Abstract
This article describes the application of modern algorithms to crack the official encryption method of the Spanish Civil War: the Strip Cipher. It shows the differences in efficiency and effectiveness between a genetic algorithm and mathematical programming, the optimisation methods known collectively as mathematical optimisation. Unlike the genetic algorithm, the programming approach has been seen to lead to high computational costs or to non-legible plain texts, which make it impractical. To improve the search for the genetic operators used, a dictionary is applied to identify possible words in each partially decrypted text and, thus, unblock the process. Results and conclusions have been obtained by analysing the outcome of the algorithms when attacking real ciphertexts found in the General Archive of the Spanish Civil War in Spain. Both the mathematical programming and the genetic algorithm approaches have merit, but the latter has considerable practical advantages.
Acknowledgements
The authors would like to thank the staff of the General Archive of the Spanish Civil War (Salamanca) and of the Military General Archive (Madrid) for the assistance provided in the location, consultation, and reproduction of the document collections used in this work.
Notes
SA = simulated annealing; TS = tabu search; AC = ant colony; HC = hill climbing.
I = number of individuals, G = number of generations, CP = crossover probability, MP = mutation probability, and ET = execution time in minutes (runtime on a laptop HP Compaq 6730 s with Intel Core 2 Duo 1.80 GHz Windows Vista Home Premium). Column L measures the results' legibility, where N = plaintext obtained is absolutely incomprehensible; I = some vowels and consonants are found correctly, but not all the text is comprehensible. A = language is comprehensible in general although some fragments or words require context to be understood; and P = text is perfectly or almost perfectly decrypted.
L = legibility; ET = execution time.