Research

The First Thing I Did With A Computer

The intellectual lineage begins with Sanders' childhood encounter with Apple Logo. The turtle made computation visible, inspectable, and changeable. Programming quickly became a social practice as students began exchanging commands, techniques, and ideas.

That early experience remains conceptually important to TurtleBlock AI: the computer is most powerful when the learner can think with it, make with it, inspect what happened, and revise.

From inquiry to a unified computer learning theory

Sanders' doctoral research at Loyola Marymount University began in 2016 and culminated in 2019 with the completion, defense, graduation, and publication of Toward a Unified Computer Learning Theory: Critical Techno Constructivism.

The research asked why educational technology so often reproduces predetermined curriculum and isolated computer use rather than supporting inquiry, collaboration, critical thinking, learner voice, and construction. Across this period, STEAMHAMLET, virtual learning environments, mixed reality, and new forms of learner-centered computing served as both conceptual provocations and design contexts.

Dedoose document analysis

The dissertation used document analysis to examine selected passages from John Dewey's Democracy and Education, Paulo Freire's Pedagogy of the Oppressed, and Seymour Papert's Mindstorms. The selected excerpts were imported into Dedoose and analyzed with a literature-derived coding system.

The dissertation preserves the relationship between theoretical precepts and excerpting codes, the frequency of codes across the three seminal works, and significant code co-occurrences. That analytical structure now provides the foundational source layer for the Dr. Bryan P. Sanders TurtleBlock AI Research Ontology.

These seventeen labels are preserved exactly as authored in the original research. Later TurtleBlock AI concepts are connected through explicit, versioned relationships rather than by renaming or rewriting the dissertation taxonomy.

Physical books become a persistent source layer

The TurtleBlock AI Research Library preserves the books and physical artifacts that helped shape the research program. A book can eventually leave the shelf while its bibliographic record, photographed evidence, intellectual connections, contents, provenance, and cataloging notes remain available to the research system.

The public catalog is read directly from D1. Source facts derived from photographed copies remain distinguishable from later interpretation, user clarification, or external verification. The collection therefore operates as research infrastructure rather than as a decorative reading list.

Tenets, questions, and actions

Critical Techno Constructivism was subsequently operationalized as seven pedagogical tenets: Personal Inquiry; Compelling Problem or Question; Technology as Tool to Think With; Formative Demonstration of Learning; Reflection as Learning; Social and Cultural Critique; and Sharing and Collaborating.

Each tenet is paired with a diagnostic question and practical guidance. In the research ontology, these remain distinct from the Dedoose codes: one layer records the original analytical taxonomy; another records an authored framework for educational practice.

From teachers-mostly-talking to students-mostly-constructing

The argument moved toward high-ceiling, open-ended computational environments with multiple entry points, collaborative workspaces, physical computing, iteration, and student-generated inquiry.

Could Minecraft Be a School?

Published in Springer's Game-based Learning Across the Disciplines, the chapter considers Minecraft not simply as instructional software but as a persistent learning environment in which inquiry, collaboration, building, conflict, and discovery can generate curriculum.

Purposeful Play – Educating with Minecraft

Published by Minecraft Education, this work describes inquiry emerging from collaborative play and positions the teacher inside the learning environment as a participant in dialogue rather than only as the source of predetermined tasks.

Vignette: STEAMHAMLET Is School 2051

STEAMHAMLET extends the research into an imagined multidisciplinary learning environment where information can become visible, editable, shared, and spatial. The environment is conversational and adaptive, while the learner remains participant, creator, and interpreter.

This line of work anticipates a central problem now addressed through WorldSpec: how to make ideas visible and revisable without reducing them to the machine's first interpretation.

A Radical Shift for the Future, Today

Published in California English, the argument turns from preventing access to machine-generated responses toward developing the habits of mind required to question, evaluate, revise, and collaborate with them.

Recursive agent design

As custom GPTs became available, Sanders moved quickly from using conversational models to designing specialized conversational systems. One early experiment was deliberately recursive: a GPT designed to help build better GPTs.

The work soon moved beyond prompt-writing. The design problem became how to represent the how of an effective GPT: instructions, source material, categories, examples, constraints, relationships, and recurring patterns of use. That led to experiments in representing agent design itself as structured data—tables, rows, fields, and relationships that could be inspected, compared, reused, and revised.

This is an important precursor to WorldSpec and the current research ontology. The question was no longer only how to phrase a prompt, but how to construct an environment in which a useful conversation could reliably continue.

From document piles to structured evidence

That agent work expanded into systems for working with large collections of public documents and government records. Meeting materials, correspondence, reports, timelines, people, organizations, places, financial relationships, and source documents were progressively ingested, normalized, linked, and made retrievable.

The model was not expected to remember or infer an entire corpus from scratch. Increasingly, the work focused on building the structured environment around the model: canonical records, entities, relationships, timelines, source layers, provenance, and explicit distinctions between evidence and interpretation.

The continuity with earlier classroom database work is direct. Meaning is made structured enough for a computer to work with while the structure remains visibly different from the meaning itself.

From prompting to participation

Sunshine Machine emerged as a model for persistent computing environments built around intentionally organized corpora: ingest, tag, sort, relate, retrieve, question, revise. The conceptual move is away from repeatedly asking a blank machine the perfect question and toward constructing an environment that already contains the relevant history, structure, provenance, and relationships.

The research increasingly moved from designing prompts to designing environments. TurtleBlock AI inherits that move: the conversation matters, but so do the persistent structures that allow the conversation to remember, retrieve, construct, and recursively change.

GPT and Me, An Honest Reevaluation: The Dawn of Co-active Emergence

Published in Impacting Education, co-active emergence describes purposeful human–machine dialogue as an intellectual process in which neither participant is reduced to a simple instrument of the other.

The writing and theory developed recursively through sustained human–machine dialogue: an idea generated a response; the response became material for evaluation; evaluation produced a new question or revision; and the next exchange changed the conditions for what could emerge afterward.

Co-active emergence therefore became both a theoretical proposition and a description of a method being experienced in practice.