25 February 2022. Re-wilding | Indigenous AI
Remaking the Lake District landscape. What AI can learn from Indigenous protocols.
Welcome to Just Two Things, which I try to publish daily, five days a week. Some links may also appear on my blog from time to time. Links to the main articles are in cross-heads as well as the story. Recent editions are archived and searchable on Wordpress. Have a good weekend.
1: Rewilding the Lake District landscape
There’s a lovely piece on rewilding in the English Lake District by Lee Schofield, who works as a site manager at the RSPB’s Haweswater site. It’s published in the Inkcap Journal newsletter, and is accompanied by Richard Allen’s paintings of the site, including some of before and after the rewilding process.
Haweswater was acquired by the RSPB a decade ago, when they acquired two adjacent farms. It was quite controversial; some influential locals, including the MP for the area, wanted them to continue farming. Schofield isn’t sure that was a realistic possibility:
As visually spectacular as the land we took on at Haweswater was a decade ago, it was damaged in all sorts of ways. Ancient woodland clearances, followed by centuries of grazing, coupled with peatland drainage, a superabundance of deer, river engineering, hedge removal, fertiliser and pesticide use had left the place in tatters.
An overgrazed, over-drained landscape is one in which water moves swiftly. The faster the water flows, the more energy it has, and the greater its capacity for erosion. Sediment reaching Haweswater reservoir, into which our land drains, is bad news for water quality.
—
(Paintings: Richard Allen. (C) RSPB/Richard Allen)
Here’s an example of one of the changes the RSPB has made to improve water catchment. The old drainage ditch—put in to drain off the bogs and improve grazing—had the effect of lowering the water table, exposing peat on the surface and increasing carbon emissions.
They’ve blocked the ditch, raising the water table again, encouraging the growth of the mosses that help peat form, and capturing carbon—turning the area back into a carbon sink again.
They’re also encouraging beavers to help improve the water systems. Readers might link to this to the critique of uplands landscape management that Fred Pearce discussed in the piece I shared earlier this week. He was discussing Europe, but that same is true in many parts of Britain. In short, farming methods had enabled water to run off the hills far more quickly, increasing the intensity of flooding when it happened.
One of the concerns is that rewilding schemes reduce the number of people working on the land, but that hasn’t been the case here. They just work it differently, and create different sources of income:
There are ten full-time staff employed across the two farms now, including me. My colleagues and I do many of the traditional things that the previous farm tenants used to do; we keep the walls up, cut the hay and feed its summery goodness to our cattle and sheep in the winter months.
We do other, new things too. We have wildlife hides and a tree nursery to run and we’re developing plans for a bunkhouse to help people enjoy the distinctive, rugged charm of our corner of the Lakes. We do science and research, surveying and monitoring.
Sitting behind this argument is an observation about the nature of what the countryside does, and how that has changed. Schofield points out that we need the countryside to different things for us now:
What we need now from our countryside, alongside sustainable food production, is carbon sequestration, water quality improvement and flood risk reduction, thriving wildlife and places to unwind and breathe the clean air.
The article is connected to the publication of a book by Lee Schofield, Wild Fell, which is published this week, about the decade of work at Haweswater. It looks lovely.
2: What AI can learn from Indigenous protocols
I was attracted to an article on Aeon by four Aboriginal Indigenous computer scientists on the subject of AI by its title: ‘Blackfella Boxes’. It turned out that the title was a good guide to the piece itself: it fizzes with fun, a bit of self-deprecation, and with gentle provocations about both artificial intelligence and Indigenous culture.
It’s probably the most unlikely article you’ll read on tech this year. And probably the most engaging. They set the tone early on:
As artificial intelligence gets better and better at writing its own algorithms, obscuring the mechanisms of everything from democracy to dating, it might seem like a strange moment to seek dialogue between Silicon Valley and Indigenous Australia. But if you set aside your unconscious bias for a moment, you’ll be surprised to find that the artificial intelligence (AI) community is far more than an army of libertarian engineers led by a handful of billionaire sociopaths.
And then they’re off, into the history of automation and automata, starting with the Golem, wandering through Frankenstein to robots, to Nick Bostrom’s paperclip maximiser thought experiment. (This last one doesn’t end well).
And because the cultural programming that run through the AI programmer community can go a bit askew, and dangerously so, we’re currently obsessing about AI ethics. Which is where our indigenous guides come in:
that’s why we’re here, as Aboriginal thinkers curious about how our own traditional knowledge systems might guide the evolution of AI. If left unchecked, AI poses existential threats to environmental stability, sovereignty, finance and more. Chances are, then, that the inhuman measures imposed on marginalised communities could also find their way to your doorstep one day. You might want to figure out how you’ll deal with your own colonisation when it comes. Trust us when we tell you that, when your entire world is taken away from you, human protocols are what keep you going.
They make an interesting equivalence here between computer protocols (“rules, agreements and procedures that allow data to be packaged up and exchanged between devices”) and Indigenous protocols.
Indigenous science, meanwhile, defines protocols as context-specific behavioural agreements that direct people on how to move, connect, interact and exchange on Country (land) in lawful ways. Protocol balances freedom and obligation, and allows for the interaction of diverse, location-specific Dreamings (beliefs) that guide individuals, clans and families within a broader continental network of custom, trade, technology and ritual connection.
—
(Aboriginal art by unnamed artist. Photo by Alan Levine/flickr, public domain)
Unlike in AI, where the protocols are largely blind to their consequences (see the paperclip maximiser one more time), these Indigenous protocols are designed to optimise wellbeing and abundance, and restrict extraction and competition. Even if they’re quite tongue in cheek about this bit:
Enthusiasm for tech projects based on Indigenous protocols has led to ideas such as putting native food products on a blockchain (‘Hey, let’s call it blackchain!’), or marketing software as if it were built on esoteric Indigenous codes – when, really, it’s just regular software rebranded by a sales team that is ‘Indigenous-led’. Our team at the Indigenous Protocols and Artificial Intelligence (IP//AI) Incubator is made up of killjoys curbing most of this enthusiasm for ancient systems.
In truth, I’m not sure if some of the experiments they describe here to test some of their thinking were real, or just an extended and humorous riff on the way we think about objects, whether smart or not. There’s an amusing attempt to run an indigenous Turing Test, a discussion of Indigenous agent-based modelling software, an exploration of an indigenous marriage protocol, and a long riff on an experiment called “Brother Fridge”, which
highlighted the mismatch of tech designed by a global minority of individualists from nuclear families, but used by a majority of the world’s population who belong to extended family kin groups.
In Indigenous societies, for example, nutrition is bound up closely with kinship relationships:
‘Who would be accountable if the smart fridge shared a man’s yoghurt with his mother-in-law? Would the man be punished, or would the fridge?’
(Fans of the Irish writer Flann O’Brien might note a similarity with his story about whether—owing to the “exchange of mollycules” the postman or the bike should be hanged after committing a murder.)
There’s humour here, not least in the lines from movies that are inserted as commentaries on the rest of text, but there’s also a deadly serious intent. The ‘blackfella box’ is a device that would automatically encrypt material that can not be shared, culturally:
A restricted knowledge protocol like this could be automated as a self-executing ‘smart contract’. Let’s call it ‘proof of Aunty’. Elders specify who has permissions to access restricted sites and information, and store that on the blockchain (a peer-to-peer public ledger that can’t be tampered with, at least in theory). Hook that up to GPS and contact-tracing applications, and you have an automated system for protecting access to sacred sites on Indigenous land.
And they also observe that self-determination, for any community, is going to involve being able to make a claim to data sovereignty. As Indigenous peoples, they know the difference between self-determination and self-administration:
In a world of automated decision-making, this distinction will become important for everybody to discern, not just Indigenous people. Being a self-managing neo-liberal subject isn’t much fun when you find yourself automatically locked out of your smart home for being late with a performance review, or muttering problematic movie quotes in the shower.
But don’t worry. As they say: they ran some Indigenous predictive modelling apps, and everything’s going to be alright. “Kind of”.
——————————-
j2t#269
If you are enjoying Just Two Things, please do send it on to a friend or colleague.