Johannes Auer: search lutz!
Nearly 50 years ago a calculator generated a literary text for the first time ever. This happened in Stuttgart, my home town.
In 1959 Theo Lutz wrote a program for Zuse Z22 to create stochastic texts. Following Max Bense’s (a Stuttgardian philosopher) advice, he took sixteen nouns and adjectives out of Kafka’s "Schloss," which the calculator then formed into sentences, following certain patterns. Thus, every sentence began with either "ein" or "jeder" ("one" or "each") or the corresponding negative form "kein" or "nicht jeder" ("no" or "not everybody"). Then the noun, selected arbitrarily from the pool of sixteen given nouns, was linked through the verb "ist" ("is") with the likewise arbitrarily chosen adjective. Last, the whole construction was linked up through "und," "oder," "so gilt" ("and," "either," "thus") or given a full stop. Following these calculation instructions and by means of this algorithm, the machine was able to construct sentences like:
EIN TAG IST TIEF UND JEDES HAUS IST FERN
(A day is deep and every house is distant)
JEDES DORF IST DUNKEL, SO GILT KEIN GAST IST GROSS
(Every village is dark, thus no guest is large)
For the performance of "searchLutz!" I use a web conversion of Theo Lutz’s program which I wrote in PHP. The web interface generates stochastic texts on the basis of Lutz’s algorithm but permits additional word input. The nouns and adjectives of the original vocabulary can be replaced by the audience during the performance by using a terminal. Furthermore words from the live search of the search engine Fireball could infiltrate the text generation process.
In 1959, computer texts were connotated as literary texts in two ways: Firstly through the "Kafka" vocabulary used, and secondly through corrections carried out by Theo Lutz. In a printed copy of a selection of stochastic texts he had edited, Theo Lutz corrected little grammar mistakes and missing punctuation marks by hand, and thus, contrary to programming acted as a "traditional" author. During the performance we refer to these literary features (or one could almost say there is no escaping from one’s humanity) of the first computer-generated texts in two ways: First we do so through the co-authorship of the audience, secondly we have a professional speaker who is reading the so produced computer texts directly off the screen and is thus performing them as they were generated.