Ay,
now the plot thickens very much upon us
George Villiers, 2nd Duke of Buckingham
There is a game you can play called bullshit bingo. (Hereinafter referred to as BB.) It involves cards similar to bingo cards but instead of numbers has phrases fondly used in business circles such as, ‘client focussed’, ‘top down’, ‘bottom up’, ‘exposure’. (So, for example, one might describe lap dancing as involving ‘client focussed, top down, bottom up, exposure’) Other favourite phrases one can encounter are ‘pushing the envelope’, ‘thinking outside the box’, ‘result-driven’, ‘out of the loop’. A very helpful website will produce random cards for you. Print out some cards, distribute them to friends and colleagues and the next time somebody gives a lecture you can all tick off the boxes. First one to complete stands up and shouts, “bullshit!”.
Of course there are a few terms that do not appear on this cards that I would like to see added, for example, theranostics, proteomics, in silico, and possibly even pharmacogenomics, bioinformatics and data-mining. In context, of course, these might have some content but in Sir Lancelot Pastit’s mouth at the annual general meeting of Pannostrum Pharmaceuticals they simply mean, “not many molecules in the pipeline I am afraid, so why don’t I give you some long words instead”. (I realise, of course, that pipeline is a candidate for BB.) However, one of my bętes noires, if you will pardon my French, or pet hates, si vous pardonnerez mon anglais, is the use of the phrase learning curve. A learning curve is a psychological term and plots knowledge gained against either time or effort. As I have said elsewhere, learning curve is the sort of phrase favoured by those who never got very high on theirs. No doubt psychologists have a use for this concept but it also seems to be a favourite in our marketing department, who clearly do not understand the term, and even some of our medical advisors have succumbed, as witness the following conversation with Dr Violet Shrink concerning our need to get to grips with proteomics.
Dr Shrink.
It’s a steep learning curve,
GMcP. Well, we have to be thankful for small mercies.
Dr Shrink.
Riddles again,
GMcP. Well statistics is difficult to learn and it is excellent but of course its excellence does not lie in its difficulty. No I meant rather that it was good news that we won’t have to spend much effort mastering it.
Dr Shrink. No, I said the curve was steep.
GMcP . I know. What does the word “bottleneck” conjure up in your mind in the context of road traffic?
Dr Shrink. (Casting me a knowing look and hauling out her notebook.) A spot where an obstacle or constriction of flow causes traffic to tailback.
GMcP. Exactly. And so the narrower the constriction, the smaller the bottleneck, so that you can’t describe a worse traffic jam as being due to a bigger bottleneck.
Dr Shrink. (Looking somewhat puzzled and doubtful.) I suppose not but I don’t see the relevance…
GMcP Now let us think about this learning curve. We are agreed, are we not, that we start, at time-zero[1] with no knowledge gained so that we start at what is called, graphically and mathematically speaking, the origin?
Dr Shrink. Correct
GMcP Now, imagine that we have learned a great deal in a very short amount of time. We have learned easily, so we shall plot a point that is not very far along the X axis but rather high on the Y axis.
Dr Shrink. Correct
GMcP. Now join the origin to the point we have plotted. What can we say about the curve?
Dr Shrink. It’s a straight line.
GMcP A straight line is a special case of a curve. That’s not my point. The curve, or if you prefer the line, is steep but we learned the subject matter easily.
Dr Shrink (Pondering a while.) I see, I think, but ‘steep’ means ‘difficult’ and I don’t think that it is right to change the meaning of words using graphs and in any case, where does the bottleneck come in?
And if you don’t believe me that there is a widespread confusion about this, type in the phrase steep learning curve on Google and see what happens. In fact, I think that the use of the phrase steep learning curve to mean the opposite of what it ought to mean is really a symptom of a more widespread phenomenon, which (another potential BB candidate here) I refer to as graphical dazzle.
Graphical dazzle is the phenomenon whereby certain persons think that their understanding has been illuminated by a graph when it has merely been blinded by it. The graph gives them the illusion of their having received wisdom, insight and understanding when all that it has done is increased the store of things they don’t understand. They nod wisely but would be incapable of explaining what the graph means. My favourite examples of graphs that achieve this are what I refer to as ‘fox and whisker plots’. (Not to be confused with box-and-whisker-plots which are simply data-displays that 99 times out of a 100 would be better replaced by a dot-plot. Box-and-whisker plots can be filed under ‘useless but harmless’.) A fox and whisker plot is one whereby separate time courses giving the ‘response’ for each treatment are plotted and a vertical whisker is added to each point to represent uncertainty. Uncertainty is certainly what it represents and the following are some points I personally am always uncertain about.
Ř Is the standard error plotted, or is the width of a confidence interval represented?
Ř What formula was used to calculate this? Is it the formula that applies, for example, for the standard error of the mean from a simple random sample?
Ř Is the person who calculated the length of this whisker stupid enough to believe that clinical trials use simple random samples of patients?
Ř Since simple random samples are not taken, what does anybody think these whiskers represent?
Ř What contribution to the primary purpose of running a clinical trial, which is to compare treatments, do such graphs make?
Ř What sort of a learning curve is the person who plotted the graph on?
I am always struck, however, as to how much more intelligent the audience is than I am. People seem to enjoy seeing these graphs without being able to answer any of my questions. However, in my opinion all that the graphs have done is increased the level of confusion in the audience, so that truly one may say, “the plot thickens”.
Stop Press: Commotion at Pannostrum AGM
Brushing aside hostile questions at the Pannostrum Pharmaceuticals AGM about drug discovery pipelines and personal remuneration, Sir Lancelot Pastit gave an upbeat assessment of the prospects for the coming year. He laid great stress on the valuable and exciting work, as he put it, in pharmacogenomics and theranostics that the company was engaged in. Unfortunately, he was unable to complete his speech, as, having just mentioned that it had been a steep learning curve, he was interrupted by obscene references to bovine manure from several unidentified sources in the auditorium.
Return to Guernsey McPearson’s Prose Selection
Return to Guernsey McPearson’s Homepage