Can we playtest the future?

Twenty slides on civic modelling: living inside our own code, what we love and what we worry about, three stories of models that went right and wrong, the superpowers arriving as fast as the trouble, and the ask. First given at Futureproof AI, Vancouver, 29 October 2026.

@image: /anselm/photos/misc/waskahigan-watchi.jpg @credit: Waskahigan Watchi, Alberta. My photograph. @sub: Futureproof AI · Vancouver · October 2026

Can we playtest the future?

What happens when anybody can build models of the future?

Anselm Hook · hook.org

Notes: Thanks. Canadian, grew up in Alberta, many years in the Bay Area; nice to be home. Tie in to Amber and Brewster.

@layout: dense @media: /anselm/talks/futureproof-2026/img/robots.gif @credit: The robots at work. My video.

Living inside our own code

Earlier this year I found myself in a hotel in Los Angeles, living and working for a few months inside our own code. We were eating, working and sleeping in a real hotel, and we had a software copy of the hotel running alongside it. We had built an orchestration layer, real robots for deliveries, security and cleaning, real guests, real staff. When an elevator got stuck because a robot could not get out of it, it affected me directly.

Our whole world is like that. We live in a landscape built out of rules, policies and asphalt. We are all living inside our own code, and we have to live with the repercussions of our decisions. What if we could playtest those decisions before we make them?

Notes: litahotel.com: built in ninety days, the Anza hotel in Calabasas; Novotel Miami next. The agent orchestration leveraged the Orbital work.

@layout: dense @media: /anselm/talks/futureproof-2026/img/lita-hotel.gif @credit: The Lita hotel twin, live: guests, staff, robots and lifts across three floors.

Real robots, real guests, real staff

Deliveries, security and cleaning, driven by the same orchestration layer as the twin. The software copy of the hotel and the physical hotel ran side by side, and every mismatch between them showed up in the corridor.

@image: /anselm/photos/2023/tubbataha/sunset-reef.jpg @credit: Tubbataha Reef, Sulu Sea, 2023. My photograph. @section: 1 · What we love

What do we love?

What do we care about, what moves us, what do we want to protect?

Notes: Diving, recently, two days' sail from land.

@image: /anselm/photos/2002/argentina/fitz-roy-torres/dsc00281.jpg @credit: Patagonia, 2002. My photograph. @section: 1 · What we love

Our world

As far as we know the thin surface of this Earth is the only place in the universe that has life. Beautiful for its own reasons, for its diversity and complexity, it may also be unique.

Notes: The opening of simulate.world, 2015.

@image: /anselm/talks/futureproof-2026/img/kalimantan-haze-2026.jpg @credit: Kalimantan under smoke, August 2026. Photo PadukaYash, CC BY 4.0, Wikimedia Commons. @section: 1 · What we love

What do we worry about?

Indonesia, this year. By the end of September nearly two million hectares had burned, the haze reached Malaysia, Singapore and Papua New Guinea, and about seventeen hundred people died of the smoke in August alone.

Notes: Figures from the Wikipedia article on the 2026 Indonesian wildfires as of today.

@image: /anselm/talks/futureproof-2026/img/fire-night-2026.jpg @credit: Pulang Pisau, Central Kalimantan, 2026. Wikimedia Commons, CC0. @section: 1 · What we love

Mixed messages

We hear two messages from the media. The first is that the Earth is a beautiful and special place. The second is that we, somehow, are responsible for destroying her.

@image: /anselm/photos/misc/lonquimay.jpg @credit: Araucarias below Lonquimay, Chile. My photograph. @section: 1 · What we love

A planetary therapist

What I want to focus on is not the crisis itself but our response to it. In many of us there is a kind of nihilism, a sense of futility; in others a frantic hysteria. Few of us are unmoved, and almost none of us can act. It is almost as if we need a planetary therapist.

@image: /anselm/talks/futureproof-2026/img/blue-marble.jpg @credit: Blue Marble, NASA. @section: 2 · Superpowers

Surprising superpowers

Just as we are starting to have a real impact on the world, we are also developing surprising powers, totally at odds with our sense of powerlessness. Language models that have read everything. The record of the past, kept. Eyes in orbit on every field and every fire. Simulation that actually works: the weather forecast is the proof. And the hybrid of human intuition with machine speed and memory, which can search a problem space faster than either could alone.

@image: /anselm/photos/2002/argentina/fitz-roy-torres/dsc00312.jpg @credit: Glacier ice, Patagonia, 2002. My photograph. @section: 2 · Superpowers

Hyperobjects

Timothy Morton calls global warming a hyperobject: something that exists on a physical or temporal scale so vast that we cannot comprehend it with an unaided mind. These tools are how we see hyperobjects at all.

@image: /anselm/photos/2002/argentina/pampa-linda/dsc00157.jpg @credit: Pampa Linda, Patagonia, 2002. My photograph. @section: 2 · Superpowers

Can we be much, much more ambitious?

Not just see the present, but run it forward. A SimCity for the real world: playful, maybe crude, but something that shifts the conversation from pure rhetoric to another tool at the table. Tools that let us playtest the code we live inside, that help us see each other and work in community, that connect us rather than divide us.

@image: /anselm/talks/futureproof-2026/img/sparrow.jpg @credit: Eurasian tree sparrow. Photo Jakub Hałun, CC BY 4.0, Wikimedia Commons. @section: 3 · Models, for better and worse

China, 1958

In the late 1950s Chairman Mao somewhat unilaterally decided that sparrows, flies, rats and mosquitoes had to be eliminated. Villagers banged pots, terrorizing the sparrows until they died. But the sparrows had been reducing the pest population, especially locusts, and as a result the locusts ate the grain and tens of millions of people died. A tragic example of knock-on effects, and of what happens when you do not think clearly.

@image: /anselm/talks/futureproof-2026/img/bali-temple.jpg @credit: Pura Ulun Danu Bratan, Bali. Photo Yohanes Nindito Adisuryo, CC BY-SA 4.0, Wikimedia Commons. @section: 3 · Models, for better and worse

Bali, 1980s

Steve Lansing was working in Bali while the Balinese government was trying a new intensive green agriculture. The Green Revolution of the 1970s had brought new intensive farming techniques to communities around the world, and for some reason these practices were failing in Bali. Lansing and James Kremer built a computer simulation of the more traditional practices the Balinese had been using, and showed that the Green Revolution practices were not serving Bali. Their work showed why the coordination embodied in the water-temple networks mattered for irrigation and pest control.

Notes: Priests and Programmers, 1991; Lansing and Kremer, 1993.

@image: /anselm/talks/futureproof-2026/img/wolf-snow.jpg @credit: Wolf in fresh snow, Yellowstone. NPS photo, public domain. @section: 3 · Models, for better and worse

Finland, this month

Google cleared an estimated quarter of a million trees for an AI data centre in northern Finland before the environmental assessment was finished, and more at a second site. The deer left the forest. The wolves followed them into the village. A data centre for models, and no model of the forest. The villagers are living inside someone else's code.

Notes: The Guardian, 22 October 2026. Residents' testimony, not a wildlife study.

@layout: dense @media: /anselm/portfolio/orbital-foundation.jpg @credit: orbital.foundation @section: 4 · Building it

What I have been building

Before this I was at Xerox PARC and a researcher at Mozilla, after a long time in games and mapping. A research grant then paid for a couple of years building Orbital, an open simulation engine and agent orchestration framework. The hotel you saw at the start runs on it, and so does what follows.

Lately I hardly touch code. I have a fleet of agents, and they suggest projects and ask me for money.

@layout: dense @media: /anselm/talks/futureproof-2026/img/bay-layers.gif @credit: Bay Area twin, an explainer rather than a simulation. NOAA, Marin vegetation map, OpenStreetMap. @section: 4 · Building it

How can we build models of the future?

There are several approaches. Steve Coy, at timelike.systems, takes an explicit approach: model a scenario as a series of discrete objects and simulate them forward in time. I am helping with this approach. Ryan Orbuch, at naturalgeneralintelligence.ai, says feed all the data to a world model and implicitly compute near-term futures; this is similar to Dr Fei-Fei Li's work at World Labs, and is also a promising approach worth trying.

Both of these approaches require significant compute. If you have an extra 300 million to spare, please feel free to message.

@layout: dense @media: /anselm/talks/futureproof-2026/img/digitaltwins-wiki.gif @credit: digitaltwins.wiki, October 2026: the home page, Everyday questions, and Vitals. @section: 4 · Building it

digitaltwins.wiki

My current project, with a few friends: not my own solutions this time, but an ongoing live survey of the digital twin space. One focus is to collect tools, best practices and thinking around building macro-scale whole-systems simulations. A digital twin turns out to be a rich stack of parts: raw data; ontologies, how we turn raw data into shared understanding; law and policy, the invisible landscape that shapes outcomes; engines that model and simulate; presentation; and above that communication and participation.

Good data is fundamental, and the success of any digital twin we build will depend on it. Contributors are very welcome.

@image: /anselm/photos/2002/argentina/fitz-roy-torres/dsc00132.jpg @credit: Fitz Roy, Patagonia, 2002. My photograph.

Can we recruit eight billion minds?

Governments and corporations already have digital twins and models. The question is how we can have citizen tools that let anybody play, try an idea, share it, explore it together. Humans know their places well; they have intuition about their homes. Machines can complement this by bringing brute force computation to model outcomes. How can we find ways to rehearse, or play with, the future together? The only way to build a collective future is for all the stakeholders to take part in writing the code we live inside.

@image: /anselm/talks/futureproof-2026/img/picard.jpg @credit: Patrick Stewart as Jean-Luc Picard. Star Trek: The Next Generation, Paramount. @sub: For Patrick Stewart, 1940–2026

Let's be bolder

Let's use our emerging superpowers to tackle the hard problems, with compassion, to have the kind of future we imagine is possible. We are living inside our own code. Let's write it together.

@image: /anselm/photos/misc/boxfish.jpg @credit: Boxfish. My photograph.

Questions

Help curate the models: digitaltwins.wiki

The long version: anselm.substack.com

Build one with us: into-the.blue

hook.org · simulate.world · @anselm · anselm@gmail.com

@layout: dense @media: /anselm/portfolio/colorado.jpg @credit: colorado.exe.xyz: who drinks the river, as an explorable explanation. @appendix: true

Appendix · Computational journalism

Explorable explanations, Bret Victor's phrase, and computational journalism are an attempt to combine the topics people already care about with participatory models: the Colorado River's water rights at colorado.exe.xyz, the Strait of Hormuz at hormuz.exe.xyz, futuresdesk.ai.

ontology.monster is aggregating news about the world and scavenging the concepts, people, places and things out of it, as fuel for building a simulation of the world.