User:Laurier Rochon/work/listener

To see The Listener live go here

The Listener

The concept of the speed show within a Belhuis is, to me, about limitations. It is also about performance (in both its senses) and simplicity. Perhaps the greatest limitation is technical, as we cannot 1)rearrange/reconfigure hardware 2)use our own hardware 3)have limited software to work with 4)and cannot install our own software. The piece will therefore have to live within a (potentially crappy) browser, and perform optimally even under terrible computing/bandwidth conditions. For that reason, I would like to build a lightweight, text-based, multi-user application that can be interacted with from different Belhuises in the city (or the world, for that matter).

Description

The Listener is a multi-person chatting program. The conversations are mediated by a "third person", a computer script that adds, subtracts and replaces words according to a certain "mode" it has randomly set itself into.

Concept

If the assertion that text and reader implode to create a 'new' third thing (often called meaning, or subjectivity) holds true when one reads a text, then it is this new creation I am trying to address. By slipping itself between the textual exchange, The Listener shifts the conversation slightly by altering certain words and sentences. In most cases, confusion ensues and new trajectories are followed by the participants, depending on the level of alteration (ranging from none to extreme). When a user sends a message to the other ones, he or she sees the original text but the others see the altered version. This technique of showing/concealing is a sneaky strategy meant to underline the subtle inner workings of computer programs when mediating the information we consume online. I have written some words on this subject here.

Technical

The chatting program code is based on Kenrick Beckett's chat engine using only 4 files - the main index.php, a CSS file, a javascript file (ajax functions) and a php functions file. Using ajax, people send textual data to a file stored on the server, and when returned to the users' screens, a filtering process (in javascript) occurs to show different contents if mediation has occurred. All conversations are logged in flat text files, eventually to be reused for a naive AI engine.

The pipeline

user input (js) -> ajax call to php function -> medtiation script (python) in CGI -> writes to conversation log file -> ajax call to php (read the text file, check for changes) -> js send to users' screens (content dependent on the author)

In action

• Splash screen

• Mediation occurs. The author sees their original text, the others see the mediated version.

• What the text file looks like - the delimiters could be a bit more intelligent, I know.