Collective scales of knowledge

By Manetta

Another extract of the interview with Matthew Fuller.

"Friedrich Kittler proposes that the early 70s was the last time any single person knew what was going on in a particular computer, now the complexity of each semi conductor, each circuit is such that it has to be jointly held knowledge or different teams of specialist within a company would be able to describe this.

That is interesting because there is a kind of threshold of knowledge that has been passed. It's often said that Leibniz was the last real polymath, who was able to operate across disciplines, you know this was in the 17th century.

And if what Kittler suggests is true, then we have a condition where a single artefact has become so complex that a single person can not describe it in all its detail anymore. And this is very interesting state being on the one hand it means that the traditional form of knowledge if humanities knowledge is being held by an individual person is no longer tenable and we have to start working in ways that are collective, to think about what is the form of knowledge that is appropriate or what is the form of research that is appropriate to the humanities and the arts if collective working becomes more and more a way of gaining traction on different scales of reality, then maybe this is something that we have to deal with? But it also that, since the systems are less knowable, there is more capacity in some ways for approaches that de-structure the possibilities for control, there is the possibility within a grounds of a more or less knowable more or less unknowable system for certain kinds of autonomies, certain kinds of freedom certain kinds of experimentation to be established, so the unknowable is also useful in this regard."

Thanks to Colm for the transcription! (link)

Link to the interview on Macba:


Pandora's boxes

By Gijs

Our research can also be described as an excersise in dealing with the uncertainty within machine learning. While the science of machine learning reveals itself through scientific papers, the practice of machine learning seems to consist of an equal amount of gut-feeling and practical experience. We imagine the data-scientist using his intimacy with the data and gut-feeling to design his net. Taming the quirks of the net in the edge cases while training it.

We wonder untill where to enter this beast? Can we ever fully understand its inner workings or psyche?

Reading a scientific paper sprinkled with jargon feels like discovering a collection of black-boxes. On the discovery of such a blackbox do you accept its advertised function or open it in an effort to understand its inner workings? While seductive, it seems but a step in the discovery of more. If we open all of them it feels like we are drilling down rather than developing a general understanding of the subject matter through a more lateral movement.

In our explorations through the field we started to recognize planes or groups of these blackboxes:

  • Frameworks like TensorFlow, SciKitLearn or Torch are tools offering machine-learning techniques through abstract, reasonable interfaces. While using them is relatively straight forward their code by itself is very specific flavoured by implementation details and optimizations.

  • Then we see clusters which can be used in neural nets almost as building blocks, in the case of a neural-network an LSTM-module / neuron.

  • On deconstruction of such blocks we encounter a collection of techniques often originated in scientific papers.

  • Finally their is the foundation in mathmatics and statistics. For example, the Algebraic techniques which are fundamental to Neural Nets and a more efficient backprop.

The borders between them are blurry and intersecting while the ordering depends on your perspective. The question remains, untill where do we dig? Can we understand ML without understanding the underlying algebra? Is the desire to understand everything productive and necessary or originating from distrust or petty ambition?


Understanding the black box

By Manetta

Web radio interviewed Matthew Fuller published on the 11th of August 2017. In this extract of the interview, Fuller speaks about different ways to think about the idea of the 'black box'.

Link to the interview (starts at 6m10s):

"The black box in contemporary science and technology studies is a term that is used to describe an entity within which we cannot know what is happening. We have to accept [it] as a stable, standard object that is internally unknowable, but externally we can inspect its functions.

The black box as a term comes from cybernetics. In W. Ross Ashby's work, where he describes the black box as a form of jargon that comes from electronic engineers. So in WWII there is an enourmous production of electronic gadgets that would be developed for things like submarines, battleships, aircrafts and so on. Because they were produced so rapidly and [because of the fact] that they were partially experimental, they weren't documented, they didn't had a great user interface -- often it would just a series of switches around a grey metal box. These objects were finished while the war was ongoing. If one was presented with one of these objects as an electronic engineer, you would have to work out what this thing was. There were various routines of providing with input and seeing what its output was. And by mapping the input to the output, with the various changes that were introduced by different forms of input, voltage levels for instance and the turning of different switches, one could see what this thing was. And at least make some kind of educated guess.

So the black box in cybernetic sense is always a dynamic object. Something which very different from a static and slightly oppressive object in technology studies. For Ashby the black box is always something that requires inquiry. [Something] that involves a dynamic relation between the experimental subject that is testing what the box is, and the function of the box itself. This is a different position in a sense.

What I'm interested in is perhaps to move away from a simply critical position that says oke, these things are black boxes we can't know them, they oppress us, and they govern our actions. It may indeed be true in some cases. But in other cases it is possible to come to them with an experimental frame of mind and to work with concatenations of input and output to see what happens. The more black boxes that are joint together, the more contingent, the more unstable perhaps the arrangement would be. Depending on particular characteristics.

The question of the black box in contemporary society is something that i think we can experiment with.

It think it relates to also the question of what the human is. If you think of Freud, the account of the human from Darwin, from Marx to Nietzsche. The question of what the human is is always subject of experiment, subject of question, and subject of a kind of renewed sense of a cascade of multiplicities that are involved in the human. Which are always partially knowable and partially unknowable and chaotic in form. Which is a fantastic place to be."

Communitity detection in the Stedelijk Museum archive

By Manetta

The project SMTP: Stedelijk Museum Text Mining Project was a study to the use of machine learning techniques in archival practises. The project is a collaboration between the Stedelijk Museum, the CREATE group of the University of Amsterdam, and Maastricht University.

The phrase community detection is an interesting way to look at the creations of common vocabulary over time.


poetically observing

By Manetta

While considering possible ways to position the book, the following angles crossed our path:

[en] poetic / contemplative / observative / artistic / functional / speculative / specular / signaling / meditating / reflecting / activating / learning / studying / a meeting book

[nl] poetisch / beschouwend / observerend / artistiek / functioneel / bespiegelend / bekijkend / signalerend / mediterend / reflecterend / activerend / loerend / bestuderend / een ontmoetend boek

een artistiek beschouwend boek / an artistic contemplative book
een artistiek bespiegelend boek / an artistic specular book
een po√ętisch beschouwend boek / a poetic contemplative book
een meervoudig beschouwend boek / a multi-sided contemplative book
een meervoudig observerend boek / a multi-sided observational book
een bespiegelend observerend boek / a reflective observational book
een bespiegelend reflecterend boek / a specular reflective book
een po√ętisch observerend boek / a poetically observing book


A conversation around algorithms

By Gijs & Manetta

We presented our project during a conversation around algorithms.

The Stimuleringsfonds invited a group of designers and artists working with algorithms in their projects, for an afternoon conversation.

The presentation we prepared for this occasion can be visited here:


Algorithmic Uncertainty

By An & Gijs & Manetta

After having worked on the topic of machine learning in various individual and collective projects in the last two years, we (An, Gijs & Manetta) are very happy to announce the start of a new trajectory that lives under the name Algorithmic Uncertainty.

Machine learning is a technique of recognizing patterns, to develop knowledge on the basis of data. It's a way to automate the process of transforming data into information. The belief that this is possible, seems to be the basis of the technology.

We developed an interest to understand how feelings of uncertainty are embedded within a machine learning model. To do this, we decided to focus on moments of uncertainty within a machine learning practise, an area that doesn't stop at the source code, but includes information about the specific situation in which the model is developed and the different people that were involved.

We are currently in a pre-trajectory of the project, testing and exploring ways to translate a machine learning model into a poetic-observing book. In the second stage we will develop this research in an poetically observing book.

This blog is a tool to keep track of our research. And of course reactions are welcome!