5 Computational methods and tools

Abstract

This chapter may well be the hardest in the book for those that are not all that computationally, mathematically, or especially graph-theoretically inclined. Textual scholars often take to text almost naturally but have a harder time grasping, let alone liking, mathematics. A scholar of history or texts may well go through decades of a career without encountering any maths beyond the basic schooling in arithmetic, algebra, and probability calculation that comes with general education. But, as digital techniques and computational methods progressed and developed, it transpired that this field of maths and digital computation had some bearing on textual scholarship too. Armin Hoenen, in section 5.1, introduces us to the early history of computational stemmatology, depicting its early beginnings in the 1950s and pointing out some even earlier roots. The strong influence of phylogenetics and bioinformatics in the 1990s is recounted, and their most important concepts are introduced. At the same time, Hoenen warns us of the potential misunderstandings that may arise from the influx of these new methods into stemmatology. The historical overview ends with current and new developments, among them the creation of artificial traditions for validation purposes, which is actually a venture with surprisingly old roots. Hoenen’s history shows how a branch of computational stemmatics was added to the field of textual scholarship. Basically, both textual and phylogenetic theory showed that computation could be applied to the problems of genealogy of both textual traditions and biological evolution. The calculations involved, however, were tedious, error-prone, hard, and cumbersome. Thus, computational stemmatics would have remained a valid but irksome way of dealing with textual traditions if computers had not been invented. Computers solve the often millions of calculations needed to compute a hypothesis for a stemma without complaint. They do so with ferocious speed and daunting precision. But it remains useful to appreciate that this is indeed all they do: calculate. The computer – or algorithm – does not have any grasp of the concepts or problems that it is working on. Nowhere in the process leading from variant data to a stemmatic hypothesis does any software or hardware realise that it is working on a textual tradition or genetic material. It has no feelings about that work and – more saliently – is indifferent to the quality, correctness, or meaning of the result it calculates. It is especially for this last reason that textual scholars should take note of the methods and techniques involved in calculating stemmata, even if the maths may not always be palatable work. Computer code and chips process data and yield some result or other. None of the nouns in the previous sentence somehow becomes inherently neutral, objective, and correct by virtue of being digital or mathematical in nature. If an algorithm contains a calculation error, the computer will repeat that error faithfully a billion times at lightning speed. Thus, it follows that we can only trust digital tools and computational methods if we can trust their theoretical and mathematical underpinnings, if