In the last 50 years, many different ways of implementing software systems have evolved. These include well over 150 programming languages as well as numerous paradigms such as structured decomposition, OO and component based systems, all the way through to today's mash-up, where you metaphorically throw software at a wall to see what sticks, a bit like spaghetti.
It is therefore a matter of considerable surprise that, independently of all these paradigms and languages, software has a characteristic signature in common with many natural systems; that of power-law distributions. This is highly analogous to the behaviour of a gas which contains many, many molecules in random motion but which together obey the general gas equation PV = RT, independently of the gas.
In this talk, I will argue that this behaviour is inevitable and will prove a new general theorem that it is intimately related to the conservation of choice in Shannon's Information Theory. This leads naturally to power-law behaviour in component size independently of the technology used and in turn leads to the well-observed phenomenon of software defect clustering. This will be supported by lots of empirical data.
Les Hatton is managing director of Oakwood Computing Associates Ltd. and is Professor of Forensic Software Engineering at Kingston University, London. He was awarded the Conrad Schlumberger prize for geophysics in 1987 before becoming interested in software reliability and switching careers in the 1990s. Although he has spent most of his working life in industry, he has held a number of academic appointments in the UK and the Netherlands.
He has published four books and many technical papers and his 1995 book 'Safer C' helped promote the use of safer language subsets in embedded control systems and paved the way for the automotive industry's widely-used MISRA C standard. After a few false starts, he has designed, implemented and/or managed the production of successful commercial IT systems, from 50,000 source lines up to the world's first portable seismic data processing package, SKS, eventually comprising some 2,000,000 source lines.
His primary interests in computing science are forensic engineering, information security, legal liability and the theory of large systems evolution. In mathematics, he is active in signal processing, medical image processing, sports biomechanics and modelling the effects of high frequency sound on marine mammals. He is on the editorial boards of IEEE Software and the Journal of Open Research Computation. In his spare time, he is the guitarist and harmonica player with the Juniper Hill Blues Band.
Film finns tillgänglig.
Alla är välkomna till UppLYSning! Du behöver inte vara medlem i Lysator eller student/anställd på universitetet. Vanligen hålls föredragen på tisdagar klockan 18:15 - 20:00 i Visionen, B-huset.
Aktuellt program finns på http://www.lysator.liu.se/upplysning/ där du också kan anmäla dig för att få utskick om UppLYSning via e-post.
UppLYSning arrangeras av Lysator i samarbete med Institutionen för datavetenskap, Institutionen för systemteknik och UNIT.