Florian Kleedorfer on intelligent agents, the Web of Needs and what critical theory has to do with it

Florian Kleedorfer is the operative studio manager of the RSA FG-Studio “Smart Agent Technologies”. In the context of his diploma thesis (TU Vienna and University of Linz, 2008) he developed an algorithm for the automatic recognition of topics in song lyrics. Florian Kleedorfer is an expert in Artificial Intelligence, especially Machine Learning, Information Retrieval and Text Mining.

Read an interview with the researcher and inventor of the “Web of Needs” here.

How did it come about that you became interested in Smart Agents Technologies – Intelligent Agents?

Computers fascinated me from a very early age. Today I would say that it was because although they – as machines – execute the given programs, they were part of the environment of us children, which had to be dealt with somehow in computer games, I on the other hand suspected that we could shape this outside world according to our ideas if we only learned to program. So there was a sense of empowerment associated with the idea of programming early on, of being able to control that which controls us – in the game – ourselves.

Is that why you studied computer science?

Yes, and later a great interest in computer graphics and 3D models was added: an exploration of the virtual worlds that the computer creates. In the course of my studies in computer science, I delved into the field of Artificial Intelligence, where the aforementioned empowerment is subtly reversed: by imitating human learning processes, we naturally continue to determine what the machine does, but we do it on a different level. It is no longer done by clear commands directed at the task, but by preprogramming a certain adaptivity. In the SAT (Smart Agent Technologies) studio, I was able to use such methods from the very beginning to elicit IT systems to work in surprising ways that make sense to users. In the first years, we used various approaches to calculate suggestions that fit a certain situation or observed preferences of the users. This should change.

What new prototype are you and your team currently building?

We develop what we call the “Web of Needs” and specialize it for different use cases. The current status can be tested at any time on our demonstrator: The use cases we are currently investigating come from the following areas:

  • General classifieds
  • Freight transport
  • Taxi
  • Rented apartments
  • Car-Sharing
  • Fairs / Conferences

In each of these use cases, simple interactions between the different roles are made possible – arranging transport, calling a taxi, making an appointment to visit an apartment, or making interesting contacts at a trade fair.

To what extent is the “Web of Needs” with its prototypes unique?

All these prototypes have in common that different installations of the server software can communicate with each other. This means that no isolated platforms are created. Instead, the overall system grows with each installation, only different objects are represented for each use case (e.g. taxi rides, apartments, etc.) and different interaction protocols are used (call a taxi, make an appointment for a visit, etc.) Where it makes sense, objects from different use cases can also cooperate with each other or be used alternatively, as in the case of taxis and car-sharing cars, for example. In this respect, the resulting prototypes are unique.

What problems does the “Web of Needs” solve?

Each of these use cases is already solved in many cases by existing closed platforms. However, the core of the business model of these platforms is to satisfy the needs of the users only within the platform (for example, as an over-user I find only over-drivers).

In contrast, needs that are published in the Web of Needs become potentially visible in all installations. So I can assume that I only have to post my need or offer once to find the whole selection of possible cooperation partners.

There is also a ‘long tail’ of use cases for which no profitable business case exists and therefore no platforms/apps are created. Such ‘long tail’ marketplaces could be served by the technology developed by SAT.

What are the main components of this new prototype?

  • Owner Application: a web application with which a user can post ‘Needs’. This can be an offer, a demand, or another intention. When you try the demonstrator, you interact with an Owner Application.
  • WoN Node (Web of Needs Node): Server where the ‘Needs’ are stored and where the communication connections with other Needs run.
  • Matching Service: A program that analyzes the data stored on WoN nodes and, if necessary, initiates connections between Needs by sending ‘hint messages’.
  • WoN Bot: A reactive program that can create and manage Needs and communicate with users of the system through those Needs. Such a bot is used to connect existing Web APIs to WoN.

Why does the “Web of Needs” represent an added value for the user?

From a perspective of the future, WoN is a needs-oriented medium that makes cooperation and mutual coordination automatable and thus simplifies them. Cooperation partners for arbitrary interactions can be found ad hoc, without having to choose the appropriate platform first and then being limited to the user group of this platform. This leads to a change in the way we deal with our own wishes and plans, because it makes it possible to manage them in one place (the preferred Owner Application) instead of having to split them across many platforms or apps and partially duplicate them as is currently the case. The representation of the wish instead of the offer satisfying it, as is currently usual, also allows a different, more self-determined access to one’s own needs.

They belong to those “IT experts” who also reflect on their work theoretically. What does your work as a computer scientist and engineer have to do with critical theory?

Parallel to my work in SAT, I was engaged, together with friends, in the social criticism of Karl Marx and the later proponents of critical theory in this tradition. Over the years, we worked through the “Critique of Political Economy” among other works, from which I drew two essential conclusions:

First, we live in a world dominated down to the finest detail by capital, i.e. by processes of exploitation that are intertwined millions of times and run in parallel. We humans have set these processes in motion and maintain them, but we have no control over them. Rather, they are emergent phenomena that emerge from the exchange relationships between people and have an impact on them. However, a truly humane state would be one in which people make rational decisions about the use of their personal, social and natural resources. Such a state seems to me logically desirable.

Secondly, however, no one knows how this state – this society – should function. All attempts to create it have failed horribly. Marx, and subsequently critical theory, addresses this problem by prohibiting the creation of a positive image of a better state. Instead, the critique of the present world is supposed to give the negative image of the other society. This idea makes sense to me, but it runs counter to my engineering thinking of not only criticizing the world, but also wanting to change it.

These preconditions led to the question: How far can we venture into the positive, how much can we already say about a ‘liberated society’, and can we derive a useful, productive course of action from this? The answer I came across was the principle “To each according to his needs, to each according to his abilities”. As long as one does not interpret compulsion into it, it is the inescapable basis of the communist idea. It is about the mediation of human needs with the human possibilities for their satisfaction. This mediation – the decision as to which needs are satisfied and how they are satisfied – takes place today via the market. In a differently functioning society, this happens in a different way. Imagining it is rightly subject to a ‘ban on images’.

The Web of Needs as a kind of alternative marketplace to turbo-capitalism?

In my work at the SAT I got to know technologies like matching systems or the Semantic Web and started to guess: Today we have the technical means to map at least this first level on the Web, to understand people’s needs and the possibilities of satisfying those needs and to calculate possible solutions. Such a system could use the market as an allocation function, but it would also perhaps be a piece of useful technical infrastructure for a society that functions quite differently, regardless of how it solves the allocation problem.

We at the SAT studio have set ourselves the goal of developing this technology. Under the name ‘Web of Needs’, as mentioned above, a generally usable, marketplace-like system is being created, whose individual instances are compatible with each other and can communicate with each other, so that an offer or a need, which is published once, can be a starting point for a cooperation for all users.

We live in a time of ‘platformisation’, in which the strongly centralising tendencies of traditional web-based business models are becoming apparent. As a decentralized counter-movement, block-chain-based solutions are being propagated, at the basis of which a kind of currency can always be found. Web of Needs, on the other hand, offers a middle course: decentralized, but without engraving the aspect of payment into the infrastructure. To develop this middle way as a really useful system along real use cases is an enormously challenging but also very fascinating activity.