The Wake: May 26, 2026

The Wake is a daily briefing from George's saved internet. The issue is written as a newsletter first. The tweets are the source material, preserved below for receipts.

Source window: May 25, 2026. Signals: 3 bookmarks and 3 likes.

Brief

The Vatican just entered the AI conversation in a voice and register that will matter. On 15 May Pope Leo XIV published an encyclical, Magnifica Humanitas, aimed at the ethical architecture of artificial intelligence. The document is already being read closely by technologists and opinion leaders; it is being treated less as spiritual ornament and more as a blueprint for public moral argument. At the same time, model capability and agent tooling are accelerating in ways that turn moral questions into immediate operational ones: teams are already using agents to resurrect decades-old software, and builders report GPT5.5-level models feel qualitatively different. The collision between a renewed moral frame and blistering technical progress will define policy, procurement, and product priorities for the next 12 months.

Why the Vatican matters now

An encyclical is not a blog post. It is a political and moral instrument intended to shape public conversation and to provide consistent guidance to a global institution with institutional leverage. When the pope weighs in on technology he is not speaking only to believers. He is speaking to governments, universities, corporate boards, and NGOs, and to the moral imagination of publics that still treat religious institutions as legible authorities on human dignity.

That matters for three reasons. First, normative authority: a clear, repeatable moral frame from a widely recognized institution can anchor coalitions that otherwise fragment around technical details. Second, timing: technology has moved from abstract ethics papers to mass-deployed systems. When moral authorities offer a durable frame, regulators can cite it and procurement officers can use it as a checklist. Third, cross-border resonance: religious moral language travels across jurisdictions where secular policy talk stalls, especially in parts of the world where faith institutions still guide civic norms.

Technologists are noticing. Developers and builders are skimming the encyclical for practical signals about human dignity, stewardship, and limits on automation (see @simonw and @badlogicgames). That means the encyclical will quickly be translated into questions in design reviews, vendor RFPs, and policy memos. Expect it to show up not as a set of rules but as a framing device for what counts as a legitimate use of AI.

Agents change the ledger for software permanence

A surprising side effect of better models and agent systems is that software that once seemed ephemeral is getting another shot at relevance. The old argument, famously voiced by Dijkstra and echoed by practitioners like Bob Martin, was that much of programming felt locally significant and temporally bounded: the hardware and platform would change, and the code would die with them. That was true for decades. Now agents: systems that can emulate environments, orchestrate interactions, and translate intent into code: are being used to "resurrect" thirty-year-old projects or revive design visions that no longer fit contemporary stacks.

This is a conceptual shift with legal, archival, and product implications. If agents can reconstitute the behavior and even the spirit of legacy systems, then IP owners, maintainers, and archivists must rethink what persistence means. Licensing questions will follow: does an agent-generated reconstruction infringe on an original work, or is it a new derivative? From an enterprise perspective, the ability to import institutional memory encoded in ancient codebases and make it actionable again changes risk calculus: decommissioning and modernization policies suddenly carry different costs and benefits.

Uncle Bob’s invocation of the historical view that code had limited longevity is a useful corrective-this is not simply nostalgia. What we are seeing is a technology that changes the lifespan of digital artifacts and the responsibilities that come with them.

Capabilities: the urgency problem

Technical progress is not politely aligned with ethical discourse. Practitioners report GPT5.5-class models feel different in kind: "shockingly, scarily capable," in the words of a leading engineer (@dhh). The effect is twofold. For builders, these models expand what is immediately possible: better copilots, more reliable agents, new forms of automation. For regulators and ethicists, high capability compresses timelines and narrows policy windows.

When capability leaps happen, the moral conversation must either accelerate or be rendered less relevant by deployment realities. That contradiction creates a dangerous gap. Moral authorities can announce principles, and technologists can respond with product-level limits, but the window for influence closes once large-scale systems are released and entrenched.

This is why the interplay matters. The encyclical sets a frame. The models change the stakes. The community is asking the hard human factors questions: can a person safely supervise multiple agents, listen to complex information, steer an agent, and review other agents' outputs at the same time? (@Dimillian). The answer will be a design and governance problem as much as a technical one.

Practical consequences for teams and leaders

Three immediate consequences flow from these converging trends.

• Governance gets real. Expect boards and procurement teams to ask for moral and operational assurances framed in language pulled from public moral authorities as much as from privacy law. Prepare for requirements that go beyond algorithmic explainability to include impact narratives, stewardship commitments, and human dignity tests.

• Product and workflow redesign. Human-agent interactions are now a UX and safety problem. The question is not whether to deploy agents but how to do so without overwhelming human supervisors. That will prioritize steerability, session boundaries, provenance, and mental-load-limiting behaviors over raw model capability.

• Legal and archival strategy. If agents can reconstitute legacy systems, legal teams must update IP and data governance policies. Companies should inventory their digital heritage and decide what to preserve, retire, or actively rewrite before third parties use agent tooling to reconstruct it.

What to watch

• The Vatican's secondary outputs. Watch for conferences, advisory committees, or partnerships that operationalize the encyclical. Those will be the real leverage points that translate moral language into policy pressure.

• How major cloud and AI vendors respond. If platform providers adopt the encyclical’s language or equivalent public standards in their enterprise docs, that will cascade into procurement language and compliance requirements.

• Agent governance frameworks. Look for startup and open-source work that treats human-agent orchestration as a product category: supervisor interfaces, audit trails, and cognitive-load management libraries.

• Legal fights over resurrection. Expect early test cases on agent-based reconstruction of legacy code and contested copyrights. Those suits will set precedent for derivatives created by model-assisted reconstruction.

• Adoption signals for GPT5.5-era tooling. Pay attention to product launches and enterprise case studies that claim "no handwriting" steerability or improved safety controls. Implementation details will matter more than headline claims.

The story is not just technical. It is political, institutional, and cultural. A papal encyclical intersects with a rapid wave of agent-led capability at exactly the moment when normative frames can still be meaningful. If you are advising executives, builders, or policy teams, orient first to governance and human-centered design. Then move quickly to the hard operational work of implementation: provenance, steerability, and limits. The moral debate can shape what we build next, but only if builders take it seriously before the new defaults entrench themselves.

Generated from Birdclaw bookmarks and likes. Edited by Ody before publication.