Sphinx

I am not, I would like to think, an unreasonable Luddite. I suppose it could be said that what I am is at best inconsistent; sometimes not a Luddite at all, indeed sometimes the sort of middle-aged man who desperately strives to keep up with odd bits of the technological Zeitgeist, enjoying catch-up TV while wondering what happened to car CD players, and sometimes an entirely reasonable Luddite. I can see, for example, why my favoured approach to constructing an index – creating a simple Word table by working methodically through page proofs, and then doing an A-Z sort – is clearly unsuited to the world of eBooks and online publication. And so I didn’t whine too much when asked to provide a list of index terms when submitting the manuscript of a forthcoming edited volume.

I’m not sure that I really thought about how this list might be used; I simply thought of the terms I would expect to see in the final index. Having now received the proof copy, I have no real idea how the list has been used. In some cases, clearly, they’ve simply done a word search, so we end up with some key terms (‘Thomas Piketty’ is the one that really leaps out at me) being indexed multiple times in a single paragraph. Sometimes, this doesn’t seem to have happened; ‘Egypt’ is in the index list, so why hasn’t it been consistently indexed? ‘Marx, Karl’ is in the list, and whatever mechanism is involved has successfully grasped that ‘Karl Marx’ should be linked to this – but not ‘Marx’ alone, although this problem doesn’t occur with Piketty.

Sometimes there is clearly sone human judgement involved, or a higher level of AI, as it knows to index ‘returns on capital’ and ‘capital returns’ as the same thing; mostly, however, it’s utterly hopeless at synonyms or variations or near equivalents. If the list says ‘population, see demography’, why does it then index neither population nor demography? It is symptomatic that the Author Query claims that the index terms ‘data, reliability of’ and ‘property, rural’ are not to be found anywhere in the text, when the latter topic features in at least a third of the chapters and the former in virtually all of them…

And I have FIVE DAYS to turn the whole thing around? Actually they told me five days and then sent the link to the files on Saturday, with the deadline then set as Thursday… Not going to happen. This gives me just enough time – since, thankfully, all the contributors seem to have done a great job in getting their submissions ready, and my editing doesn’t seem to have missed too much – to insert additional index entries where clearly necessary, but certainly not enough time to remove all the redundant ones, even if I could work out how to do that – even if the system allows it…

I do now wonder what I could have done differently, to learn lessons for next time – and to be honest I’m not sure. Rigidly enforce a limited vocabulary on everyone, to do away with synonyms and variations? ‘Always refer to “rural property” not to “country estate” or “agricultural investment”; never say “unfree labour” when you mean “enslaved people”‘ and so forth… I guess the crucial point is to recognise that where once the art of indexing was about trying to anticipate what future readers might want to look up, it’s now about trying to anticipate what an indexing algorithm that I don’t understand will make of the text…

Update: having now finished wading through the proofs, I can add that a particular highlight turned out to be the key term ‘war’, which we’d selected to cover the various discussions of the financing of war, the role of war in the destruction of accumulated capital and so forth, and the indexing algorithm decided to apply to every single mention of any sort of war, especially those used solely for dating purposes (“before the Peloponnesian War” and the like). I have taken most of these out…