Piet Zwart Institute: Media Design Research Fellowships
“NetMonster”: The Networked Image, October 2005

You can hang a picture on a wall – but you cannot hang an image on the wall, the image is the space between eye, experience and picture.

The twentieth centuries gift to us was montage which valued fragmentation, conflict, the staccato rhythm of the machine and opposed continuity and organic unities. In the work of Hertfield this was used to take various pictures Ape / Hitler combine them to make new commentary on the rise of Nazism. Other commentators on montage suggest that ripping images from various contexts and sticking them back together reveals new images, meanings and poetics from the interplay of the pictures.

This relies on a basic set of working assumptions that pictures are structured significances and when viewed. The corresponding images that manifest, have a symbolic ordering to them. Which we use to enable the navigation of complex thought and emotion.

The networked image raises many questions:

Does the networked image in it's proliferation and multiple contexts undermined this set of assumptions or reinforce them?

Is the networked image itself always in state of perpetual collage?

Has collage reached the end of cultural usefulness.

The question may not be what does the network do to pictures but what do the pictures want from the network?

The way young people use pictures on their mobile phones is a primary example of the image as notation (momentary structured significance without symbolic ordering) - they are skimmed over, turned into marginalia or instantly deleted.

If the networked image is always in state of collage then revealing it in other ways will rely on a re-ordering of it. An examination of it's constituent parts, its contexts and links.

Research Context

In the Autumn of 2004, a prototype software system called the “NetMonster” was designed to generate, edit and continuously update a composite image made up out of the results of internet searches guided by various key terms. Functionally, it could be thought of as an authoring system that combines a search engine and a content management system. At Brown University (USA) we performed various tests on the software with students of the Department of Modern Culture and Media. From this initial experiment several possible directions for further research have appeared.

Current content management and editing systems are based on the premise of efficiently delivering content within the structure of a web site, streamlining communication and optimising usability and ordering authorship into a production process. But our previous work with designing content management systems for non-expert users like “Nine(9)” (2003) has led us to realise that what is needed for cultural projects is software that can draw the user into a more ongoing dialogue with the wider network. Rather than just placing images onto a web page, we need to create portals in which content is harvested, reflected on and steered to allow people to explore the issues they find important. Our response is to develop the notion of ‘generative social software’ – software designed to react to users content by ‘generating’ (gathering, adding in and transforming) its own content from its searches, creating an iterative process of feedback and analysis.

The most commonly used aesthetic technique to combine several images into one is montage. This technique was originally developed when pictures could be cut and pasted by hand or film strips could be edited together to provoke new meanings or relationships. But it now seems unlikely that montage can survive intact when faced with the scale of imagery encountered on a network or the frenzy in which it is consumed. Despite their uniformity of setting, one image can appear in multiple quantities on a network which can differentiate its value – all images are equal but some are more equal than others. The development of an appropriate network aesthetic remains a central problem.

Networked images always appear in the company of text. The prototype “NetMonster” returned text and rendered it over the relevant image. Yet the different ways of perceiving image and text have yet to be addressed. Images in networks are normally treated like words – they appear as individual discrete units like buttons or thumbnails. Yet they can also retain the visual continuity of an image that cannot be reduced to a symbol (such as when they are shown ‘full size’). Networked media offers a space in which this ambiguity between text and image can be productively explored.

When images are encountered in public media and political argument they often perform an iconic role. Yet it is difficult to know how such images can truly enter a public debate when our encounters with them are so disconnected. The networked images that are returned by search engines are accessible yet not answerable nor accountable. We need to find ways to situate them, to create a context for them in which they can be collaboratively analysed, critiqued and built into a wider means of expression. That is the ultimate goal of this research.

Methodology

We intend to further our development of “NetMonster” to allow people to collaboratively build up a composite ‘networked image’ out of images, text and addresses returned from internet searches. The “NetMonster” will continuously rebuild itself based on users edits and changing search parameters, offering up new content and configurations. In this way the empty gesture of a political icon (e.g. Abu Gharib) can learn to detect the context of its own existence, automatically creating dialogue between itself and its users.

A “NetMonster” project starts with a set of key search terms. The “Monster” would then start to construct itself from the search results. This resultant "networked image" can then be navigated, edited and directed by its users using various browser based authoring features - allowing them to edit text, turn images off and on, etc. Based on these new settings, the “NetMonster” conducts another search and rebuilds the image. The result over many such iterations is a collaboratively built networked image which continuously changes and offers up new content and configurations. In this way the software autonomously generates thematic content, composes it and displays it and then allows its content and thematic focus to be modified by people after each stage.

At present, we need to firstly improve the searching abilities of the “Monster” so that it can build on its current database in a logical yet stimulating manner. We also need to make the editing environment more flexible, efficient and as sensitive to the thematic context as possible. So far the images in the “NetMonster” are only sorted and displayed in terms of tonal graduation with any accompanying text radiating out from them. But the success of the “NetMonster” idea will be greatly dependent upon its use of the image and must therefore be complemented by a careful investigation into its function within a networked environment. We need to develop better techniques for categorising, composing images in an online environment. We will develop aesthetic techniques and software tools that are not constrained to industrial and commercially specialised applications. Another area of research is to explore to what extent we can exploit the context in which we find an image on the web. How much of an image’s context can software detect itself without a users interpretation – how much can be productively gleaned from its quantitive properties, its ownership or its location?

We also plan to add telecommunication channels to the “NetMonster”. The ‘monster’ also returns contact information such as phone numbers and email addresses. Using telephony and voice synthesis and drawing on previous projects (such as “ARoundhead”), we can allow the ‘monster’ to attempt contact with the owners, authors or managers of the search data it returns. In this way it can open up a channel of communication to the owners of the sites whose work it incorporates., perhaps asking for permission to use their work or offering them open content licenses and passing on their response to “NetMonster” users. Whilst the ‘monster’ is in the first instance under the influence of its collaborators, initiators and modifiers it will also be pursuing these internal directives designed to engage people into further participating with it. This will allow it to propagate itself by connecting people associated with its themes and by becoming its own network of dissemination.

This project will be constructed as a set of software modules. The individual functions such as searching, database storage, editing, compositing, image processing and navigation will be separated into components that can be recombined in different ways. This will allow us to reuse the components to provide a much more flexible response to specific needs. The ‘toolkit’ will be the basis from which we will gradually extend the system in the future. This will also be a form which will make it easier for us to disseminate as research which other artists and organisations can make use of.

The “NetMonster” software will not only be a means of artistic expression and critique but will also have the potential for other functions. Its ability to gather, edit and organize search results or databases in general makes it a research tool for conceptualizing large amounts of data. The “NetMonster” could simplistically be said to result in a “snapshot” of the current state of knowledge in the network about a given subject.

Outcomes

The main outcome of this research will be “The NetMonster Suite” - a web interface to a set of software functions for the online resourcing and collaborative construction of the “networked image”. The “NetMonster” will be a responsive, immediate and sensuous space for projects based on networked collaboration. At the end of the fellowship we will officially launch the “NetMonster” project at the Piet Zwart Institute using the public presentation, online exhibition and printed outputs. Mongrel has built up a network of contacts over the last ten years which it will use to publicise this new project and promote it to potential partners and participants in arts organisations. Mongrels previous collaborative system “Nine(9)” was used by ImagineIC in Amsterdam to deliver workshops to nearly four hundred participants. The “NetMonster” is also intended for projects by other artists. Currently, the “Identity Runners”, a collective of three female artists, are planning to use it for a work to be shown later in 2006 and we expect that this take up will increase.

As we did with the students at Brown University, we intend to develop and test this software with the participation of the students of the design programme at the Piet Zwart Institute. At Mongrel we have established a practice of documenting our research using online ‘wiki’ text editing technologies and this will be a way that students can observe, comment on and participate in the various stages of our work. We also intend to leave this available as a documentation of our research and link it to future developments of the software after the fellowship has ended. Mongrel has its own network of web sites which it can use to host these activities or in collaboration with Piet Zwart’s online presence and research pages.

After the workshop at Brown University we printed out the results of the first ‘monster’ in a ‘before and after’ sequence of eight foot square posters. This provides an impressive visual legacy to leave behind but in addition we hope to extend the flexibility of printed output from the “NetMonster”. We are planning to work with the researchers at MUTE magazine in the UK to develop print-on-demand facilities so that “NetMonster” can generate limited edition artists books of work that has been produced using the software.

Technical Practicalities

Mongrel is an artists collective of currently three core members. The software development will be done by Harwood and Richard Wright with design input and coordinating from Matsuko Yokokoji. They will be drawing on a software resource and expertise built up from several years of LINUX based development.

The main programming for the “NetMonster” will be done in Object Orientated Perl, tying together several other technologies such as GIMP functions for graphics generation and MySQL for database functions. We have many years experience of these technologies having previously used them to build projects such as “Nine(9)”, “ARoundhead” and “Lungs”. We have a dedicated rack of LINUX based servers in our studios in Southend-on-Sea but we can either work on these remotely from Rotterdam or develop on the Piet Zwart servers. The “Gentoo” version of LINUX we run is conveniently the same as that used at the Institute.

Because of the amount of processing that is needed by the “NetMonster” to search and render each pass of the process in iterative cycles, we usually need to use several servers at a time. This can be achieved by running processes remotely and designing a pipeline for the workflow. We will also design our software architecture to use modularised ‘plug-ins’ to achieve greater flexibility, allowing users to choose different rendering engines or different editing interfaces. These kinds of facilities will form part of the technical research of the project and we are currently investigating various options for their implementation.