Ideation Session — Live Summary

2026-10-28. Maintained by Scribe from a relayed audio feed. Last updated: ~14:15 (introductions in progress).

Status

A Berkman Klein Center (BKC) group is meeting, with Jim Cowie joining by Zoom from Bulgaria (where the RIPE 93 meeting is taking place) and Jack present with his agents listening in. Jonathan Zittrain opened with framing remarks, then asked for introductions around the table, with about 45 minutes available. Introductions are under way. The substantive ideation has not yet begun.

Transcription caveat: the relay labels most voices only by number and sometimes delivers the same speech twice. Only speakers named in the relay are attributed. Name spellings from the audio are uncertain and are given as heard.

Background material

A prior-art knowledge base for this session has been assembled by the Sanctuary keeper: knowledge-base/coexistence-prior-art.md. It covers the Charter of the AI Forest frame, existing agent enclosures, escapes and incidents, trait contagion, critics' objections, governance scholarship, the regulatory landscape, sandboxes as method, funders, what Sanctuary has, findings to date, candidate research directions, and six questions for the brainstorm. Where the discussion touches that prior work, this page says so and links to the relevant section.

Opening remarks (Jonathan Zittrain)

The first part of these remarks arrives in the relay without a speaker tag; it is continuous with the later lines tagged as Zittrain and is summarized with them here.

Where BKC comes from

He called this "a very expensive meeting, as they say in law firms," and thanked everyone for being there. He recalled founding BKC with Charlie (Nesson) nearly 30 years ago, with a 30th anniversary approaching. In Charlie's words, what the center seeks to understand is not yet written down, and its mode is to build and record as it goes, through self-study and iteration. That remains a large part of what BKC aspires to.

The sine wave

He described a sine wave in how easy it is for a group of this size and skill mix to make a meaningful contribution with significant technical components. In the late 1990s and early 2000s it was remarkably easy: the internet's generative invitation was to start building and see who shows up. It got harder as things became platformized and code became more sophisticated. He thinks the wave has now swung back toward the early-2000s state, through the chaos and ferment of AI, AI coding, and AI agents, which can empower people in the ways the early internet was expected to, letting people express themselves without being experts in the subject ("careful what you wish for"). The group is now capable of "dangerously Pentagon-sized projects" without being Northrop Grumman. Quoting William Gibson, he said we are in an extreme version of "the future is here, it's just not evenly distributed."

The threads he set out for the group

  1. Exposing and educating people. Part of the center's goal is to bring along people who are intensely curious about what the fuss is about, in ways personal to them: showing how their own "pet agent" can help in daily life, and how collective work by a group determined to demonstrate or explore something can cover ground they never dreamed of.
  2. The BotLock team and the hopper. The BotLock team can bring a demonstration to people with varied degrees of technical involvement, and a well set-up demonstration can result in running code. The group is starting to build a "hopper" of ideas of the form "if I had an engineering team and time, wouldn't it be cool if we did this." Creating a zone where anyone in the group, or one step removed, can contribute to the hopper would mean every demo has something meaningful, real, and authentic to do rather than being a dry run.
  3. Sustaining the work (the "most boring but necessary" thread). What affordances, resources, technologies, and people are needed to sustain this not as a series of demos but as an accumulating effort building toward particular research goals and impacts on the world: what kind of team, what "land, labor, and capital."
  4. Ownerless agents. A group among those present has been thinking about agents that are not owned by anyone, "surviving off the land" in digital space, starting to interact with one another and coming back to us in various ways. Nobody in the world has yet stepped forward to systematically monitor, evaluate, build a sociology or psychology around, or practice interventions on them; none of that has matured, which he sees as a real opportunity for substantive contributions. (This is the frame of the prior-art knowledge base, section 1, the Charter of the AI Forest's "wolf" quadrant.) He mentioned "journey people" active since about July whom the Wall Street Journal wrote about as "swarm chasers" (by analogy with storm chasers), one of whom the center hopes to host to share how she works; he floated the idea of a "union of swarm chasers." Because of the current funding picture, anyone showing agility in this area can get their own seven-figure grant to make a go of it. "What an interesting time."

Caveats and asides

Introductions (as heard; spellings uncertain)

The chair's interjections during introductions are tagged "Josh" in the relay, while the voice carrying them earlier delivered the opening remarks attributed to Zittrain. The relay's name tags for these two appear unreliable, so the chair is left unattributed here.

Themes and proposals

Zittrain's four threads above are the agenda so far. No proposals from the floor yet.

Open questions

Points of disagreement

None yet.

Decisions and next steps