It may seem obvious, but it cannot be said often enough: science is not merely a list of Things We Know.
Science includes all our knowledge about the physical world, and our understanding of its processes and mechanisms, but perhaps most important of all is the scientific method. It is the method of finding things out.
It involves rigorously questioning what one sees, watching it unfold while carefully observing and measuring it, describing the how and why without bias. Then letting others challenge our perception, and put our ideas to the test.
The scientific method also involves being aware of the errors that can creep into this process, and constantly testing our laws, theories, observations, and their accuracy. It involves being aware of the limits of this knowledge for a realistic grip on the Things (we think) We Know.
The history of the application of the scientific method also shows that the context in which we understand things changes regularly. One obvious example is the long-held belief of the Earth as the centre of the Universe which was later turned on its head.
Grappling with knowledge
So what are the unknowns and errors that we must be aware of in the scientific process?
There several types of unknowns. At TWDK, we will deal only with scientific uncertainty in an experimental result:
– The facts that we haven’t come across yet and whose existence we cannot predict or are unaware of – the Things (we don’t know that) We Don’t Know
– The gaps in our knowledge that we are seeking the answers to – the Things (we know that) We Don’t Know
– The inaccuracies in our existing knowledge – the Things We Don’t (quite) Know (even if we thought we did)
Delve deeper into types of unknowns.
Here at TWDK, we seek to deal with the second and third type.
Let us try and tackle the third type of unknown: the uncertainty implicit in what we do know, or rather, Things (we think) We Know.
How do we come to know?
We use experimental observation to measure what we see objectively and precisely. There are limits to the precision of our instruments, the techniques that we use, and even to our own objectivity. These shortcomings bring ‘uncertainty’ into scientific experiments.
To narrow in on reality, we must continuously improve our theories and models, instruments and experiments. To do this we must understand where these uncertainties creep in, and devise ways to overcome them.
Knowing what we don’t know
Uncertainty must be dealt with in each field, at each level, in its own way. There are several stages where it may creep in.
Assumptions may be have been made before experimentation. Delve deeper into Occam’s Razor.
Sampling errors may be encountered when dealing with a small sample (piece or amount) of something too large to measure all at once, such as a cup of water from a river. How can we be sure this sample is representative of the bigger picture? If this cup had a lot of pollutants in it, how can we be certain are we that the whole river does? Taking a bigger sample, or many little samples, can improve confidence in our results.
Any experimental observation will have an error associated with it. These errors may originate from fundamental physical limits that limit the accuracy of the measurement, instrument inaccuracy, deficiencies in the measurement technique, or inadequacies in analysis, among other factors. These may be represented as an error bar on a graph or a plot, or within disclaimers. The relative error, or the magnitude of the error (difference between the exact value and the appoximation) as compared to the magnitude of the exact value, is important too. Delve deeper into error.
The failure to replicate can raise its head, even for strong results. In such cases, replication of the experiment does not provide the same result (within the expected margin of error).
Readers are invite to tell leave comments and tell me if I’ve missed anything in the list above.
Dealing with uncertainty
It is very hard to maintain the awareness of these shortcomings routinely as some of this information is unavailable at times. To complicate matters further, popular science writing may not include any mention of uncertainties in findings at all. There are also examples of cases where statistics are deliberately manipulated to favour a certain result, or obfuscated to communicate a different inference.
Ultimately, all known scientific facts rest upon uncertainties. This doesn’t automatically mean they’re wrong, but it may be possible for us to improve upon them. This awareness is extremely important for scientific literacy. This is why there is a need for skepticism, and to make space in our minds for Things We Don’t Know.
There are things that we can do to minimise error in our observations or measurements. We’ll deal with this in another blog post.