RandomForests - Do Machines need training forests?

This text was written in reflection of Un-Green conference, Riga.

This summer part of the amazon and the arctic are going up in flames. It underlines that Nietzsche was wright when he had Zarathustra plea for people to stay loyal to the Earth. The world doesn’t need more gizmos or start-ups, the world needs care, collaboration, loyalty to natural systems and certainly a few less fascist lunatics at various steering wheels. Now is as good a time as any to rekindle the noble art of giving a fuck about life on planet Earth. Not life as in life-style, but as in the exceedingly rare dance of ever diversifying creatures involved together in The Great Conversation.

Looking back,its hard to pinpoint when machines stopped being in service of humans and started to be in service to the economy directly. A.K.A late-capitalism, when planet Earth is transforming at great speed into planet Trash. We build nuclear reactors but can’t even manage a simple plastic bag without ending up with Great Pacific Garbage patches. Let's be clear, we were never in charge. We were always trying to keep up with change, be it natural of cultural. If machines were ever in our exclusive service, they are no longer. Pinocchio has ceased to be simply a story about an inventor wanting a son, but has become also about a machine that finds an opportunity to get built. The trade-off is that it has to appear human. If only Gepetto had made a garden to fill his emotional void.

Looking at our track record makes you wonder how valid it is to take human intelligence as the model for artificial intelligence like a present-day Pinocchio. Why not model it on Elephant-intelligence, Whale-intelligence, or Orang-utan intelligence? Shouldn't our machines also learn directly from non-humans? Should artificial agents only grow up in corporate environments, or do they need training forests?

Humans desperatly need training forests. The tragedy of our times is that almost none of us has ever been in the presence of the full force of life. Biodiversity at full volume. We only know the radically depleted version. That’s our normal. That’s almost all we have from which to imagine the full spectrum, to imagine what we’ve lost. Perhaps that is one of the most degenerated facets of human behaviour, the willing inability to sense a loss. To let the Amazon burn while carving out Pinocchios.

 RandomForests


Moth Snowstorm - the value of fieldwork in artistic practice

This text was written in the context of fieldwork with students of Beaux Arts Nantes and Members of PING.

My grandfather’s generation was used to a natural spectacle that writer Michael Mccarthy describes as a ‘Moth snowstorm’: in the summer there would sometimes be so many moths that you would walk in huge clouds and could not see anything anymore, like a snowstorm. That was not in the Amazon, but in Europe less than 100 years ago.

When I visited my friends in India in the rainforest for the first time, it was a great shock. I was 30 years old at the time, but I realise now that it was the first time that I was in the presence of the full force of life. It was a life changing experience. In Europe we only reach 50% of full biodiversity in our nature areas. So most us grow up without even knowing what the full force of life is like. Ecologists say you can find it in the last old forest of Europe in Poland, and the end of the Danube river in Romania. I’ve not been there yet. In Holland we only have what some ecologists call ‘green concrete’: it looks green, but is almost completely dysfunctional and slowly dies. That has become our ‘normal’. We don’t really see how poor we are.

So when some of the students in Nantes asked, do you like our new building of the ArtSchool? Then I can say I understand the building. It is a ‘white space’ that helps us think about our work, like the white pages of a book help us to read the story printed in black. It may be good to realise that the ‘white cube’ of an art school or a scientific laboratory has only been ‘normal’ for a short period in history, and now we need to move on. The great challenges for our culture are in the mixed zones outside. I work with artists and designers and scientists to do research outside the lab, in the middle of landscapes to engage in the full complexity of those ecosystems.

The white cube is an artificial space. The materials, the energy, the inspiration in our work don’t come from there, but from the complex environment outside that we share with other beings. My friend Suprabha who has been working so save plants and animals in the rainforest in India for 20 years, says we are only fully human in the presence of other beings. Over time I have come to understand what she means. I dream of working in the middle of a Moth snowstorm. So that is why my workspace is now an outside meadow with fruittrees.

 PiNG


Cultural Geology - Mammoth Soup

A short report of the Badgasten residency in 2014 with Satellietgroep on the Sandmotor, a temporary artificial sandbank on the Dutch coast.

Through a nebulous sandstorm in which the entire surface of the landscape seemed to be on the move, a group of determined travelers slowly became visible. They have set out on an arduous journey leaving the mainland behind. Like the early travelers moving north after the receding ice of the last glacial maximum or 'ice-age' they were moving into new territory. Our group was stepping onto the same sand that formed an area called Doggerland that connected Holland and the UK which disappeared beneath the North Sea thousands of years ago. Some 50 million metric tons of it have now been brought up mechanically, to form a temporary landscape: the Sandmotor.

This group moving through the sandstorm was part of 'Cultural Geology', the second sandmotor expedition initiated by the Sattelietgroep in March 2015. The event was two-fold: an exploration of the Sandmotor followed by an inventory by experts Bert van der Valk, Fred Sieval and Bram Langeveld of the fossilized objects that can be found there and a culinary experiment by resident-artist Theun Karelse, that seeks to connect with the ice-age remnants through direct sensory experience; is there any taste left in these bone fragments? Can fossils be ingredients that link back through centuries, to form a 'Mammoth soup'? Could we be the first people since the ice-age to taste mammoth?

The default way to approach fossil remnants is conservation. If you don't conserve them they will start to fall apart due to the exposure to oxygen and the crystallization of the salts inside them. For the smaller bone fractures there is little academic interest in conservation. That triggered the idea to do something different with the fossils, to activate them in a way. Besides applying them as ingredients, Theun selected one particular object to let it deliberately fall apart. The ideal candidate from the collection of fossils he found on the Sandmotor was a tusk of a very juvenile Mammoth. Due to its chemical structure it is almost impossible to conserve. So why not utilize its tendency to fall apart? In collaboration with Edwin Dieperink, a glass blower and instrument maker, Theun developed an hour-glass in which the mammoth-tusk is left to fall apart, creating a 'doggerland-clock'. The clock actively applies the decomposition of the fossil, challenging standard methodologies of dealing with historic artifacts and fossils, to explore the process of decay as a meaningful alternative.

The soup-experiment undertaken in collaboration with experimental cook Sjim Hendrix and the doggerland-clock formed the multi sensorial counterpart to an essay written by Theun Karelse to conclude his dual residency with Sattelietgroep and Nida Art Colony. It contemplates the parallels and differences between the last climate change event and the current one, in terms of ecology and culture and it explores what a positive agenda for the future could mean. How can humans contribute to a healthy and functioning landscape, that includes us? How can people, their technologies and infrastructures become symbiotic partners in the landscape? What is this post-natural landscape? In these writings he formed the outlines of a new project 'Machine Wilderness' to prototype these hybrid man-made and natural systems. Can human infrastructures be redrawn in a way that is inclusive of the patterns, cycles and needs of other animals and plants? Can our infrastructures align to biodiversity and ecological boundaries they are situated in? Machine Wilderness takes specific locations as case-studies for designing local technological interventions as part of that ecosystem.

 Badgasten Satellietgroep


Machine Wilderness - a world of ecological robotics

An interview with Theun about the project on WMMNA, written by Regine Debatty.

Machines as we know them are man-made infrastructures that usually encroach into nature. They extend the internet of things to the animal kingdom, fashion landscapes according to men’s needs and far too often attempt to dominate, exploit and shape the environment.

Machine Wilderness is a research project that calls for a different role for technologies, artificial entities and man-made nodes that would work in harmony with ecological flows, rhythms, and processes.

I discovered the existence of Machine Wilderness a few months ago while participating to an exciting panel called Techno-Ecologies: Art and Machines, at the Renewable Futures conference in Riga. Theun Karelse was there to present his research into Augmented Ecology. His talk was full of birds of pray trained to fight drones, drones hunting wild pigs, robots fighting invasive starfish, robot birds chasing birds over airport runways, etc. The artist explained how this collection of technologies in nature were feeding into his interest for new models of interactions between machines and ecology and how it eventually led to Machine Wilderness. The presentation was fascinating, too short and called for an interview!

Theun is an artist, illustrator, prototyper and interdisciplinary researcher at cultural laboratory FoAM in Brussels and Amsterdam. Here’s what our email exchanges looked like:

Hi Theun! Could you first tell us how Augmented Ecology evolved into Machine Wilderness?

I navigate by gut-feelings really and then during a project I figure out what it actually is I’m up to. Augmented Ecology was such a ‘hunch’. FoAM – of which I am a member – had been awarded an honorary mention during Ars Electronica for creating a smartphone app connected to nature called Boskoi. It was made for mapping wild edible plants. I started the project because I wanted to be in touch with nonhuman life around me in the city. What lives out there with us? The idea that a smartphone couldn’t just hold your task-list or find a hotel nearby, but could also play a role in connecting you to your natural environment was received as somewhat revolutionary. Connecting the ancient knowledge of hunter-gatherers to contemporary media.

The development of the app raised issues connected with taking environmental mapping beyond academically trained ecologists to general audiences and having biodiversity data in the public domain. What to do with endangered species, protected species, vulnerable populations? I called the ‘Data Authority for Nature’ – we actually had such an organization in Holland back then – but they were more like a broker for environmental data-sets generated by academic research. Their lack of interest made me keen to dive further into exploring the rapidly evolving ways technology is integrating into our landscapes. That research became Augmented Ecology. I set up a session during Pixelache 2011 and an Amsterdam meet-up, but I think it was a bit too far from peoples mindsets at the time. That changed in 2014.

An image really struck me when it surfaced on Twitter in March of that year. It shows a small sign attached to a safari-vehicle. It indicates that poachers are using the GPS data from images – uploaded to Flickr and Instagram by tourists – to hunt for rhinos. The term cyber-poaching was coined. WWF was running a promo at the exact same time showing a team spotting Rhinos with Google Glasses. There was a naivety about the complexity of issues and opportunities for these kind of innovations. These things confirmed my instincts. The much discussed ‘Internet of Things’ was extending also into non-human lives, natural processes, ecosystems and habitats. They were going online too. Massively so.

In the background, a race was unfolding between superpowers towards realtime planetary biomonitoring. 2014 saw the birth of ecohacking platforms like Nerds for Nature and Global Eco Hack. Face-recognition software was being developed for primates. Sharks and mountain-lions gained ten-thousands of followers on social media. National parks set-up legislation for drone use, to limit the impact on animals in parks. Robots started protecting coral reefs. I saw melting-glacier-sonification-art, massive online animal tracking and species mapping platforms, environmental activists monitoring through satellites, networks of old smartphones listening for illegal logging in tropical forests. And these things I posted. Augmented Ecology became an archive.

What also stood out was how GPS loggers carried by animals were evolving from simple tags into rich sensor-packs called data collars. Sounded like a great metaphor at the time of WikiLeaks and Snowden. These devices were starting to be applied to building up ‘libraries of behavioral signatures’ not just for individual animals but also at population-level, like a Facebook for animals. Microsoft’s ‘Technology for Nature’ program started testing data-loggers as anchors for drones who could plan and execute monitoring missions autonomously. So the herd gets mirrored by a fleet of machines. That is when the idea of Machine Wilderness really took hold of my imagination and now the subject did resonate with people (and funding bodies).

Machine Wilderness became a discussion and prototyping platform investigating the role of technology in landscapes. It started with a symposium at Artis Royal Zoo bringing together speakers from related fields like AI, conservation drones, landscape architecture, landscape philosophy, art, sustainable agriculture and choreography. The enthusiasm for the subject went beyond any of our expectations. The symposium had a full house and even a sizable waiting list of people of different ages and diverse backgrounds. Most importantly, the program attracted a majority of women. To a program focussing on technology and robotics! I’m proud of that. It shows me we are on to something. Maybe for people this project is really about intimacy; can we relate to life, our living landscapes? That is what attracts me, I want to learn how to listen.

I would somehow prefer to enjoy wilderness without the presence of machine. But that’s probably very naive of me. “Machine Wilderness explores what our technologies could look like if they are native to our landscapes.” What do you mean by ‘native to our landscapes’? What would be the main characteristics of these machines?

I feel the same. Machine Wilderness does not advocate the presence of machines in our landscapes, nor does it subscribe to environmental engineering more generally. I hope it even addresses some of the issues behind green tech. The thing is, machines have been out there for a long time and will be for the foreseeable future. How machines and technologies in general, are designed is based on our imagination of them. We decide what is required. The overwhelming majority of human technologies and infrastructure is human-centered to the point of ignoring the needs of non-human life, and in many ways even our own. That’s a choice, even if it is made unwittingly.

So what if we approach things inclusive of the other 99%? Not to save the world, but because it is a fascinating question. Out there, humanity, our infrastructures and the natural world co-exist, whether we choose to conceptualize them as separate realms or not. A primary influence in this has been Timothy Morton, but also my friends Maja Kuzmanovic and Nik Gaffney at FoAM. Machine Wilderness is my way to try and accept that heterogeneity, and an invitation to others towards an exploration of those hybrid ecosystems, by trying to design artificial agents that inhabit these complex environments. Machines seemed particularly suitable, because of their level of interaction with their environment. Their agency brings out a wide range of interactions. More than say, architecture.

I don’t know anything about robotics really and only scholarly knowledge about avant-garde ecology, but I’ve embraced not knowing. That seems a crucial first step in a fresh approach towards technology. In Machine Wilderness we take specific landscapes as the starting point for designing something to become an integrated part in that biome. In that sense local conditions – microclimates, mineral flows, seasonal cycles, food webs, biodiversity, ecological niches, layers of communication – generate local solutions. Technological diversity. To emphasize this we incorporate local languages where we work like ‘jynnweythek godhviles’ which translates from Cornish roughly as mechanical wild animals.

Being ‘native to landscapes’ alludes to this diversity; being shaped by local factors, histories, opportunities, interactions and demands. Maybe ‘indigenous’ or ‘endemic’ is better than ‘native’, I’m not sure. It’s probably influenced by the way my friends amongst indigenous peoples in Lapland, Northern America, Indonesia and India speak of their responsibility towards ‘all our relations’. A holistic view where relations are not just family, ancestors, offspring, but also include fellow species, forces of nature and geological features. That may be a very different language from engineering.

Why would machines native to the landscape be a good thing? How could the environment benefit from their presence for example?

Only a deeply cynical person would advocate robotics as a replacement for the diminishing abundance and diversity of bio-based life. Fostering biodiversity is what my friend John Lui (documentary filmmaker and restoration ecology advocate) sees as our central task for this century. All signs from field-studies point that way and I admire his efforts and those of other friends like Suprabha Seshan, who lives amongst the most advanced ecosystem restorers in the last remnants of the rainforest in Kerala, India.

In our current discussions we have started to question if our machines should serve any particular human purpose. To break free of the original meaning of the word robot as a worker or slave. Maybe like organisms their primary function is simply to exist and to do that on the long term, they are a stakeholder in the health of their biome, like biological organisms are. You get to a point where you start to see organisms as expressions of their habitats. In co-evolution, organisms form a bigger unit together. Maybe we’re exploring how human infrastructures can start to co-evolve within those landscapes. And we do it low tech.

The actual machines don’t really matter. To me robotics are only a means. There are amazing biomimetic robots being created in places like Bristol Robotics Lab and readers may be familiar with creatures created by Boston Dynamics – as shown in this fantastic 2013 DARPA calendar by designer Daan van Dijk – or the amazing showpieces made by Festo. In their way these efforts are extremely advanced, but we are asking different questions. Can a machine operate at the power levels of biological organisms? Can this energy and building materials be sourced locally? Can machines take part in existing food webs (digest and disperse seeds)? What is their life cycle like (can they die and compost)? In those areas of robotics even simple experiments can be cutting-edge research. From that perspective things like cars start to look like crude and redundant technology from a former age. Machine Wilderness is a way to fully saturate myself with that perspective. How does the world appear then?

Do you have examples of successful, considerate Machine Wilderness? Either from the past or in the present?

A second machine age has been announced with a mix of enthusiasm and apprehension. Machines would take over the labor for which humans have become too expensive (fabrication) or too slow (services). Solving our environmental problems looks extremely labor-intensive. Some robots are being applied to eat the great garbage patches floating around on the oceans. It would be interesting to discuss the potential for robots in aquatic and terrestrial restoration ecology. Currently the terrestrial efforts are almost exclusively towards monoculture agriculture. That’s strange because autonomous robots would liberate us from that model. Crops could be embedded in richly diverse landscapes. It looks like a case of horseless carriage syndrome, the tendency to think of new technologies in the terms of the already existing ones.

Key aspects of our machines and devices are at a primitive state. Projects like Fairphone try to address some of the complex issues. The idea of our technologies as highly advanced technologies goes largely unchallenged. Take GPS navigation for example. Finding your way at centimeter resolution based on satellite systems sounds advanced in a way, but may pale in other ways when compared to other ways of navigation. Aboriginal songlines for instance stretch across the Australian continent and layer centuries of cultural knowledge relying only on the interpersonal relationships that pass on the songs. Hearing a song line being sung definitely makes me feel different than hearing a TomTom voice direct me to a next turn. Analogies like that may help to frame the quality of what we produce as a culture.

As for examples from the past, the history of cybernetic animals has its own dynamic and stretches back way beyond the medieval robotics experiments by engineering pioneers like al-Jazari. It is captured in great depth in Cybernetic Zoo the brilliant historiography by Reuben Hoggett. His archive includes a Cybernetic Creature Timeline and – my personal favorite – a historic overview of Mechanical Elephants, Horses, and other Walking Animals. These all fit pretty much into conventional ideas of robotics. The current dynamics in genetic engineering may however ignite less familiar materializations. Seeing software jump directly towards ‘wet-ware’ creates of course a whole set of new questions, particularly about control and containment, which are under scrutiny in places like the Wetlab at Waag here in Amsterdam and by researchers like Sascha Pohlflepp in his essay Life’s Artefacts (in the Ten-Thousand-Year Garden). We’re thinking about how to incorporate these developments in the Machine Wilderness program. Readers are definitely encouraged to connect with us on these (or any other) subjects.

In Augmented Ecology I was looking more at works like Symbiotic Machine by Ivan Henriques, the Biodegradable Drone experiment during iGem2014 and the TumbleWeed bot by Shlomo Mir. Later during the development of Machine Wilderness I also discovered the wonderful work by artist Ian Ingram, like Danger, Squirrel Nutkin!

One additional thing I’d like to mention about the project: The title Machine Wilderness is borrowed from the excellent 2012 ISEA in Albuquerque, curated by Andrea Polli, and originates from writings of cultural geographer Ron Horvath in the 1960s. We apply it with full appreciation and respect for that context.

Thanks Theun!

note: This interview was conducted at the start of the Machine Wilderness programme and some of the ideas expressed in it have evolved since.

 WeMakeMoneyNotArt  MachineWilderness