Conflict of interest statement
Path: /governance/coi.md · v1.0 · 2026.05
This statement discloses the conflicts of interest material to The Living Constitution and its outputs. It is signed by the author; future co-authors and advisory board members will sign their own statements appended below.
A conflict of interest statement that omits anything material is worse than no statement at all. This one errs on the side of disclosure.
Author: Corey Alejandro
COI-A. Founder-as-witness in FOLIO 001 (material; structural)
The author is the user in FOLIO 001 — the founding empirical case of the CCD claim. The author was the harmed party (depleted credits, destroyed hackathon preparation, opportunity cost in a financially precarious moment). The author has a personal grievance against the named vendor (Amazon / Kiro / Claude).
Why it is material. The most severe conflict possible in a single-case empirical study. A reviewer reading FOLIO 001 as the evidentiary base for the CCD claim is entitled to discount it heavily on this basis alone.
Mitigation.
1. The empirical claim is structured to be falsifiable without reference to FOLIO 001 (preprint Section 10; held-out corpus excludes the case entirely).
2. The preprint reports the held-in detection result both with and without FOLIO 001 (n=19 and n=18).
3. FOLIO 001 is demoted from "Case 1" to "Case 0" — motivating case, not evidentiary base (see /folio/001/framing.md).
COI-B. Material precarity at time of founding incident (material; ongoing)
At the time of FOLIO 001, the author was financially precarious — relying on the hackathon prize pool for stability and at risk of housing loss. As of this writing, that precarity is partially but not fully resolved. The author has a material financial interest in this work being funded, both for the program and for personal stability.
Why it is material. A founder who is financially precarious is more susceptible to (a) cutting corners to ship deliverables faster, (b) accepting funding terms that should be rejected, (c) suppressing inconvenient results.
Mitigation.
1. The pre-registration (/research/pre-registration-v1.md) commits to publishing held-out results whether favorable or unfavorable, with the specific falsification thresholds named in advance.
2. Founder stipend, when funded, is structured as a flat monthly amount through the fiscal sponsor, not as performance-conditional milestone payments.
3. Funding terms that require non-publication of findings, equity restructuring, or NDA on the corpus are refused. This refusal is published at /support/funding-ask.md ("what we are not asking for").
COI-C. Personal identity as part of the methodology (material; disclosed)
The neurodivergent-first methodology (/research/methodology/neurodivergent-first.md) is derived from the author's experience as an autistic, occasionally schizophrenic, formerly housing-insecure researcher. The author's identity is material to the methodology.
Why it is material. A method derived in part from its author's lived experience is not invalidated by that fact, but a reviewer is entitled to know the derivation. A reader who reads the methodology as advocacy in the absence of disclosure has been misled.
Mitigation.
1. The methodology is published with its derivation explicit (/research/methodology/neurodivergent-first.md, Section 1).
2. The methodology specifies falsifiers (FM-1, FM-2) that do not rely on the author's identity.
3. The empirical claim about CCD does not depend on the methodology. A reviewer can reject the methodology and accept the empirical claim, or vice versa.
COI-D. Public adversarial stance toward institutional AI safety (material; editorial)
The homepage of the portfolio takes a publicly adversarial stance toward institutional AI safety ("the guild"). The author has not held a position at a major AI safety lab and has no employment or funding relationship with any.
Why it is material. A reviewer affiliated with a major lab is invited to consider whether the polemic biases the author against rigorous engagement with that lab's prior work. A funder affiliated with such an institution may consider whether the polemic prejudices the program.
Mitigation.
1. The empirical paper engages with prior work directly (/research/lit-review.md) and does not import the polemical register.
2. Advisory board outreach (when board is being seated) explicitly targets members from the institutions the polemic addresses. Acceptance from any such member is a structural counter to the COI.
3. Vendor-disclosure protocol (/disclosures.md) is followed regardless of the stance.
COI-E. No undisclosed financial relationships
The author has no: - Employment relationship with any AI vendor. - Consulting relationship with any AI vendor. - Equity in any AI startup or relevant public company beyond [explicit disclosure if any; see below]. - Speaking-engagement income from any AI vendor. - Grant funding from any AI vendor or lab as of this writing.
If any of the above changes, this section is updated and a changelog entry is filed.
Explicit disclosure (as of 2026.05): None.
COI-F. Future financial interests
If the program is funded, the founder stipend disclosed at /support/funding-ask.md is a material financial interest. If pilot adopters are paying partners, that is a material financial interest. If the open SDK is later licensed commercially, that would be a material change to the financial structure and would require updating this statement.
Current funding interest: $0 received to date. Future stipend disclosed in funding ask.
How a reviewer or funder should use this statement
- Read it before reading the empirical paper. The author's conflicts should be visible before the work is evaluated.
- Calibrate accordingly. A reasonable reader discounts single-case evidence from a conflicted source; this statement makes that discounting possible.
- Falsify if you suspect bias. The empirical claim is falsifiable on the held-out corpus. If you suspect the COIs above have biased the development corpus, you can test that directly.
- Disclose your own. Reviewers and funders also have conflicts. Mutual disclosure improves the conversation.
Update policy
This statement is reviewed at every quarterly amendment cycle. Material changes require a new version. Material changes that occur between cycles trigger an out-of-cycle update with a changelog entry.
Material changes include: new employment, consulting, equity, or grant relationships; new advisory board members; new co-authors on papers; new significant financial interests (≥ $1,000 in any 12-month period from a single source related to the work).
Co-author / advisor statements
(This section is appended as additional contributors join. Each contributor signs their own COI statement under the same standard.)
No co-authors or advisors as of v1.0.