Scientific models can accurately predict behaviour of what we have measured. Popular science programmes imply that we can assume that what we haven’t measured is equally predictable.  Are they correct?

We can record data from all sorts of events.  A man walks to work, we can record how far he has got at which time, and we can plot a graph of it.  We can then derive an equation to fit a curve through the data points that we have recorded. Programs like excel do this automatically.  Some equations will fit the data very badly; others will match each point of data exactly.  So we now have some equations that match the data, but those equations do not predict what happened before and after we recorded our data.  They do not predict how the man got up and walked around his house before leaving for work, or how he sat on his chair for three hours before walking to get a coffee.  In this case it is easy to see that the equations are only valid as a model for the data that has been recorded.  We would be completely wrong to use them to predict all the walking that the man does in his life.

We have all seen a graphical representation of a sound.  Whenever any sound or noise is recorded it can be represented by a graph.  Once we have the raw data, it is possible to define equations that describe the shape of the data.  This is known as Fourier Analysis.  So, we have our raw data, and we have our equations, and we can find that the equations almost perfectly match the behaviour of the raw data.  In our Fourier analysis, we can take a short stretch of  completely random signal, and we can analyse it and model it with equations that match it almost perfectly.  But if we try to use those equations to predict the precise signal in the section of the noise before or after what we have analysed we will get completely the wrong answer.  The sort of shape will look similar, but the detail will be completely different.

Both of these are examples of what science does.  It records data and then it determines equations that match the data that has been recorded.  We use these equations to immense practical purpose and most of the time they hold true.  When measurement doesn’t match the equations then we tend to dismiss the measurement as faulty.  Nobody would believe me if I claim to have invented a perpetual motion machine!

However, we must recognise that we may simply be in a short stretch of ‘white noise’ and it would be bad science and bad logic to insist that our equations hold true outside of the domain in which they were developed and tested.  Commentaries about potential other universes,  events before Big Bang, or even events in the distant past of our own universe or planet fall into this category.  It is not an act of science, but an act of faith to assume that the behaviour of the material universe has always been and always will be the same, as that which we see today.