Early Draft 2026-06-25

Summary

Draft PDF of Towards Superintelligence Alignment (subtitle: Boundaries, Values, and Correction). Research manuscript in progress—not a claim that alignment is solved. Many chapters remain stubs; see Current Status in the PDF.

What changed since Early Draft 2026-06-20

Frontmatter

Split Preface (why / who / how to read / authorship), Executive Overview (skim layer), Introduction (five claims + roadmap), and Current Status (WIP + chapter map).
Acknowledgements added and placed early in frontmatter.
Introduction no longer front-runs Part I: standard picture, “which alignment?”, and the civilizational loop diagram moved to Chapters 1–2.
Former “where the argument is shaky” bullets distributed into chapter What Would Change This View sections.

Manuscript & appendices

Generated Appendix A (notation index) and Appendix E (assumptions ledger); operational glossary expanded.
Global bibliography; INSTRUCTIONS slimmed for agents/contributors.
Appendix I (Lean proof spine): dependency diagrams refined; definitions linked to chapter anchors.
Adversarial WWCTV falsifiers added book-wide; notation and ledgers refreshed.

Chapters

Author drafts integrated for Chapters (successors, selection, safety-case arc, civilizational limit).
Ch. 25: correction grounded in handle control; CCI consolidated.
Ch. 32: fitness model aligned with Lean handles.
Attributed epigraphs across chapters; part/chapter titles simplified; heading style normalized.

Formal spine (formal/)

Lean proof spine synced with manuscript notation; UAD bridge papers imported; proof dependency graphs updated.
Build
Source: main @ HEAD
PDF built with ./build.sh → dist/pdf/towards-superintelligence-alignment.pdf

Still open

Ch. 5 partial; Ch. 39–40 largely stub.