Check assumptions before ruling things out.

Sherlock Holmes is a famous fictional detective renowned for his
logic, deduction and sharp intellect, analyzing apparently intractable data and ‘diagnosing’ the likely perpetrator of the latest heinous crime: he’s up there in the pantheon of fictional detectives along with Hercule Poirot, Philip Marlowe, Jules Maigret and the rest!

One of Sherlock’s famous aphorisms is [1]:

When you have eliminated the impossible, whatever remains however improbable, must be the truth.

This sounds eminently sensible, as ruling out impossible things should be easy, but we do have a small problem… how can we be sure what we are ruling out is, in fact, impossible and not just an assumption of impossibility? For example, in the world of transformer test and assessment, I’ve heard webinars relating to online bushing monitoring where the host states with authority that it is an impossibility for insulation power factor to decrease… and ultimately to go negative. In fact, this is just a commonly held assumption as the works of Denis Kopaczynski and Long Pong demonstrate [2, 3].

I recall from many years ago, testing an old 50 MVA distribution transformer in the UK, and finding that the three bushings had power factors of 0.4%, 0.0% and -0.4%… which was disconcerting as at that time I was working under the assumption that power factor can’t be negative. I even followed the ‘Lachman Rule’ and put all the test gear and shorting leads away, then got it all out again to rerun the tests only to get exactly the same results. I called colleagues for a consult, got updated on negative power factors in practice, and the transformer went back in to service without issue, noting that the bushings likely needed changing ASAP. As ever, check assumptions before ruling things out. Another well-known Sherlock aphorism is [4]:

It is a capital mistake to theorize before one has data. Insensibly one begins to twist facts to suit theories, instead of theories to suit facts.

This quotation also sounds eminently sensible until, once again, you dig a little deeper… how do we know what data to collect if we don’t have a theory to work with? But this one is actually, to a degree, supported by the well-known financial wizard Warren Bufffet, as he has several relevant quotations [5]:

What the human being is best ad doing is interpreting all new information so that their prior conclusions reamain intact.

A great many people think they are thinking when they are merely rearranging their prejudices.

If a person is to be biased, he is better to be biased by facts than by prejudice and fantasy.

Confirmation bias, where we seek out supporting data and disdain contradictory data is one of many biases we humans are subject to – but being aware of the biases is a step towards countering them. I’d suggest Feynman likely had it best in a commencement address at Caltech in 1974 [6]:

… have ideas, but don’t hold on to them too strogly… rigorous self-doubt and self-testing are required.

Confimation bias, where we seek out supporting data and disdain contradictory data is one of many biases we humans are subject to – but being aware of the biases is a step towards countering them.

One of the effects of bias, of drawing conclusions based on limited data, or just ignoring data altogether relates back to Conan-Doyle, the creator of the Sherlock Holmes, who was not only highly logical, but was also a fervent spiritualist and believed strongly in the supernatural. He was asked to review some photographs taken by two young girls between 1917 and 1920, photographs which purported to show fairies in their garden [7]. He championed the photos as genuine evidence of the supernatural! It was more than 60 years later that the girls admitted the photos were fake, and the ‘fairies’ were paper cut outs of drawings copied from a children’s book, but by then Conan Doyle was long since dead. It’s a surprise – to me at least – that the creator of Sherlock Holmes could be deceived by some photographs. But in another case involving his friend Harry Houdini, he flat out refused to believe the available data set before him: Houdini demonstrated an illusion where, live on stage in front of a paying audience, he seemed to walk through a brick wall [8]. Even when Houdini showed Conan-Doyle how the illusion was created and performed, involving a large mat on the stage and a hidden tunnel under the wall, Conan-Doyle continued to claim that Houdini could, in fact, dematerialize and then rematerialize on the other side of the wall. His adamant approach led to a significant disagreement between the two gentlemen and the end of their friendship.

There are often challenges in dealing with data collection, data analyses, and in ensuring our assumptions and biases are identified and addressed, and in accepting that we may have to change our mind. I like to think that I know a few things related to condition monitoring and transformer test and assessment, but I also know that there is nothing wrong with a ‘consult’ and a discussion of results and analyses with someone who may have a different point of view.

I recommend questioning data, analyses and conclusions frequently – even when it seems an obvious trail from data to conclusions… maybe especially when it’s an obvious trail! But I’d suggest the last word belongs to Feynman (6)

The first principle is that you must not fool yourself – and you are the easiest person to fool.

Sources:

Public references supporting key technical and safety statements in the article:
[1] https://www.bestofsherlock.com/top-10-sherlock-quotes.htm#impossible
[2] “Negative Power Factor of Doble Insulation Test Specimens (An Analysis)”, D. Kopaczynski, S.J.Manifase, 55th International Conference of Doble Clients, Boston, USA, 1987
[3] “Review of Negative Power Factor Test Results And Case Study Analysis”, L. Pong, 70th International Conference of Doble Clients, Boston, USA, 2002
[4] https://www.goodreads.com/quotes/92128-it-is-a-capital-mistake-to-theorize-before-one-has
[5] https://www.goodreads.com/quotes/8956691-what-the-human-being-is-best-at-doing-is-interpreting
[6] https://calteches.library.caltech.edu/51/2/CargoCult.htm
[7] https://en.wikipedia.org/wiki/Cottingley_Fairies
[8] https://wordsworth-editions.com/houdini-and-doyle-a-modern-ghost-story/