A tense exchange

Time for another lazy excerpt from private correspondence! This time we visit that most viscerally thrilling and scientifically crucial of subjects: what tense(s) to use in your scientific paper. Daring, I know! But surprisingly controversial, and I'm motivated to write it after reading and reviewing umpteen notes, drafts, and published papers in which the tenses seem (to me) perverse. In particular I think there's a need to write such a thing after being told by one physicist "I think there's a convention in science writing that we always use present tense". Piffle!

I think this sort of view, which is not uncommon, is an interesting phenomenon -- junior physicists read badly written documents, then after initial recoil they rationalise the bad style as "how it's done in science". Later then internalize and view the weirdness to some degree as a badge of honour; and finally they become irrational defenders of the spontaneously arisen faith. The same sort of thing can happen for oddball journal style rules like using "Section" if the first word in a sentence but "Sect." otherwise -- such rules are different for each journal, of course, and if you ask me it's the journal staff's responsibility to enact their own crazy rules if they think that's a good way to justify their continued existence. Bah, humbug. Some people seem to consider these arbitrary rules -- many of which are hang-ons from the utterly irrelevant game of printing and binding scientific articles and stacking them on the physical shelves of an actual library -- as important as the science being described. ATLAS has a wonderfully huge collection of style rules which I suspect do slightly improve the quality of our output, but in inverse proportion to their level of detail. Most fresh in my mind is the many months of collaboration review of the new ATLAS underlying event (UE) measurement paper, where I was the main editor: the internal review process absolutely improved this paper, but I promise that it wasn't the endless text iterations that made the difference. Naturally I anticipate with bated breath the journal reviewers' expert insights on how many commas there should be, and where they should go!

I recall a course taught in Cambridge many moons ago on experimental methods. It was rather a mixed bag: some half-arsed stats here, something about op amp circuits (really!) there, and finally something about how to write a paper. After spouting the usual rigid stuff, the lecturer finally gave up. "Just read Jane Austen until you know how to write", he implored. I take issue with the Austen, but the sympathy seems right: find authors whose style you admire, be they scientific or secular, and get forging. I don't advise channelling Hunter S Thompson into your next HEP paper, but if I find it I will certainly read it with more enthusiasm than the many stiff, colourless, unexceptional manuscripts that our collaboration issues. Such is the bleaching effect of large-scale text review by hundreds of borderline autistics desperate to pick a hole, any hole, in your document.

My own view of appropriate balance is that good scientific writing should be clear and unambiguous to any reasonable reader, but you shouldn't notice the writing at all unless a) it's unusually elegant and refreshing, or b) you snigger at the occasional little joke or flourish. In the end, your work does need to be read by humans, not robots. I'm rather pleased with the UE paper, but not as pleased as I was 18 months ago when it still had a bit of individual character. Such is life in a collaboration.

Ok, back to the "always write in present tense" thing -- in principle at least, it is the subject of this blog. This "rule" just leads to bizarreness; you may have got used to reading science papers that say "blah blah are measured" in the abstract, but it's not really a natural language usage. Your English teacher would tell you off, saying "no, they have been measured". Obviously the measurement is not taking place as the paper is being read. You might get away with "are presented", "are compared", "are used", etc. but the actual measurement, simulation, collection etc. is definitively in the past. One of the reasons for the verbal contortions is the strict application of "passive voice" and not being allowed (at least in ATLAS collaboration papers) to say "we", as in "we present measurement of...". This would be far more natural, and the passive form strikes me as rather spooky, as if the work of ATLAS is being reported by a spectre who only observed but did not participate. Which, as PhD students will tell you when appropriately bribed, is an apt description of their professor's involvement...

This dance of past and present tense isn't as simple as choosing one or the other, as the above example demonstrates. Certain operations that actually involved running the beam, writing events to disk, performing numerical analyses on them, etc. are actions located in the past. But the presentation in the form of the words and pictures of the paper, is current. The comparison of one plot to another is being made by the reader in the present. And if we did our job well, the scientific conclusions drawn from the made-in-the-past data plots are themselves timeless, everlasting, and deserving of the present tense. So the tense in a paper will, if appropriately applied, bounce back and forward a bit within the various sections depending on what you are describing. Which is hardly surprising; we do it all the time in natural language, but as I mentioned before, we have often internalized some odd rules which we would only ever consider using in scientific writing, and we would typically do well to unlearn them.

In our paper we used past tense for aspects of our analysis (or of previous work) which were done in the past relative to the reading of the document, and the present for the timeless things or those which refer to the way in which the reader is studying the document. But the summary section then requires a further nuance; until this I only used the past tense for things which happened weeks or months ago as part of the analysis, but by the time the reader is at the conclusions, their personal studying of the presented plots (which were present tense as they were first seen and described) is in the past. The recent past, but the past nevertheless. And I think that small-scale reference time is the one that feels most natural in concluding remarks; as they read the summary they are not currently looking at the plots etc. but rather remembering them from the previous pages. I find it extremely perverse to read a conclusion section which starts with "An analysis is presented..." -- what, another one?! At that point you are clearly reviewing -- the analysis has been presented. For what it's worth, "has been" feels so much more right than "was", but I'm damned if I can tell you why.

Have I convinced you? Probably not -- that's the thing about good (bikeshedding)[http://en.wikipedia.org/wiki/Parkinson's_law_of_triviality] opportunities, they are a gift that just keeps giving because everyone has something to offer. But my version is right, so there :-P