The profits of the null hypothesis.

The null hypothesis has a useful property – it can never be proven.  Test results either reject or fail to reject the null hypothesis, they never prove the hypothesis.

The science of statistics allows companies to test products and make claims as to what the failure rate will be for a product even though no one can know which one example of the product will actually fail when it is sold. Legislation, regulations and economics will set the standards for what testing is done and so the failure rate but the mathematics wasn’t the problem if your purchase fails even though the mathematics predicted that something could fail.

For most people this means that when they use a product then it will more or less work as described but you can never be certain of this. There will be failures and the failure rate can be very predictable across a batch of products but that is of little comfort to you with your own single faulty example.

The best that can be is that you can be proud of the fact that your broken product is but a reflection of a fundamental property of this universe in that things can be both random and yet predictable at the same time.

A good example is the predictable nature of radioactive decay rates: Whilst we can use the regularity of radioactive half-life to measure long term events these predictable measurements are made up from the sum of many unpredictable events of when the atoms of the material actually decay to the fission by-products. This underlying random nature of our universe means it is not a trivial dichotomy of determinism or indeterminism, of predictable or unpredictable, but it is both of these at once.

The same logic of predictable and yet random also applies to lottery results (because lotteries always, or should always, use natural events to add randomness to their draws) in that when the distribution of results deviates from an expected mean then the mathematicians that are hired by lottery companies to ensure that the results are random become suspicious of the draw machines. The mathematicians do not know what the draw results will be but they know what the distribution of results should be.

What does all this mean ? Well the implications of what is in this one post are as profound as your imagination can fly.

At the simplest level then stuff can always break and there can always be catastrophic failures. Risk of failure can never be removed from a system. It can be designed around but there will always be a residual risk that cannot be removed.

Whilst you probably will not win the lottery (though you may win) if something tries to influence the lottery draw results then this would probably show in the mathematical distributions of the results. Something could alter the time you bought a ticket if that time altered the numbers you picked but it could not reasonably alter the draw results. So though god, for instance if such a thing existed, may know the future, it cannot actually do anything to the draw itself that is natural without the mathematics of statistics showing this influence.

Just as with the lottery draws where you can have a predictable mean result if you average the random events that are at the foundation of the draws then equally in the natural world random events average out to appear stable. The weather is a good example and equally interesting is our influence and thinking about this. Though many see a single random event (a butterfly wing for instance) as a potential trigger to cause a vast perturbation in a metastable system, natural systems (such as weather) exist because they are metastable and absorb such random fluctuations to a mean.  The problem is that the closer you look at the effect of the butterfly wing then you could envisage how the tiny energy of this wing is added to trip a metastable system out of its current state to find a new metastable state.  The false thinking here is presuming that the metastable system is actually at a “saddle” whereby a small nudge would break the last remaining bond with the metastable state and the system would re-arrange to a new configuration. This is giving this one input far too much weight. That said unless you knew exactly all the bonds that held a system into a metastable configuration it is presumptuous to not think that a new input could not have an effect though it is improbable if the energy is many orders of scale smaller. Understanding how nature works is understanding how anything can exist when the origins are absolutely random. Be it physics, biology, chemistry or the weather, the secret to what exists is to understand how the metastable system you are studying has evolved.

Going back to the null hypothesis this also means that there is probably no-god though equally it (well one or all of them) may also exist. There can be no certainty here either way but without evidence then the probability is with the no-god. Now this is a fact but people’s reactions to this fact is a bit like the lottery in that though the probability is that there is no-win, people will still bet because there is a tiny chance that there is a win. Probability is that there is no-god but people bet on there being god.  It’s not really a matter of  faith but of our  human nature of being poor with statistics.

Some misunderstand how to use the null hypothesis. For instance Christian Piatt has a ridiculous presentation of this here on the Huffington Post in which he considers that if god is a hypothesis then atheism is the null hypothesis. As everyone says, the null hypothesis cannot be proven, then for some reason Piatt thinks that the “null” when the problem involves his god is “atheism” and so ergo “atheism” cannot be proven. Piatt’s god is not the only game in town. There are many gods documented throughout history. Equally there are many other supernatural entities that are not gods as such that are also recorded. Finally there is the honest default position for the Secular Humanist/Methodological Naturalist which is that the events are all naturally occurring and easily explained without pulling a genie out of a bottle. We must assume that Piatt’s poor understanding of the mathematics of the null and alternate hypothesis and ignorance of the competition in the marketplace of gods simply means this isn’t his field, which appears to be selling Jesus to people who probably could do better to think about reality for free. I wonder what creative commons license Jesus would use ?

If your null hypothesis has anything to do with the supernatural then you are not going to get any results that are proof but if the tests are constructed right then you should end up with statistically significant results that are unlikely to have occurred by chance either one way or the other. To date the results are significantly against there being any supernatural entities be it angels, demons or gods. Human nature unfortunately also attaches an unfair weight to the subjective reports and historical results that have a positive claim that supports what they wish for. One tabloid newspaper claim would overcome any number of scholarly papers that showed no statistically significant results. In other words, because lottery draws do not list every result but only advertise the winners, people expect they could win when in fact they should expect to lose.

This discipline of science and statistics is a fairly modern invention in the terms of human history. In time it may be that knowledge of statistics becomes our cultural norm in the same way that today historical subjective reports carry a lot of weight to the average person. But I honestly do not hold up much hope for this because I envisage that whilst we remain as a superstitious humanity then even in a 1000 years time people will still be buying lottery tickets and will still be ticking a box that says that they  believe in God because the residual risk of there being God never goes away and I envisage that there will be religions that will still exist to sell them the tickets.