TAXONOMY OF GLITCHES

Outline of talk given at READ_ME Software Art Festival 2003, Helsinki.

Thanks to Olga and Alexei.

Slide 01
What I do at BEFLIX:
Finding glitch-like phenomena.
Minimal processing of images.
Aim to produce aesthetically pleasing images.

Slide 02
Quote from Kim Cascone:

 Today's digital technology
 enables artists to explore
 new territories for content
 by capturing and examining
 the area beyond the boundary
 of "normal" functions and uses of software.

Slide 03
This talk is more about art than software.

Slide 04
The tool isn't the message.
The result might be the message.
The genre is partly the message.
Why does there have to be a message?

Quote from schoolteacher:
 The picture came from you,
 and you are not wrong,
 so the picture cannot be wrong.

A contrasting non-quote:
 The picture came from a computer,
 and computers always crash,
 so the picture is wrong.

Slide 05
Old-style definitions of glitch mention:

Sudden discontinuity.
Slip, slide.
Electrical spike.
Background noise.
Outcome is unknown.

[supporting pictures follow]

Slide 06
Move emphasis...

Away from appearance.
Away from electricity.
Away from things localized in time or space.

Towards processes.
Towards computers.
Towards textures and fields of influence.

Slide 07
A glitch is not random.

[supporting image]

Slide 08
Glitches are not like fractals.

Not trying to mimick nature.
Not trying to explain anything.

[supporting image]

Slide 09
Highly structured - Some structure - No structure
    Fractal       -     Glitch     -  Randomness

[supporting images]

Slide 10
New proposed definition of glitch.

Glitch = 
 Any outcome,
 Generated by complicated process,
 Whose output is deterministic.

Slides 11-15
Image: PC RAM
Image: Unusual system initialization
Image: Executing garbage code
Image: Cookie index file
Image: Hardware crashes

Slide 16
Allow random events
to trigger
deterministic processes.

Slide 17
Random events include:

Cosmic particles
Code rot
Consequences of Heisenberg Uncertainty Principle

Slide 18
Glitches occur after an initial error-event
during a system's operating session.

Slide 19
Random inputs to deterministic machines:

Image: Calculator errors
Image: IC manufacturing errors

Slide 20
Deterministic inputs to deterministic machines:
Image and live example: Arcade emulator executing itself
Image and live example: Arcade emulator with wrong bios

Slide 21
Glitches aren't always "outputs" of something.

Slide 22
Data Visualization can be Intrisic or Non-Intrinsic.

Slide 23
Intrinsic Data Visualization.

Type I:
Glitches occur in a hardware device with built-in display capabilities.

Type II:
Glitches occur in auxilliary data which is fed into software.
External display device required to see the glitches.

Slide 24
Non-Intrinsic Data Visualization.

Data can be segmented differently to give different images
via colour lookup-tables.

Slide 25
One-to-one colour mappings give different aesthetic effects
although the data remains unchanged.

Slide 26
Code Bending is Intrisic Type II Data Visulaization, 
but we can classify further.

Slide 27
"Low-risk" code-bending attacks only the graphics engine.

"High-risk" code-bending attacks all the code including system routines.

[supporting animation of high-risk code-bent game]

Slide 28
Emotional appeal of glitches.

Slides 29-33
Image: Relatively new imagery for consumption.
Image: Images open to personal interpretation.
Image: High-density material.
Image: Retro overtones.
Image: Some people like straight lines.