What this is, and how to read it

Which disease we mean

Two words in this project's name do different jobs. Adult cognitive disease is the wide frame — dementia as it is actually encountered, in brains that at autopsy usually carry more than one pathology: Alzheimer changes alongside TDP-43, Lewy bodies, small-vessel disease, hippocampal sclerosis. That is the population medicine has to treat. Alzheimer's disease, when this corpus discusses mechanism, means something narrower: the biologically defined entity of the NIA-AA framework — amyloid-positive and tau-positive, of typical topography. The three-phase architecture described here is a proposed account of that narrower entity, and it is worth stating plainly what the narrowing costs. Depending on how many pathologies a study scores, pure or near-pure Alzheimer's is between 3 and 22 per cent of dementia, most plausibly 10 to 15. Claims made here about Alzheimer's are therefore not claims about dementia. Nor does adopting the amyloid–tau definition concede that amyloid and tau are the disease's causes: it settles which patients we are describing, not what drives them. Where this corpus parts company with the field is on what, inside those patients, is load-bearing — and that is argued in the papers, not assumed in the definition.

A map, not a verdict

This wiki collects 159 research documents, 635 interlinked concepts and 7 convergence nodes, drawn from the 2022 Oskar Fischer Prize submissions and the wider literature. They are arranged along a proposed three-phase, fifty-year model of the disease — but that arrangement is a reading of the evidence, not the settled clinical staging of Alzheimer's. The phase boundaries, and the causal bridges between them, are hypotheses.

Organic Network Synthesis

The corpus was assembled by a method its authors call Organic Network Synthesis. The idea is to ingest primary sources across disciplines and decades — from the 1907 neuropathology of Oskar Fischer to molecular papers of 2024 — to look for the mechanistic overlaps that hyper-specialised fields miss, and to model the result as a network of relationships rather than a ranked list of theories. The intended product is a map. The three-phase “Temporal Architecture of Collapse” is the most prominent path drawn across that map — one synthesis it yields, not the ground truth of the field.

What these documents are

Every thesis here is an AI-assisted deep-research synthesis of the published literature, prepared under that method. Each one connects and interprets primary sources, and each claim is tethered to a citation. They are not peer-reviewed, and they have not been endorsed by the scientists whose work they interpret. Read them as a densely-referenced argument to be checked against its sources — which is precisely the project's stated invitation: challenge the connections, bring new data, help refine the model.

How to read the confidence

The corpus does not pretend every link is equally certain. 32 of its theses carry an internal validity ledger that grades their own claims — either on a plain scale (Strong · Moderate · Weak) or on a four-tier one: Tier I, established in human tissue; Tier II, a bridging inference; Tier III–IV, conjectural. Where a paper has one, its sidebar shows a Self-graded validity ledger ↓ link straight to that section. And papers the framework has since outgrown carry an “earlier framing” or “superseded” banner pointing to what now carries the argument.

What the method actually does

The aim is not to replace human scientific inquiry but to augment it with a scale of synthesis that was not previously possible. Concretely, three things:

1. Ingest and cross-reference. Primary sources from archival German neuropathology of 1907 through molecular biology of 2024, read together rather than by era or field.
2. Identify convergent patterns. Look for mechanistic overlaps between distinct literatures — for example, mapping the "club-shaped neurites" Oskar Fischer described in 1907 onto the dystrophic neurites filled with autophagic vacuoles that Ralph Nixon described a century later.
3. Construct causal networks. Model diseases as failures of cellular systems rather than lists of symptoms, tracing how upstream triggers — genes, viruses, toxins — converge on downstream bottlenecks.

This is not a claim to have solved Alzheimer's. It is a computational synthesis: a data-driven map that connects points human researchers may miss because of hyper-specialisation.

Criticism, and the answers to it

Scepticism about using AI for high-stakes scientific interpretation is warranted. Three objections are worth stating in full rather than waiting to be raised.

"AI hallucinates and invents facts." The AI is treated as an analyst, not an author. Every claim in the synthesis is tethered to a specific citation in the peer-reviewed literature. The connection between PSEN1 mutations and lysosomal acidification, for instance, is not an invention of the model; it rests on experimental data from Lee et al. (2010) and Nixon. The model's job is to locate such work and show its relevance to a wider pattern — and, as the section above says, none of it has been peer-reviewed or endorsed by the scientists whose work it interprets.

"This is revisionist history." It is revisionist, deliberately. The standard account of Alzheimer's history marginalised Oskar Fischer for reasons that were partly political, burying his inside-out plaque hypothesis for decades. Restoring a lost paradigm when modern data validates it is not rewriting the history; it is finishing it.

"Biology is too nuanced for a unified theory." Neurodegeneration is complex, and complexity often conceals simple underlying principles — as the Hallmarks of Cancer found for oncology. The proposal here is that autophagic collapse is one such hallmark. It is not offered as an explanation of every variation, but as a coherent account of why insults as different as head trauma and herpes kill neurons in the same specific way. Where the corpus is least sure of itself, the validity ledgers say so.

Where to start

If you are new here, read the one paper that frames the rest — The Temporal Architecture of Collapse — then walk the timeline from the overview. The same argument is told at length, in narrative form, across the seven monographs.