cent mille milliards de... pourquoi encore?

In 1961, Gallimard published Raymond Queneau's 'Cent Mille Milliards de Poémes; a feat of unequalled combinatorics in literary production. Queneau presented these interchangeable sonnets as constituting an inexhaustible source of reading matter for the reader.

Each of the ten sonnets is comprised of 14 lines with each line being interchangeable with its correlate line in any of the other sonnets giving 1014 permutations. Queneau's calculations point out that it would take a single person (reading 1 sonnet per minute (the time which a close textual reading demands ), 8 hours per day, 200 days a year) over a billion years to complete the reading.

Some fifty years later, the question remains - How could an attempt to read all of the poems, if one were to consider it, take place and how would one proceed, systematically, in the attempt?

Print versions of the poems used horizontal divisions of the lines on the pages allowing the reader to cut the poems into strips thus providing access to different permutations but the problematics introduced by paper made this unfeasible. Sometime later Jean-Michel Bragard + Robert Kayser produced a mechanical device using 14 rotating decagonal cylinders.

With the advent of hypertext software in the late 1980s, versions of the poemés were assembled so the reader could view different combinations more easily than in the printed version and more compactly than Bragard+Kayser's machine. With the rise of the internet, many html and, more recently, javascript and other implementations of the poemés have appeared. These range from the simple random generation of each line from an array when the page is loaded to the slightly more advanced graphic selection of lines using a 10x14 square grid.

As is the case with many hypertext implementations of printed literature, whilst these versions provide for quick access to permutations, they do nothing more than the printed or machine versions and so neglect to utilise the potential of the software; namely attempting to provide a solution to the impossible task of a systematic and complete reading of the poems.

The ongoing development of more economic algorithms in the field of combinatorics, especially those concerned with permutation, accelerates the relatively simple task of exhausting the potential of Queneau's poemés; the creation of a piece of software that runs through each permutation and displays it on screen is achievable by the novice programmer / hypertext author with a basic understanding of recursive techniques.

However, the issue still remains that no single person could read all of the permutations, even with the assistance of computers. Queneau's calculations were based on a defintion of 'a' reader. It is ironic that the OuLiPo, created in 1960 by Queneau and François Le Lionnais, potentialised collective endeavour but never revised these calculations based on the model of a distributed reading.

However, this concept does not appear to have been explicitly excluded from the programme of the OuLiPo; George Perec was not solely responsible for the production of the lipogrammatic 'La Disparition': Whilst the novel is solely attributed to Perec, everyone he came into contact with was pressured into submitting words, phrases or sentences that fitted within the constraint. Perec thus provides an OuLiPian model for the realisation of all of the poems; instead of a multiple anonymous authorship under the umbrella of Perec, we have a multiple anonymous readership under another.

The utilisation of distributed programming software such as that implemented to perform tasks too lengthy for super computers, e.g. THINK , can offer a solution to the problem of a systematic and complete reading of the poems. By distributing blocks of permutations to participating readers, every permutation will have been read by some reader after some period of time, the duration being dependent on parameters outlined below.

Some Statistics:

Using Queneau's calculations, a solitary reader could get through 96,000 per annum and therefore would take 1,041,666.67 millennia to complete the reading.

Reading non-stop would reduce this to 191,304.13 millennia.

Print Run: Using an estimate by Tom Soder (http://www.recycledproducts.org/treecalculation.html), it would take 24,000000 trees [1 million tons of paper] to print each sonnet on a single side of A4. On a 20ppm printer, this amounts to 63,593.81 millennia in print time.

The world's population [currently around 6,240.576,000] would have to read 16,025.64 poems each and reading 60 per day, the project would be complete in 267.09 days. NB: this does not include the time spent in translation, banishing illiteracy, print time and delivery. Nor does it include the exponential growth of the world's population.

Moving to more reasonable calculations, the number of english speakers with internet access is currently around 172 million (http://www.worldlingo.com/resources/language_statistics.html)and therefore each would have to read 581,395.35 poems. At 60 per day [1 hour] a completed reading would take 26.62 years.

The visitor statistics for Grammatron (one of the more popular hypertext projects) are estimated at 1 million. Laughably, this would give each reader 100,000,000.00 each to read, thus taking, at 60 per day, 32.05 millennia to complete. 


dir_builder.exe - this application, coded by tara mclean creates a directory structure capable of housing 100 billion individual files. download [40k] - the last estimate for completion of the program was approx. 49 years [using win2k pro/athlon 950mhz].

download executable (win32 only)

cent mille... other versions