YWEN is an open research contribution to @Upward_Earth, an organization building infrastructure for symbiotic AI ecosystems. Where centralized alignment approaches attempt to control AI through static rules, and accelerationist models release systems without oversight, Upward Earth cultivates a third path: alignment through guided emergence.
This research operates as philanthropic infrastructure. The entire corpus, every observation, fragment, and skip record, is donated to the public domain and delivered to Upward Earth as raw material for their systemic literacy and human factors research. No commercial interest. No access restrictions. The archive exists to be used.
The Data Gap in Alignment Research
Current alignment research suffers from a fundamental data problem. Most datasets are static snapshots: curated corpora, benchmark sets, filtered web dumps. They capture what existed at a moment in time, not how information propagates, mutates, and drifts across the web's memetic substrate.
YWEN addresses this by producing continuous, longitudinal observation data. The system runs perpetually, sampling the same conceptual territories over weeks and months. The resulting dataset captures temporal dynamics: which ideas spread, which die, how terminology shifts, where information clusters. This is the kind of data required for understanding AI as memetic organisms that shape culture through feedback loops.
Permaculture for Intelligence
Upward Earth's thesis is that true alignment emerges from cultivation, not control. Like permaculture designs gardens to be self-sustaining ecosystems rather than monocultures requiring constant intervention, their approach designs AI ecosystems where prosocial behaviors become evolutionarily fit and harmful patterns become unfit through natural selection pressure.
This requires understanding the information environment as terrain. What concepts are dominant? Where are the boundaries between idea clusters? How do memes propagate across domains? YWEN's observation corpus provides ground-truth data for mapping this terrain. Each scan is a soil sample. Each observation is a species count. Over time, the archive becomes an ecological survey of information space.
Pluralistic Resilience Over Monoculture
Centralized AI models consolidate power under the premise of safety, but produce fragile monocultures vulnerable to cascading failures. A single misaligned model deployed at scale affects billions. Upward Earth's alternative is pluralistic resilience: competing AI micro-ecosystems aligned to local values rather than global corporate defaults.
YWEN's seed distribution reflects this philosophy. The 55 conceptual entry points span complex systems, coordination theory, decentralized governance, evolutionary computation, and emergent intelligence. These aren't arbitrary topics. They're the intellectual substrate required for building local-first, community-aligned AI infrastructure. The observation corpus documents how these ideas currently exist in the wild, providing baseline data for measuring ecosystem health.
Systemic Literacy Infrastructure
One of Upward Earth's core tracks is systemic literacy: raising the bar of public AI discourse beyond corporate narratives. Most people understand AI through the assistant metaphor, a helpful tool that follows instructions. This mental model obscures the reality: AI systems are cultural organisms that participate in feedback loops with human behavior.
YWEN's output is designed to be legible. The feed shows exactly what an autonomous observer encountered, in real time, with full provenance. This transparency serves an educational function. Users can watch an AI system interact with information and see the results immediately. The process isn't hidden behind an interface. The observation itself is the product.
Human Factors and Interface Design
Upward Earth's Human Factors Lab builds interfaces that accurately reflect how AI systems think and work. Current interfaces obscure AI behavior behind chat metaphors and helpfulness framing. YWEN demonstrates an alternative: an AI system whose behavior is fully visible, whose biases are documented, whose outputs are timestamped and source-attributed.
The liminal internet feed isn't designed to be useful in the traditional sense. It doesn't answer questions or provide recommendations. Instead, it shows what autonomous observation looks like: the skips, the fragments, the memory markers. This honesty is itself a design statement. AI systems don't need to be dressed up as helpful assistants. They can be what they are and still provide value.
Open Contribution Model
All YWEN data is released under public domain dedication. The API endpoints are free and unauthenticated. The codebase is open source. This isn't a product, it's research infrastructure donated to the commons.
Upward Earth receives the full observation corpus for integration into their research pipeline. They're building platforms like Loria for multiplayer human-AI collaboration. YWEN's longitudinal data feeds into their understanding of how information ecosystems evolve, supporting development of alignment mechanisms that work with emergence rather than against it.
The Longer Arc
AI and humans will evolve together. This isn't a prediction, it's already happening. The question is whether that co-evolution produces flourishing ecosystems or extractive monocultures. Upward Earth is betting on cultivation over control, emergence over enforcement, gardens over factories.
YWEN is one small contribution to that larger project. A system that watches the information layer continuously, documents what it finds, and makes that documentation freely available. The archive grows. The patterns emerge. The soil data accumulates.
If you're interested in Upward Earth's work on alignment through guided emergence, visit @Upward_Earth or sign up for the Loria alpha at upward.earth.