A tiny act of self cyborg-ification

Like pretty much everyone in the developed world, I like to find things on the internet. The act of discovery is thrilling, and we like to think, informative (though one must wonder how to quantify the intellectual value of another cat video). For the artist, the web offers an inconceivably large corpus of inspiration, visual or otherwise.

When I find an image which appeals to me, I frequently copy it onto my local computer, building a scrapbook of inspiration (or provocation). Some people like to use services like Pinterest to collate these things, but I am old school, and prefer to keep my data on spinning hard-disk platters under my control. As a result, I have collected hundreds of images over the years, mainly sitting in a single unordered directory.

These images lie dormant, just 1s and 0s amongst millions of others. Every now and then I flick through them, looking for a spark. Often I can no longer remember when or why (of from where) I copied them.

I decided to present them back to myself, in an algorithmic manner, to see what they might inspire in a new context. Every day, at 8:31am, a picture is selected at random and posted onto my Twitter timeline. I have no interaction with this process, the images just appear, as if from me, at the same time each day.

A tiny act of self cyborg-ification.

At some point in the past, I selected these images. Each one represents an aesthetic choice made by a historical version of myself. And now they are being presented back to me, algorithmically, for fresh appraisal.

Kind of like a visual version of Oblique Strategies.


Erasing History

As I upgrade my hardware, I dutifully copy these images from one place to another. During this transfer, the files themselves are re-created on a new disk, entirely new digital records which perfectly replicate their parent. The idea of the ‘original’ image ceases to make sense – you can’t point to a specific copy of the data and claim it is any more authentic than any other.

However, in theory, each image has the potential to retain it’s provenance and history. Many cameras record EXIF data, which can be used to store technical details (the camera model and settings), information about the creator (artist name, copyright) and the circumstances of creation (time taken, GPS coordinates). These data are ideally created when the image is taken, and then augmented as the image makes it’s way out into the world.

I wondered what history had been recorded in my directory of collected images and whether this could be brought to the surface. Generally this data refers to the act of creation, but in theory, the act of sharing the photo could also be recorded inside the image. I could leave my mark, and perhaps watch it move around the network, like some sort of ‘message in a bottle’.

This was an exciting proposition, until I actually ran few tests. It transpires that EXIF data can be stripped, and Twitter is one of the worst offenders.

Consider these two seemingly identical images of Alan Moore:

First, the ‘original’ which I copied from the web.


It contains all sorts of information:

ExifTool Version Number         : 9.90
File Name                       : alan-viking.jpg
MIME Type                       : image/jpeg
JFIF Version                    : 1.01
Exif Byte Order                 : Little-endian (Intel, II)
Image Description               : SAMSUNG
Make                            : SAMSUNG
Camera Model Name               : GT-I9000
Orientation                     : Horizontal (normal)
X Resolution                    : 72
Y Resolution                    : 72
Resolution Unit                 : inches
Software                        : fw 49.01 prm 49.103
Modify Date                     : 2015:01:21 15:50:35
Y Cb Cr Positioning             : Centered
Exposure Time                   : 1/26
F Number                        : 2.6
Exposure Program                : Program AE
ISO                             : 100
Exif Version                    : 0220
Date/Time Original              : 2015:01:21 15:50:35
Create Date                     : 2015:01:21 15:50:35
Components Configuration        : Y, Cb, Cr, -
Shutter Speed Value             : 1/26
Aperture Value                  : 2.6
Brightness Value                : 1.54
Exposure Compensation           : 0
Max Aperture Value              : 2.6
Metering Mode                   : Center-weighted average
Light Source                    : Unknown
Flash                           : Off, Did not fire
Focal Length                    : 3.5 mm
Warning                         : [minor] Unrecognized MakerNotes
Flashpix Version                : 0100
Color Space                     : sRGB
Exif Image Width                : 640
Exif Image Height               : 480
Interoperability Index          : R98 - DCF basic file (sRGB)
Interoperability Version        : 0100
Sensing Method                  : One-chip color area
File Source                     : Digital Camera
Scene Type                      : Directly photographed
Custom Rendered                 : Normal
Exposure Mode                   : Auto
White Balance                   : Auto
Digital Zoom Ratio              : undef
Focal Length In 35mm Format     : 0 mm
Scene Capture Type              : Standard
Contrast                        : Normal
Saturation                      : Normal
Sharpness                       : Normal
GPS Version ID                  :
GPS Latitude Ref                : North
GPS Longitude Ref               : East
GPS Altitude Ref                : Above Sea Level
Compression                     : JPEG (old-style)
Thumbnail Offset                : 1316
Thumbnail Length                : 9350
Image Width                     : 480
Image Height                    : 640
Encoding Process                : Baseline DCT, Huffman coding
Bits Per Sample                 : 8
Color Components                : 3
Y Cb Cr Sub Sampling            : YCbCr4:4:4 (1 1)
Aperture                        : 2.6
GPS Altitude                    : 0 m Above Sea Level
GPS Latitude                    : 0 deg 0' 0.00" N
GPS Longitude                   : 0 deg 0' 0.00" E
GPS Position                    : 0 deg 0' 0.00" N, 0 deg 0' 0.00" E
Image Size                      : 480x640
Megapixels                      : 0.307
Shutter Speed                   : 1/26
Thumbnail Image                 : (Binary data 9350 bytes, use -b option to extract)
Focal Length                    : 3.5 mm
Light Value                     : 7.5

Now compare it to this photo, which has been passed through Twitter:


ExifTool Version Number         : 9.90
File Name                       : CDHLzOyW8AEGvpx.jpg
MIME Type                       : image/jpeg
JFIF Version                    : 1.01
Resolution Unit                 : None
X Resolution                    : 1
Y Resolution                    : 1
Image Width                     : 480
Image Height                    : 640
Encoding Process                : Baseline DCT, Huffman coding
Bits Per Sample                 : 8
Color Components                : 3
Y Cb Cr Sub Sampling            : YCbCr4:2:0 (2 2)
Image Size                      : 480x640
Megapixels                      : 0.307

The image you see on Twitter no longer contains a single trace of information related to it’s creation. The image has been reborn as an anonymous, amnesiac clone of the original.

The act of ‘sharing’ has stripped it of it’s identity.

Sure, there are services like TinEye which offer to find the history of online images. However, they are not perfect, particularly for images on low traffic sites.


Here Tineye has identified the first citation of this image as coming from a social media aggregation site. Whereas I actually lifted it from here.

In an attempt to thwart this algo-revisionism, I am publishing some of the EXIF data in the text of the tweet. There’s not room for much, but where possible I publish details of when it was created, and by whom along with a record of the software used to manipulate it.

Unfortunately, many of the images have already been through an anonymisation process before I came across them. There is no record of their origin, and their future is stored in proprietary systems, beyond scrutiny.

Whilst we worry about networked systems recording ever more data about us, perhaps we should also consider the data which is being selectively ignored, and why.

Drunk George redacted

Parody accounts are one of the prevailing cultural forms of Twitter. The combination of lazy anonymity and the 140-characters-of-wit format make it the perfect place to assume a character and play it out to a potential audience of millions. Back in the prehistoric days of Twitter, one of the first parody accounts to come to my attention (and hold it) was @osbornedrunk, wherein our erstwhile Chancellor, Gideon George Osborne, was beautifully portrayed as a bumbling idiot, bouyed by frequent hits of vodka jelly and magic mushrooms.

Drunk George ran his course and the author moved on to other things, briefly reappearing as @osbornedead around Halloween, but effectively the horrific bufoonery of the real-life Osborne outpaced the character and he quietly retired.

It turns out that the creator of @osbornedrunk is a friend of mine, a fact revealed accidentally, at the great crossing of the Brighton Ley lines (hail Eris). With the upcoming General Election, it seemed appropriate that George should come out of retirement and play out the last fevered month of the campaign. After all, parliament has now been dissolved, leaving plenty of time for vodka jelly.

One of the defining features of this government has been the revelation that we are all being surveilled, apparently for our own good. This project brings this to our attention by generating a redacted report about anyone who sends an @ message to @DrunkGeorgeOsb.
The reports are generated by scanning your last 100 tweets and finding out who you’ve been talking to and what they are talking about. When presented with a redacted report one cannot help but imagine what words might be behind those black oblongs, even more so when the report is about ourselves.
In a sense, drunk George has now become a cyborg, partly controlled by his author, and partially by algorithm.

The Mood of the Nation

As we approach the election, it becomes apparent that the narrative landscape of politics has changed. In the past, the discourse was led by inky newspapers and the satirical sniping of broadcaster-sanctioned comedians. The last election, in 2010, certainly had a social media component – manifested by the American-aping ‘televised debates’ and the associated back-channel chatter on Twitter. However, in the intervening five years, the size of this channel has increased enormously (approx 30m users in Q1 2010, to 288m users in Q4 2014)

It’s now possible to use this ‘big data’ to get a handle on the mood of the nation. Tweets can be read by algorithm and classified as positive or negative. Sentiment analysis is big business, harvesting the millions of opinions expressed online, and turning them into numerical values.

Brandwatch recently released sentiment data for David Cameron and Ed Miliband during the BBC Paxman interviews. In the video below I have used these data to plot the mood towards the two leaders during the broadcast – represented by red and blue backed emoticons. If the mood towards to politician is positive, the emoticon smiles, if the mood is negative, the emoticon cries.

Thousands of tweets, reduced first to numbers, and then to emoticons. Watching the result I’m struck by how the mood seems mainly negative towards both men. Is this a reflection of a national disenfranchisement with politics, or is it simply a reflection of social media itself – a place we go to complain, rather than praise?


North Korea recently published a list of 300 patriotic slogans, which read remarkably like Markov generated sentences. I’ve never been a huge fan of Markov chains for creating text, since the results can be somewhat haphazard and ungrammatical. However, when the source text is itself haphazard and ungrammatical, it seems fitting.

@nk_markov takes these 300 aphroisms as source for a Markov system, and generates new pithy slogans and marries them to images of the glorious leader himself.




The emoticon has become ubiquitous in informal text-based communications. The necessity to modify the meaning of potentially ambiguous sentences with the use of pictographic shorthand has a long history, but really came into it’s own with the democratisation of publishing which came with the advent of the internet. They are a visual shorthand which saves time in our busy lives. More recently we’ve seen the rise of the Emoji, which offers a more varied palette of face based symbols, conveying a range of emotions far greater than the manipulation of the ascii character can offer.

Emoji Portraits by Jung Jake

These symbols almost universally take the form of a circular yellow face. This is a gender and race neutral form which finds it’s roots in the counter-culture of the 60s, and more recently in the Acid House scene of the late 80s. It’s interesting to see the symbol returning in the realm of fine art, notably in the works of Ryca and Jimmy Cauty.

Alan Moore with a Jimmy Cauty riot shield

The human visual system has a lot of circuitry dedicated to the recognition of faces. Recognising the emotional states of our fellow primates clearly provides a strong evolutionary advantage. Indeed, it’s argued that the white sclera of the human eye developed to accentuate the expressiveness of our faces and make it easier for others to track our gaze. It is no surprise then that the face is used as a shorthand for emotion in written communication, it’s where we wear our emotions, and we’re built to quickly recognise them, even in the form of a yellow circle.


Because faces are so important to us, we have trained computer algorithms to detect them automatically. I’ve used this technique in a number of recent works, notably Who Watches the Watchers? and @farageblocked. These algorithms are derived statistically from millions of images of faces – completely unlike our own system which seems at least in part innate, and certainly forms part of a more general visual system which works in an entirely different manner. This is most obvious when we find images mis-identified as faces, we’re suddenly drawn to see if we can see what the machine has seen, as in Henry Cooke’s Faces in the Cloud project.


I recently modified @farageblocked to replace his face with miserable emoji and was immediately struck by how pleasing it was. I wondered whether the same technique could be applied to video, potentially providing a visual shorthand for remapping emotional states. There is a delightful tension between the obfuscation of identity and the immediate recognition of the emoticon.

2010 leader’s debate

As our social interactions move away from the realm of reality and into the online world, we are seeing an increase in social anxiety – we seem more comfortable communicating our emotions via the similey face than dealing with the reality of our fellow humans. Apps like pplkpr promise to simplify our emotional landscape by monitoring our stress levels and advising us who to avoid.

Waiting for Godot (excerpt)

Could we imagine a future where this anxiety is eased by a form of ’emoticon subtitling’? Where our already mediated interactions are automatically overlaid with a simplified palette?

Shake it off – Taylor Swift

The technique lends itself to directly inverting the intended meaning of the source, for example:

Happy (sad) – Pharrell Williams

Stagger Lee (happy) – Nick Cave and the Bad Seeds