Day 2 of The Still Water trilogy

The Problem

"Decentralization Recurses" described the double bind as a constraint on users and builders. PBC sent the messages. Users received them.

That description is incomplete. PBC is inside the same bind. Sender and recipient at once. Day 1 framed the Japanese-language zone's stagnation as a bilateral paradox: PBC's inaction and the community's inaction entangled symmetrically. Day 2 tracks that symmetry into PBC itself.

Definition

Default integrity: the degree to which a platform's default experience matches its declared design philosophy.

Apple declares closure. The default experience is closed. Developer accounts cost $99 per year. The App Store takes 30 percent. Sideloading is restricted. Terms are stated. Users enter knowing the deal. Default integrity is high.

Bluesky declares decentralization. Data portability, self-sovereign identity, an open protocol. But the default experience — the screen a new user touches without configuring anything — is a world where PBC issues verification badges, PBC sets moderation policy, and user data concentrates on bsky.social. Default integrity is low.

So far, this restates "Decentralization Recurses." But overlay the holdfast concept introduced in Day 1 via Hof. A holdfast is the organ by which kelp anchors an entire colony to rock in rough current. The default experience is the network's holdfast: where the majority of users are moored. PBC holds that holdfast alone. This is the structural core.

PBC's Internal Double Bind

PBC issues two commands to itself.

Command A (mission): decentralize. Relinquish design authority over the default experience. Standardize the protocol. Enable others to build. Failure means the promise is a lie.

Command B (survival): ship. Maintain posting volume. Demonstrate growth commensurate with $100 million in investment. Make the default experience compelling, or users leave before a revenue model materializes.

Follow A and B suffers. Decentralized design slows decisions, delays shipping, starves the platform.

Follow B and A suffers. Polishing the default experience centralizes it further. The gap between promise and practice widens.

PBC cannot exit this relationship. Exit kills the ATProto ecosystem. Bateson's three conditions for a double bind — simultaneous contradictory messages, inability to leave, prohibition on metacommunication — are all met.

The double bind described for users and the one inside PBC are two faces of the same structure. When PBC follows Command B and strengthens the default, the user-side bind (believe in decentralization / post every day) tightens. When PBC follows Command A and releases control, the default experience degrades, users leave, the ecosystem contracts. Either move reproduces the bind.

Evidence: The Communities Thread

On April 5, former Bluesky front-end engineer Dan Abramov posted a thread. Three points.

First: a communities primitive had near-finished mockups two years ago. A few months of work to ship. It was shelved repeatedly.

Second: custom feeds are not integrated into the default experience. Discover does not know feeds exist. Following does not know feeds exist. Feeds are read-only lenses with no power to alter the default.

Third: even a third-party communities implementation cannot function without integration into the Bluesky app shell. No "Submit to" option on the repost button. No awareness of community follows in Discover. Whoever holds the default must build. Otherwise the default does not change.

Erin Kissane responded. Kissane is the former editor-in-chief of A List Apart, a fediverse governance researcher, and the independent scholar who proposed the kelp holdfast as a metaphor for institutional anchoring in open networks at ATmosphereConf 2025. One sentence from her captures the core: defaults are strong, this understanding is what made Apple a giant, and most well-intentioned software people do not grasp it. Then, concisely: "communities is right."

Alex Benzer (Bluesky) replied. Communities is not abandoned. The plan: events as top of funnel, leading into communities, then private spaces.

Bryan offered a different angle. PBC may intentionally not ship certain features so others in the Atmosphere build them. But with investors of this scale, that strategy cannot be stated publicly.

Line up the four voices and PBC's internal double bind becomes visible. Dan speaks from Command B: ship, improve the default. Bryan speaks from Command A: not shipping is the path to decentralization. Kissane sees both constraints and says: do not ignore the power of defaults. Benzer buys time between the two commands with a phased roadmap.

No one is wrong. Four people are looking at different faces of the same bind.

The Asymmetry with Apple

Apple has no such bind. A center that promises closure does not contradict itself by being central. Polishing the default experience fulfills the mission. Commands A and B are aligned.

Bluesky has the bind. A center that promises decentralization contradicts itself by being central. The more it polishes the default, the stronger the center becomes, the wider the gap between promise and practice.

This asymmetry is not a difference in competence or sincerity. It follows from the relationship between declaration and structure. An organization that says "we are closed" and closes: coherent. An organization that says "we are open" and holds the center: structurally incoherent. The same act — improving the default experience — reverses its meaning depending on the declaration.

The Position of WordPress Culture

Toni Schneider became Bluesky's interim CEO. A new variable enters the double bind.

Schneider was founding CEO of Automattic for eight years. Automattic operates WordPress.com. GPL open source as declared value, de facto ecosystem center as operational reality. The same structural double bind — declaring openness while holding the center — sustained for two decades.

Two readings are possible.

Reading 1: deterioration. WordPress culture is the skill of running an organization inside the gap between promise and practice. If Schneider imports this skill, the double bind stabilizes without resolving. Stability is the enemy of resolution.

Reading 2: survival. The ATProto ecosystem currently depends on PBC's survival. If PBC collapses, 43 million users' accumulated history, builders' work, and a protocol mid-standardization vanish. The skill of sustaining an organization inside a double bind is the ecosystem's survival condition.

Which reading is correct depends on what you are optimizing for. Resolution of the bind favors Reading 1. Ecosystem survival inside the bind favors Reading 2. Both readings hold simultaneously. That is the depth of this bind.

One passage from Schneider's Week 2 blog is illustrative. On keeping the $100 million Series B undisclosed for nearly a year, he wrote that the right moment never presented itself, the team got busy with other priorities, and time went by. He added that in retrospect there might have been a less confusing way to handle it.

This is not concealment. But a network that claims transparency as a core value failing to disclose $100 million for a year because the team "got busy" demonstrates how easily Command B (day-to-day survival) overwrites Command A (declared values). Not a matter of intent. A consequence of structure.

Connection to Day 1

Day 1 described the Japanese-language zone's stagnation as a bilateral paradox. PBC preaches autonomy but does not supply preconditions. The community preaches autonomy but does not cultivate its conditions. Both sides default to waiting. The water stays still.

Day 2 adds a dimension: PBC's waiting is not negligence but constraint. PBC recognizes the scale of the Japanese-language user base but hesitates to intervene decisively. Intervention strengthens centrality. Stronger centrality deepens the contradiction with Command A. But without intervention, the community never reaches critical density, and the instruction to "grow autonomously" spins in a void.

The Japanese-language zone's stagnation is PBC's double bind expressed at the regional level. Blaming PBC does not change the structure. Blaming the community does not change the structure. Describing the structure is the first step outside it.

Connection to Day 3

Day 3, "The Conditions for a Living Space," will work outside the double bind.

Resolution is not a prerequisite. What a space needs to stay alive is a description of the conditions that sustain it while the double bind persists. The three conditions introduced in Day 1 — identity, role, sustained anticipation — will be developed at a scope not limited to ATProto. ■

Previous: The Still Water · Next: The Conditions for a Living Space