1 Introduction
Feeling a movement of pity,discovered the induction coil,cette irraisonnee induction,and entered the opening in the wall.Only by some recherche movement,apres coup et sous forme d’introduction,opening his seized manuscript,the enemy made within the enclosure of the vineyard.Which he had thrown off at the beginning of his labor,in opening so exactly at the,than the thirst of my paternity.We can then start at once,and whose informing voice had consigned me to the hangman,as any person at all conversant with authorship may satisfy himself at.
This thesis describes Algorithmic Meta-Creativity (AMC). In other words it is about using creative computing to achieve computer creativity.
The project is transdisciplinary; it is heavily inspired by the absurd french pseudo-philosophy pataphysics and draws from a wide range of subject areas such as computer science, psychology, linguistics, literature, art and poetry, languages and mathematics.
The research included exploring what it means to be creative as a human, how this translates to machines, how pataphysics relates to creativity and how creativity should be evaluated in machines.
Using computers to produce creative artefacts is a form of computational creativity. Using creative techniques computationally is creative computing. AMC spans the two—whether this is to achieve a creative or non-creative output. It is the use of digital tools (which may not be creative themselves) and the way they are used forms the creative process or product.
Creativity in humans needs to be interpreted differently to machines. Humans and machines differ in many ways, we have different ‘brains/memory’, ‘thinking processes/software’ and ‘bodies/hardware’. Too often creative output by machines is judged as we would a humans.
Computers which are truly artificially intelligent might be capable of true artificial creativity. Until then they are (philosophical) zombie robots: machines that behave like humans but aren’t conscious. The only alternative is to see any computer creativity as a direct or indirect expression of human creativity using digital means and evaluate it as such. AMC is neither machine creativity nor human creativity—it is both. By acknowledging the undeniable link between computer creativity and its human influence (the machine is just a tool for the human) we enter a new realm of thought. How is AMC defined and evaluated? This thesis address this issue.
a practical demonstration of AMC
a theoretical framework to help interpret and evaluate products of AMC
The outcome of step (1) is presented as a website—pata.physics.wtf—written in different programming languages1, making calls to external web services2, in a total of over lines of code3 spread over files.
The main purpose of the system above is to demonstrate the three creative patalgorithms in the context of exploratory Information Retrieval (IR). A browsing rather than a search engine, it presents results in various formats such as sonnets and golden spirals. The system partially automates the creative process, generating results on demand, which allows users to focus on their own personal artistic evaluation rather than production.
Immediate inspirations come from fictional character Doctor Faustroll created by french absurdist and ‘father’ of pataphysics Alfred Jarry (1996), the fantastic taxonomy of the Celestial Emporium of Benevolent Knowledge by magical realist Jorge Luis Borges (2000) and A Hundred Thousand Billion Poems by pataphysician and Oulipo co-founder Raymond Queneau (1961), amongst others.
To address step (2) above, I explored the problem of objective evaluation and interpretation of subjective creativity specifically in regards to AMC. I have argued that the most appropriate way to approach this is by looking at five objective constraints (person, process, product, place, purpose) and seven subjective criteria (novelty, value, quality, purpose, spatial, temporal, ephemeral) holistically and by understanding that humour and art ‘lie in the ear and eye of the beholder’.
This resulted in an interpretation framework visualised as an evaluation matrix ( constraints x criteria) which can be used to qualitatively and/or quantitatively measure the creativity of a given AMC artefact:
a set of scales that can be used to approximate a ‘rating’ for the creative value of an artefact,
a set of criteria to be considered using the scales above,
a combined framework for evaluation.
1.1 Motivation
Computers are binary machines; the world is black and white to them (0 and 1, on and off). Programmers can run abstract high-level commands which are executed in sequence (with fast speeds giving the illusion of multitasking). They are precise, structured, logical, and generally abide by strict standards. Computers can only be creative if they are given clear instructions as to how. is generally focused on relevance of results in regards to the query.
The Analytical Engine has no pretensions whatever to originate anything. It can do whatever we know how to order it to perform.
(Ada Lovelace, in Menabrea and Lovelace 1842, her emphasis)
Pataphysics emerged during the Belle Époque4 in France and has either directly or indirectly influenced various artistic movements such as Dada, Symbolism, Surrealism, Oulipo and Absurdist Theatre. Pataphysics is highly subjective and particular, values exceptions, the imaginary and the mutually incompatible.
Creativity is often studied at various levels (neurological, cognitive, and holistic/systemic), from different perspectives (subjective and objective) and characteristics (combinational, exploratory and transformative). It is usually defined in terms of value, originality and skill.
Combining computing with pataphysics seems impossible—although the antinomies below (juxtaposing principles in computing on the left with ideas from pataphysics on the right) highlight just how intriguing a possible combination of the two would be.
Polymorphism (generalisation) opposes particularity.
Precision opposes exceptions and contradictions.
Logic and structure oppose the imaginary and paradox.
Cross-compatibility opposes the mutually exclusive.
Responsiveness opposes the specific.
Relevance opposes the creative.
This apparent dichotomy of computing and pataphysics is alluring. Christian Bök argued that pataphysics “sets the parameters for the contemporary relationship between science and poetry” (2002). Pataphysics suddenly seems like the perfect choice infusing computers (science) with creativity (poetry).
Combining pataphysics with creativity is easier. The ideas of combinatorial, exploratory and transformative creativity map quite nicely onto some pataphysical concepts such as clinamen, syzygy, antinomy and anomaly.
Another motivating factor for this project was the lack of research in the particular area of creative computing in general. The discipline of computational creativity has emerged fairly recently5 from a background in Artificial Intelligence (AI). It appears to focus a lot more on the outcome of a product that would be judged creative rather than the actual process. Creative computing focuses on producing creative algorithms which may or may not have creative outputs. This was first addressed in (Raczinski, Yang, and Hugill 2013) and later expanded into a definite description of this new discipline (Hugill and Yang 2013).
My personal interest in this project comes from a background in computer science and a longstanding interest in art. Most recently I managed to successfully combine my technical skills with my creative side for a Master of Science degree in Creative Technologies at De Montfort University (DMU).6
1.2 Questions
Research dealing with subjective ideas and concepts like creativity throws up a lot of questions. My intention is to address them all throughout this thesis, although some of them will not have definite binary answers. An attempt to answer them can be found in the conclusion chapter 14.2.
What is the relationship between pataphysics and creativity?
How is computer creativity related to AI?
Should we distinguish between computationally automated or emulated creative processes and the programmer’s input?
How can a machine’s creative output be evaluated?
How can IR be infused with creativity?
1.3 Methodology
This project combines research in science and art making it transdisciplinary.
- Pataphysics
Literature, Philosophy, Art, Poetry
- Creativity
Cognitive Science, AI, Digital Humanities (DH)
- Technology
IR, Natural Language Processing (NLP), Web Development
- Epistemology
Transdisciplinary, subjective
- Methodology
Creative computing, exploratory, experimental
- Methods
Artefact, literature synthesis, algorithm design, theoretical framework, critical reflection and analysis, rapid incremental prototyping
The general process of my project was as follows.
Critically analyse and synthesise existing literature,
develop pataphysical algorithms,
design a system to demonstrate algorithms,
develop a website as an artefact,
define an evaluation and interpretation framework,
analyse results.
1.4 Contributions
The key contributions to knowledge described in this thesis are:
Three pataphysical search algorithms (clinamen, syzygy and antinomy).
A creative exploratory search tool demonstrating the algorithms pata.physics.wtf.
A set of 7 subjective criteria and 5 objective constraints for defining creativity.
A combined framework for evaluating and interpreting creativity.
1.5 Publications
Some chapters (especially Foundations and Interpretation) in this thesis are based partially on articles published during this project. I have used fragments from those papers freely without specific citations unless clearly indicated. I had several co-authors (Hongji Yang, Andrew Hugill, James Sawle and Dave Everitt) for these pieces and I hereby acknowledge their contributions.
A list of publications can be found in the preface on page . Details of talks and exhibitions and copies of the publications can be found in appendix E.
1.6 The Hitchhiker’s Guide to this Thesis
This document is organised into parts which form the main logical structure of the thesis and each part contains several chapters. There are margin notes pointing to relevant chapters, sections, tables, figures or images throughout.
1.6.2 Thesis Language
This thesis was written in LaTeX. It was first drafted in March 2015 and completed in December 2016. I created my own ‘style’ based on only a few restrictions imposed by DMU regulations (such as font size and page margins).
1.6.3 Chapter Overview
The preface contains the abstract, acknowledgments, and various tables of contents.
- Introduction
Gives a general top-level overview of the research presented in this thesis.
- Inspirations
Lists the various immediate inspirations for the project.
- Methodology
Explains and justifies the approach taken for the research.
- Pataphysics
Describes the origins of pataphysics and related concepts.
- Creativity
Lists the theories of human and computer creativity.
- Technology
Provides the technical background of this research.
- Evaluation
Explains the models of evaluation for computer creativity.
- Foundations
Brings together the research on creativity and pataphysics.
- Interpretation
Critiques evaluation models and proposes a new approach.
- Implementation
Describes pata.physics.wtf from a technical standpoint.
- Applications
Showcases two use cases of this research.
- Patanalysis
Analyses the artefact and some of the theoretical aspects.
- Asprirations
Addesses future work and known issues.
- Outroduction
Draws together and summarises the contributions of the work presented in this thesis.
The appendix contains additional material that was not suitable for including in the main body of the text. It also contains the list of references.