network crab fishing workshop

starlab/FOAM 20.11.2000

these notes are a brief overview from a series of discussions, tangents and diversions which was informally based around the issues involved in using computer networks as an integral part of collaborative projects. Most of the discussion focused around the mechanisms + implementation issues on a quite general level. mostly questions involving 'how' rather than 'what', 'why', 'when' or 'where' were covered.

looking at various ways in which multiple users could affect the output, or contribute to the output of a single data stream. whether in fact it makes sense to think of this as a stream, or as a larger whole, in which users/collaborators are affecting a part of, which may feedback into the rest of the system in various ways. having the possibility for many ppl to affect a system in such a way raises a range of issues in presenting an interface to the system, collecting data + resolving the effects of the data.

if parts of a causally related system require data about what is happening in other parts of the system in order to develop transitions, intermediate states (in which the information is assumed, but not available) can be calculated to give the impression of continuity. however, if there are sufficiently large gaps between events, this intermediate state can take on a life of its own. the longer the gap, the more divergent the intermediate state becomes. this is applicable with high speed networked games, where the simulated world should be consistent for all players, in which critical action can take place in the same timeframe as 'lag time'. the problem of creating a seemingly continuous experience with discontinuous flow of information. pseudo divergent realities, collapsing wave functions.

this lead to a brief, yet convoluted discussion on the nature of time, causality and how to represent discontinuous (and/or inconsistent) time embedded in a single time dimension. most of these questions are open ended and not particulary intuitive, but could lead to some novel representations of dataflow, interaction + influence which would extend current network visualisation + sonification techniques.

the version control + extended 'undo' features as presented in osmic (part of the xanadu system) provide an example of multilinear structure for tracking changes over linear time.

treating all data as indistinguishable, using audio data as video, audio to control graphics, control rates as audio, etc. the main problem is one of scale, data-density or frequency. the creative decisions on how to translate or arrange influences is left to the author(s).

assuming data is equivalent (or forcing equivalence upon it) leads to a range of problems and methods of translation. by looking at processing effects, some of this became clearer. which effects (or techniques of manipulation) are perceptually equivalent in the visual domain + audio domain. what are the 'obvious' mappings, and what are not. fr example, delay is obvious + apparent in a+v domains, reverb probably isn't, audio frequency spectrum → colour ranges works with continuous data, where changes + consistencies can be observed.

discussing various ways of representing data in a domain other than the one it was 'intended' for. algorithmic transmogrification, human perception and “see what happens if…” seem to be the common threads/mechanisms. even though a seemingly obvious point of distinction, we did return to it several times, and it does mark definite differences in approach, not only for us, but in a wider context also. discussions about network topology + representations of network activity covered existing systems + possible future developments.

the role of algorithms which develop their behaviours over time and in relation to particular thruput was suggested, but not developed extensively. it seems there is a range of applications which could benefit from such an approach. one specific application was the modelling of the complexity of an arbitrary data set, with minimum required dimensionality. an instrument would then be generated with the same control dimensionality) to play the data.

the problems of describing + selecting sounds in an evolutionary system were discussed briefly, along with extending existing systems (tierra + mutagen) to produce sound output in parallel or instead of their present output.

the most successful method for developing distributed instruments (which can be used for performance, or to automatically generate output) so far seems to be using the Open Sound Control protocol to pass data between separate machines running max/msp and/or pd patches (programmes). this also enable the control interface to be separated from the computationaly intensive rendering of sound + image data. naturally, anything which works over a local tcp network, can just as easily be used over a wide area network such as the internet.

this technique of separating the 'interface' from the 'rendering' opens a range of possibilities on local networks for utilising peripherals attached to another machine on the network (eg. video cards, multichannel audio outputs, printers etc).

the resolution and unreliability of midi devices lead to a consensus that it is a protocol best left to musical instruments (as intended). the problems tend to increase proportionally with the number of machines connected using midi. (no exact relationship has been formulated)

in discussion of using feedback + noise to determine the nature of an unknown system, the subject of process physics came up. in brief, process physics aims to understand the nature of reality by modelling processes. one of the central concepts in this modelling is the use of self referential noise to model a pregeometric universe. while essentially a diversion, this discussion hilighted some of the differences between designed system, and growing systems.

we briefly discussed several systems for distributed performances, or networked based collaboration. our focus was on realtime, audiovisual systems, which are aimed at performance, however there are a range of tools used for asynchronous collaboration which were not mentioned, and are beyond the scope of most of the systems discussed.

open sound control; developed at cnmat which facilities the exchange of data over a udp network. optionally timecoded messages, can be sent within a heirarchic address space, to multiple recipients.

  • pd; graphical programming language.
  • max/msp; graphical programming language based on pd.
  • kromozone performance system; built within max for specific types of performances.
  • keystroke “cross media synthesizer”; a system to enable media elements (or range of data formats) to influence each other, or be influenced over a tcp network. still in development, runs on macintosh only, but in principle extendible.
  • the matrix; hardware solution for connecting a range of machines using different protocols, developed by future lab (aec). in development?

general discussion on the lack of adaptable systems, the range of solutions to specific problems, the ability or difficulty to generalize these systems.

a proposal for a system of interconnection + interoperability with a range of networked subsystems was discussed. some specific developments were proposed, one was the idea of using protocol dependency trees which would facilitate graceful degradation. another was to extend the ideas of class inheritance (from the object oriented programming paradigm) into a content description or property inheritance structure.

gravitational attractors

  • methods of transforming data into other domains/dimensions
  • equivalence of control/data/audio/video/etc
  • mappings
  • interfacing with public/performance/private spaces
  • differences between 1st,2nd+3rd person perspectives
  • differences between human perception and machine methods * in the process of translation(s) * ease and density of interpretation(s)
  • transitions between translations
  • multidimensionality
  • surroundiness/ambience/seamlessness

primary chaordic attractor

  • connecting everything to everything else

mechanisms

  • trial + error
  • feedback
  • evolutionary methods
  • learning algorithms

links

sutChwon Article from TimesUp Newsletter » http://phl.cx/sutChwon/ctl_article.txt

http://www.rfc-editor.org/repositories.html

RFC:2533, A Syntax for Describing Media Feature Sets

RFC:2506, Media Feature Tag Registration Procedure

RFC:2295, Transparent Content Negotiation in HTTP

RFC:2703, Protocol-independent Content Negotiation Framework

conneg Working Group http://www.imc.org/ietf-medfree/

– transcribed by Nik Gaffney


http://xdv.org/weblog/975069610/index_html

Re: further adventures in intertwingling by Timmy B on Wednesday November 29, @12:46

Time: 2D and other times relate to the parallel processing problems, things like partially commutative monoids and othe objects - there is a bunch of french research about this from the late 80s that I have somewhere buried in the archives… time to break out the shovel

Noise: feeding coloured noise into a system is used for detecting nonlinearites in its responses. I think the feedback principle (FEEBDACK) can be used to accelerate and emphasize these tweaks in the response. connect mixers to themselves and follow the crests between feedback and silence….

no URLs yet

an interesting article published in leonardo describing the processes involved in making music with dna + protein sequences.

http://mitpress.mit.edu/e-journals/Leonardo/isast/articles/lifemusic.html


Re: farther ut ventures in*2 tert wingling by xgz on Wednesday December 06, @11:50

Hi Maja, Nick, Tim etc

read through the files on xdv.org also took the references etc.. these are really interesting and i will try to expand them with my own references and links later on ..

one thing that bothers me though is the idea that there are 'hidden' structural similarities between to me very disparate (desperate) things, and therefore, the search for mapping is to me a rather futile one, I try to put effort much more in looking for algorithms/protocols that generate similarities - I mean intermediate organisms - a thing that I expect to be more closely related to artistic and scientific activity… Isn't mapping ocurring automatically when we deduce values in both physical domains (audio and visual) and try to set up probes and synthesisers for either output?

getting bored with transformations of reality lately - think we need to replace the whole sampler aethetix with anything that is more promising… I want to make new things

feedback etc → refs to david tudor's work - I just found this quote, but i guess there is a lot to say about his way of dealing with electronics and sound material - something that comes closely to a more “ecological” approach?

“Rather than think of tone generators or recordings of natural sounds etc., I experimented with principles of amplification, trying to make amplifiers oscillate in an absolutely unpredictable manner.” “In the end, it turned out that I didn't even need any amplifiers because most electronic equipment uses the principle of amplification. You need filters, modulators and mixing equipment which have gain stages. By piling these components up, I was able to work without any sound generators and I made several pieces in that manner. ”

romanticism = http://hollerbickerstaff.com/1999review/MusicBox2/Mbox2.htm collaboration and current network situations yes closing the loop - one remark - I haven't been keen on the play-metaphor because it obscures a lot and puts everything in rather a different light and I hate this lightheartedness and easiness - and you are quite right to say that the academics flip on this: on top of that - they haven't had much to say over the last years I remember… I mean besides Axelrod or Holland and talking about Holland, I quite enjoyed Hidden Order: How Adaptation Builds Complexity and in a way the different parts of the model that he is proposing in that readable work, is pointing more to some kind of “collaborative” environment than I understood myself at first (one ref = http://www.santafe.edu/~shalizi/notebooks/john-holland.html) Another favorite of mine is Gary William Flake, The Computational Beauty of Nature because it is like an anthology of modern scientific code / concepts

different layers: as an abstraction I can live with this, but I think dealing with online collaboration it is all very complex and cannot be pinned down to exactly these layers, … and since i try to avoid lists and matrixes, dividing activities in different layers mmmm I like more the idea of a tubing set…

and then i discovered the extended description oh I see .. yes I agree and don't have to add anything to it - yes I hate midi as well and I like osc!

now about distributed performances, and this is the socio-cultural factor, is the role of the “audience” most of the time they are negative evaluators, reasons may be diverse

  • it is too new and they evaluate it as a traditional performance
  • they hype the network aspect and when this is not explicitly visible they are disappointed (they want to see something simple happening)
  • they hype the network aspect and when this is explicitly visible they are disappointed (they want to see something complex happening)
  • frustration of not being able to participate in it?
  • etc ??

my idea is that this is at the heart of the design of the system and f you want to develop something you will have to come up with a better solution than whatever was tried in the past

btw did not have time for keystroke yet .. but i like supercollider as well: http://www.audiosynth.com/

Guy van Belle


Re: in*2 tert wingling mapping metabolismi by i22 on Wednesday December 06, @12:19

one thing that bothers me though is the idea that there are 'hidden' structural similarities between to me very disparate (desperate) things, and therefore, the search for mapping is to me a rather futile one, I try to put effort much more in looking for algorithms/protocols that generate similarities - I mean intermediate organisms - a thing that I expect to be more closely related to artistic and scientific activity… Isn't mapping ocurring automatically when we deduce values in both physical domains (audio and visual) and try to set up probes and synthesisers for either output?

it seems there are two aspects to this, one is perceptual, the other exploratory. of course there is a huge ammount of this 'mapping' going on all the time, humans have an excellent ability to impose patterns, to actively try to find patterns (seeing/hearing things in tv noise is a good example). im interested in finding patterns/systems that bring these abilities into play.

the exploratory aspect is related to this idea of 'hidden' similarities. however, im not sure if the process of mapping things into different domains is a seach for related structures, or a search for ways of finding patterns that are not apparent, or 'visible' in the original domain.

i like the idea of these intermediate organisms, i also like the idea of an organism that takes something + transforms it into something else. the mapping is the result, the metabolism is the algorithm (or process of mapping)

Speech Synthesis

Hidden Order

SukcSpit

lagging behind skipping ahead

– needs serious refactoring


content negotiation

protocol negotiation

protocol sketching » http://etiquette.sourceforge.net/)

service discovery

MediaSock http://mediasock.org/

Service Location Protocol http://www.openslp.org/

RFC:2608 - Service Location Protocol, Version 2

RFC:2609 - Service Templates and Service Schemes

RFC:2610 - DHCP Options for Service Location Protocol

RFC:2614 - An API for Service Location Protocol

ad-hoc networking

ad-hoc data exchange

ubf (a and b) is a language for transporting and describing complex data structures across a network. http://www.sics.se/~joe/ubf/site/home.html

media sharing