Sensemaking Commons

Late one Sunday in August, I was reading Dave Winer. He had been watching The Pitt and liking it for a simple reason: the show behaves like a simulated community. You come to know the characters. You keep thinking about them when the episode ends.

Then he looked a little further ahead. Before long, he said, AI will make shows like that behave more like software. New episodes of The Wire or The Sopranos or The Pitt on demand, assembled around familiar characters the way a model assembles fluent text. The catch he named in one line: we would be missing the fun of discussing the latest episode with friends. Or, he added, we might invent bot friends who pretend they watched the same private episode we just generated.

The bot-friend line stuck because it solved the wrong problem.

That second option is the tell. The imagined response is revealing: when the shared episode disappears, we invent a simulated commons rather than ask what happened to the real one.

When the episode is only yours

Conversation needs overlap. Not identical taste. Overlap. A scene both people saw. A memo both people read. An incident both people can point to without reconstructing it from scratch. When every feed is private and infinite, the default is no longer a common week. The default is parallel weeks that only look adjacent in a meeting invite.

You do not need a monoculture to have a commons. You need enough held-in-common material that disagreement still has a place to stand.

I call that shared layer a Sensemaking Commons: the inputs a group needs in order to think together. Shows, memos, incidents, documents. Whatever people can actually point to. A commons does not require shared conclusions. It requires shared referents.

The overlap you can build on purpose

Personalization is usually sold as care: more of what you want, less of what you do not. At the level of a single viewer, that can be true. Personalized systems do not necessarily eliminate shared inputs. They make common input non-default. The coordination cost rises because nobody can assume that anyone else saw what they saw.

That is where this sits in the Digestion Gap: plenty can enter each person’s stream; less becomes something a group can point to together.

Winer was imagining what happens when an accidental commons of pop culture comes apart. I started looking for how a group might manufacture some of that overlap on purpose. Indi Young’s work on recurring characters, episodes, and thinking styles gave me a clue. In her June 2026 piece she describes a structure for collaborative ideation, closer to a television series with a recurring cast than to an open brainstorm. The characters are grounded in researched archetypes, then placed in different scenarios so a team can test ideas without demanding that anyone change how they think.

Sensemaking Commons is my name for the organizational version of that move, not hers.

A product team can put a familiar customer character through a hard tradeoff (cut the feature, raise the price, ship the workaround) and argue about the episode, not about whose judgment is defective. An executive team can do the same with a named incident: the same sequence of events, replayed as a shared scene, so dissent lands on what happened rather than on personality.

The method still has to stay grounded. A generated approximation of “the customer” or “the incident” is not the same as researched cognition and a packet people can reopen. If the shared object is only remix, disagreement has nowhere solid to stand.

Disagreement with an address

Five executives reading five personalized summaries have plenty of information. They may even receive accurate information. What they lack is a shared evidentiary surface. Nobody knows exactly what the other four encountered, omitted, emphasized, or inferred.

A shared source changes the nature of disagreement. Someone can say, “Go back to paragraph seven.” Now disagreement has an address.

A functioning commons seems to need three things that travel together: shared referents, visible provenance, and social permission to interpret them differently. Shared referents create common ground. Provenance makes that ground inspectable. The Merge Tax named the inverse problem: a clean story without a trail stops being checkable. Interpretive permission prevents the commons from becoming enforced consensus.

The practical questions are few. Do your meetings start from a common artifact, or from five fluent private summaries of it? When two leaders disagree, can they argue over the same paragraph? If the AI brief disappeared tomorrow, would a shared packet still exist?

Once supply is infinite and private, shared inputs become a design choice.

None of this argues against personalization. It argues against treating the commons as a free byproduct of more content. Keep a few inputs shared on purpose. Prefer one joint brief over five fluent summaries nobody else saw. Leave room for a shared episode, literal or metaphorical, that people can discuss without inventing a bot friend to fake the overlap.

What still needs to be experienced together?

Madam I’m Adam