follows: A grown-up person Tom asks a child Jerry to investigate a table in an empty room, and asks Jerry to write down the position of the table in the room after his investigation. Note, when Jerry performs the investigation, he is alone in the room with the table. He cannot communicate with anyone (including Tom) his discovery. Only after he leaves the room can he disclose the position of the table. However, as soon as he leaves the room, Tom (or some other people) goes into the room from a back door and changes the position of the table (assume they can do it quickly enough). Because the test should be “repeatable”, Jerry does multiple tests and find the table at a different position each time. Even though each time Jerry correctly writes down the position of the table, the best conclusion Jerry can get is that the table appears randomly in the room.
Fortunately, most of the physical values at the macro level are fairly stable. They either remain constant or evolve slowly enough that people can derive the difference of the tests using known knowledge. (However, we cannot guarantee it will remain this way.) When we go into the micro world, things are totally different. Quantum mechanics is based on the experimental discovery of the “uncertainty principle”. The theory gives up identifying the exact position and momentum of a particle, but focuses on their statistical relations “over time”. Can we test the position and momentum at every single time? Maybe we can, but we get different result in every experiment (think about error!). We are just like the child Jerry in the example above, unaware there is a Tom there changing the position and momentum each time. No wonder Einstein questioned: “Does God throw dice?”
[1] Wikipedia: Naturalism.
[2] Wikipedia: Scientific method.