“Not everything is possible and truth does not depend on perspective”​

I start this article with a strong opinion. You may say I am an opinionated person and I may agree with you on that. But can we decide who is right and who is wrong or we will remain in an endless discussion about the topic? One that can end up we both of us celebrating diversity over a toast or killing each other because of our differences?

I could probably prove to you that I’m right, but I know that words have a precise meaning and I hesitate to flexibilize the meaning of the word ‘proof’ just to fit my narrative. Proof is a strong concept that can be achieved only in mathematics.

“In math, the concept of proof is much more rigorous than what we use on a daily basis and even more precise than for physicists and chemists” says Simon Singh in ‘The last Fermat’s Theorem’. “A scientific theory can never be proven in the same absolute way that a mathematical theorem. A scientific theory is merely considered highly probable, based on the available evidence. […] the scientific proof depends on observation and perception, both fallible, providing only approximations to the truth.” 

Simon Singh in ‘The last Fermat’s Theorem’

Physics has shown us that measurements inevitably carry uncertainty. But while we cannot completely eliminate uncertainty, many times we can reduce it more and more and more, until it gets so small, that it doesn’t matter anymore for the purpose of our daily lives.

For example, the precise weight of a measure of coffee, to the fifth decimal letter, may be very, very hard to determinate. However, because a change even in the first decimal letter on the weight of a coffee spoon doesn’t change to the taste of my expresso, to my daily life experience, this uncertainty on the weight is irrelevant. It doesn’t make my expresso experience less tasty, less real or less reliable.

Scientific truths may carry some uncertainty, but are extremely reliable. Gravity, the speed of light, the exchange of heat, and many more (the laws of physics), do not depend on species, race, location, origin, weather, intelligence, creed, alcohol or anything else that our brain categories can create.

The degree of certainty of a given truth is based on the number of times that this given truth has been observed. How many times an apple falls from a tree? Many. Many, many, many times. Actually, every time it has been tested. How many times it has floated in the air? None. Not one single time. So, it is possible that one day we will see an apple detach from a branch an float in the air? Well, it may be possible, but it is extremely unlikely.

Likelihood has also a defined inflexible meaning. It is an statistical term that identifies the probability of something happening based on the number of times that it has already happened. The likelihood of the sun rising tomorrow is so very indescribably high, that we could even guarantee that the sun will rise tomorrow.

Again, likelihoods are calculated based on observations. Every time that we asked if the sun was going to rise, it did. With our intelligence, we have taken the observations and extracted patterns out of it. With these patterns, we have created models that help us predict future events, like the next sun rise. Nowadays we know a lot about of the universe, but it all started with simple observation of an event and collection of the outcome of this observation. What we call, data.

The only thing that can help distinguish a probable from a possible event, and establish the degree of confidence on a particular truth, is data.

Data has qualities. Four come to my mind now: replicability, precision, accuracy and abundance. Replicability means that provided the means of observation, other people have to be able to collect the same kind of data. If I use use a ruler to measure things, you should be able to use the same ruler to measure the same things and get the same results. Precision means that different measures of the same thing, with the same tools, should give similar, very similar, outcomes. Accuracy means that the data has to resemble reality and if I know that I’m measuring the volume of 1 L of water (which I check by a different measurement, like weight), I should get a read of 1 L. Finally, abundance is important because fewer data does not allow for reliable conclusions (the statistical theorem of small numbers can tell you why, so I won’t do it here). Data can be judged and it is possible to separate good from bad data based on these criteria.

Patterns behind data observations allows us to create models, another important tool to reach the scientific truths. But also models dependent on data. “Shit in, shit out”, you may have heard before. Even the best of models provide shit conclusions if you feed them with bad data. They are able to reach a conclusion, but not the truth, without good data.

The biggest enemy of the truth is bias. People love their pre-concived ideas and will fight for them bravely.

The only way to fight bias, is with data. Yes, you can select data with bias, which is the worse (and unfortunately frequent through out history) of all crimes a scientist can commit. But biased measurements usually cannot be replicated by another observer and frequently more data closes the disputes.

To try to disqualify scientific truths by arguing on philosophical bases our inability to achieve proof of the truth, do not make you look smarter: it makes you look unable (or afraid, or lazy) to acquire and analyze the necessary data.

If no data can change your mind, you are not wise, you are a prejudice fool.

Originally published on LinkedIn in February 2017