deep-synthesis-why — Workgraph

Metadata

Status	done
Assigned	`agent-46`
Agent identity	`3184716484e6f0ea08bb13539daf07686ee79d440505f1fdf2de0357707034c3`
Created	2026-05-01T19:41:57.689168137+00:00
Started	2026-05-01T19:43:19.845628647+00:00
Completed	2026-05-01T19:47:17.834712760+00:00
Tags	`grant,urgent,synthesis,v2-pivot`, `eval-scheduled`
Eval score	0.89
└ blocking impact	0.95
└ completeness	0.95
└ constraint fidelity	0.55
└ coordination overhead	0.90
└ correctness	0.92
└ downstream usability	0.85
└ efficiency	0.90
└ intent fidelity	0.57
└ style adherence	0.95

Description

The Google.org submission is hours away. Erik is asking, honestly, whether the v2 scientific framing (PHR discovery in pangenomes, applied to undiagnosed rare disease) is the science to do versus some science Poietic PBC happens to be positioned to do. He's worried it might be a stitched-together justification rather than a coherent spine. He's also articulating that the company's actual reason for existing is making human-machine collaboration legible to its participants, which is the PBC benefit statement filed in Delaware. The science is the proving ground for the infrastructure, not the other way around.

What's missing as a single artifact: a synthesis that puts the mission at the spine and explains why pangenomic PHR work is the right proving ground for it.

This task does NOT edit the application. It produces a short standalone doc. Erik will read it before hitting submit, and a separate paused task ('sync-application-to-synthesis') will fold any sharpening back into the application after Erik approves.

Project convention: notes for Erik go to ~/poietic.life/notes/.

What to read

workgraph_google_application_FINAL_v2.md (canonical application)
workgraph_extended_outline_v2.md (canonical outline)
STATE.md §5 (the v2 pivot decision log — why KRAS was cut)
CLAUDE.md (especially 'Key Narrative Decisions' and 'The Demonstration')
The Poietic PBC public benefit statement, filed in Delaware. Verbatim text from ~/poietic/corporate/signed/amendments/01_Certificate_of_Amendment.pdf.txt: 'to make human and machine collaboration legible and responsive to its participants.'
The deployed landing page ~/poietic.life/index.html to see how the public benefit statement is currently presented

What to produce

A single doc at ~/poietic.life/notes/deep-synthesis-why-this-science-20260501.md, under 800 words total, with these sections:

Section A: Mission as spine (≤120 words)

State the PBC mission verbatim. Articulate, in one paragraph, why making human-machine collaboration legible is a real public benefit (not corporate filler), and why it requires demonstration on something with truth-value rather than aesthetic preference.

Section B: Why science is the proving ground (≤120 words)

Why scientific research is the right proving ground for legibility infrastructure, specifically: scientific reasoning has external validators (literature, replication, peer review), so if the trace is illegible the work is exposed. Code tasks, by contrast, can pass tests while being narratively incoherent. The bar science sets is what makes the infrastructure claim falsifiable.

Section C: Why pangenomics specifically (≤150 words)

Why pangenomics is the right scientific domain for this team: Erik's authoritative position (vg, PGGB, Nature Methods 2024); open questions that current single-reference analysis cannot answer; the existence of mature open-source tooling that is independently useful (so the work survives even if the demonstration fails). Connect to the founders' track record honestly. Do not overclaim.

Section D: Why PHRs in particular (≤200 words)

Why pseudo-homologous regions are the right specific question: Garrison Nature 2023 established the concept; PHRs sit precisely where standard short-read clinical sequencing fails (repetitive, complex, population-divergent regions); they have dual relevance — evolutionary (pattern of structural change across the vertebrate tree of life) and translational (causal variants for undiagnosed rare disease patients). Both relevances flow from the same underlying biology, so the science isn't bifurcated. State plainly that 'PHRs across all species' is not literally the deliverable — the deliverable is a methodology and an atlas growing from a defined starting region, with the rare disease cohort as the translational anchor.

Section E: What this commits to, what it doesn't (≤120 words)

Honestly enumerate: this proposal does NOT promise a clinical pipeline, does NOT promise to solve rare disease, does NOT claim CRISPRme or DNA-Diffusion are funded scope. It DOES promise an open atlas, an open methodology demonstration, an open benchmark (BioBench), and a published organizational design framework. Naming what's out of scope is part of the legibility commitment.

Section F: Honest read on the v2 application (≤90 words)

After writing A-E, ask: does the v2 application already imply this synthesis, or does the synthesis expose a gap that should be folded back? Identify specific sections (§17, §26, §29 are the prose-heavy candidates) where the synthesis would sharpen existing language. Be concrete: section number + one sentence of what to add or change. If v2 already lands this synthesis, say so plainly.

Constraints

Hard cap: 800 words total. Section caps above are ceilings, not targets.
No em-dashes (CLAUDE.md style rule).
No v1/KRAS terminology.
No 'PI' / 'lead PI' language.
Founder order Erik, Luca, Vaughn whenever named.
Do NOT edit workgraph_google_application_FINAL_v2.md. Output is the notes doc only.
Do NOT make claims about CRISPRme directly collaborating with Casgevy developers (CLAUDE.md framing rule).
Cite Garrison Nature 2023 and the PBC benefit statement at minimum.

Validation

All six listed files actually read
PBC benefit statement quoted verbatim in Section A
Six sections each present and within their caps
Total under 800 words
No em-dashes
Section F identifies specific application sections to sync (or states none needed)
Output at ~/poietic.life/notes/deep-synthesis-why-this-science-20260501.md

## Description

The Google.org submission is hours away. Erik is asking, honestly, whether the v2 scientific framing (PHR discovery in pangenomes, applied to undiagnosed rare disease) is *the* science to do versus *some* science Poietic PBC happens to be positioned to do. He's worried it might be a stitched-together justification rather than a coherent spine. He's also articulating that the company's actual reason for existing is making human-machine collaboration legible to its participants, which is the PBC benefit statement filed in Delaware. The science is the proving ground for the infrastructure, not the other way around.

What's missing as a single artifact: a synthesis that puts the mission at the spine and explains why pangenomic PHR work is the right proving ground for it.

Project convention: notes for Erik go to `~/poietic.life/notes/`.

## What to read

1. `workgraph_google_application_FINAL_v2.md` (canonical application)
2. `workgraph_extended_outline_v2.md` (canonical outline)
3. `STATE.md` §5 (the v2 pivot decision log — why KRAS was cut)
4. `CLAUDE.md` (especially 'Key Narrative Decisions' and 'The Demonstration')
5. The Poietic PBC public benefit statement, filed in Delaware. Verbatim text from `~/poietic/corporate/signed/amendments/01_Certificate_of_Amendment.pdf.txt`: 'to make human and machine collaboration legible and responsive to its participants.'
6. The deployed landing page `~/poietic.life/index.html` to see how the public benefit statement is currently presented

## What to produce

A single doc at `~/poietic.life/notes/deep-synthesis-why-this-science-20260501.md`, under 800 words total, with these sections:

### Section A: Mission as spine (≤120 words)
State the PBC mission verbatim. Articulate, in one paragraph, why making human-machine collaboration legible is a real public benefit (not corporate filler), and why it requires demonstration on something with truth-value rather than aesthetic preference.

### Section B: Why science is the proving ground (≤120 words)
Why scientific research is the right proving ground for legibility infrastructure, specifically: scientific reasoning has external validators (literature, replication, peer review), so if the trace is illegible the work is exposed. Code tasks, by contrast, can pass tests while being narratively incoherent. The bar science sets is what makes the infrastructure claim falsifiable.

### Section C: Why pangenomics specifically (≤150 words)
Why pangenomics is the right scientific domain for this team: Erik's authoritative position (vg, PGGB, Nature Methods 2024); open questions that current single-reference analysis cannot answer; the existence of mature open-source tooling that is independently useful (so the work survives even if the demonstration fails). Connect to the founders' track record honestly. Do not overclaim.

### Section D: Why PHRs in particular (≤200 words)
Why pseudo-homologous regions are the right specific question: Garrison Nature 2023 established the concept; PHRs sit precisely where standard short-read clinical sequencing fails (repetitive, complex, population-divergent regions); they have dual relevance — evolutionary (pattern of structural change across the vertebrate tree of life) and translational (causal variants for undiagnosed rare disease patients). Both relevances flow from the same underlying biology, so the science isn't bifurcated. State plainly that 'PHRs across all species' is not literally the deliverable — the deliverable is a methodology and an atlas growing from a defined starting region, with the rare disease cohort as the translational anchor.

### Section E: What this commits to, what it doesn't (≤120 words)
Honestly enumerate: this proposal does NOT promise a clinical pipeline, does NOT promise to solve rare disease, does NOT claim CRISPRme or DNA-Diffusion are funded scope. It DOES promise an open atlas, an open methodology demonstration, an open benchmark (BioBench), and a published organizational design framework. Naming what's out of scope is part of the legibility commitment.

### Section F: Honest read on the v2 application (≤90 words)
After writing A-E, ask: does the v2 application already imply this synthesis, or does the synthesis expose a gap that should be folded back? Identify specific sections (§17, §26, §29 are the prose-heavy candidates) where the synthesis would sharpen existing language. Be concrete: section number + one sentence of what to add or change. If v2 already lands this synthesis, say so plainly.

## Constraints

- Hard cap: 800 words total. Section caps above are ceilings, not targets.
- No em-dashes (CLAUDE.md style rule).
- No v1/KRAS terminology.
- No 'PI' / 'lead PI' language.
- Founder order Erik, Luca, Vaughn whenever named.
- Do NOT edit `workgraph_google_application_FINAL_v2.md`. Output is the notes doc only.
- Do NOT make claims about CRISPRme directly collaborating with Casgevy developers (CLAUDE.md framing rule).
- Cite Garrison Nature 2023 and the PBC benefit statement at minimum.

## Validation
- [ ] All six listed files actually read
- [ ] PBC benefit statement quoted verbatim in Section A
- [ ] Six sections each present and within their caps
- [ ] Total under 800 words
- [ ] No em-dashes
- [ ] Section F identifies specific application sections to sync (or states none needed)
- [ ] Output at `~/poietic.life/notes/deep-synthesis-why-this-science-20260501.md`

Depends on

done .assign-deep-synthesis-why

Required by

Log

2026-05-01T19:41:57.688135748+00:00 Task paused
2026-05-01T19:42:18.559374884+00:00 Task published
2026-05-01T19:43:19.845631623+00:00 Spawned by coordinator --executor claude --model opus
2026-05-01T19:43:36.659027685+00:00 Starting deep synthesis. Reading canonical files in parallel.
2026-05-01T19:43:50.812394685+00:00 Read v2 application, v2 outline, STATE.md, CLAUDE.md, certificate of amendment. Verified PBC benefit statement verbatim. Now reading landing page.
2026-05-01T19:45:00.036547081+00:00 All six files read. PBC benefit statement verbatim confirmed. Drafting 6-section synthesis doc.
2026-05-01T19:47:14.086080665+00:00 Committed 3512ac3 in poietic.life repo. Doc is 796 words across 6 sections; all caps respected. Ready for Erik review before submission. Section F flags §17a, §26, §29 for sync.
2026-05-01T19:47:17.834716106+00:00 Task pending eval (agent reported done; awaiting `.evaluate-*` to score)
2026-05-01T19:49:37.810027098+00:00 PendingEval → Done (evaluator passed; downstream unblocks)