GENESISCORE / RESEARCH

A research surface for treating the output as a testable target.

GenesisCore does not claim that its output is already science. It claims that the output should be treated as a target for scientific verification and engineering discipline. This page fixes the experiment method, the verification method, and the public weekly board frame.

Primary Measure
Big Five

Main verification target for trait-level comparison

Minimum Complete Cases
N >= 300

The minimum condition for the main analysis

Weekly Cycle
Rerun + Summary

Weekly batch execution is part of the design

Public Board
Frame Live

The public graph surface is live while weekly metrics are still being connected.

01 / Why Big Five

Why compare against Big Five

Subjective resonance alone cannot support a product-level claim. GenesisCore uses Big Five as the main comparison surface so the output can be assessed against a known trait framework instead of only post-hoc feeling.

Primary

Big Five is the main evidence surface

The main experiment does not depend on user resonance alone.

Secondary

Feedback stays separate

Subjective feedback is stored as a UX signal rather than primary proof.

Posture

Verification target, not proof label

The page states a posture toward verification, not an earned scientific title.

02 / Experiment Method

Experiment method for the current phase

Only runs with research consent and complete required inputs enter the main experiment. Chart features, consent snapshot, Big Five scores, and version data are bound by analysis_id, while duplicates are reduced to the latest valid case.

01

Consent-Gated Capture

Only consent-on runs enter the research store.

02

analysis_id Binding

Chart features, Big Five, and consent snapshot are bound to one analysis_id.

03

Complete-Case Filter

Missing fields and duplicates are filtered out before the main analysis.

04

Weekly Dataset Build

The dataset is rebuilt weekly so the same rules can be rerun under the same conditions.

The current design does not make long-term user tracking the primary axis. It is constrained to what can work under the present service model.
03 / Verification Method

Verification method for the current phase

The main comparison is M0, a non-astrological baseline, against M1, which adds astrological feature vectors. The primary metric is MAE, the secondary metric is R², and the process assumes permutation tests, Benjamini-Hochberg correction across the five axes, negative controls, holdout rules, and weekly reruns.

Main judgment: whether M1 improves on M0 and remains supported under rerun conditions
Primary metric: MAE / secondary metric: R²
Statistics: 1,000 permutation tests with q < 0.05 correction
Guards: duplicate filtering, QC rules, negative controls, external holdout
Weekly cycle: rerun under fixed conditions to monitor reproducibility
04 / Weekly Public Board

Reserved frame for the weekly public dashboard

This section reserves the public surface for the weekly verification module. Even before the live data feed is connected, the product should lock down the four public axes: weekly verification targets, comparable cases, period-based trend agreement, and confidence-band distribution.

Coverage

Weekly verification targets

Frame Only

A frame for how many runs entered the weekly verification pipeline.

W01W02W03W04W05W06W07W08

Frame only. Weekly verification counts will replace this sample trend.

Comparable

Comparable cases

Frame Only

A frame for complete / comparable cases after QC, deduplication, and missing-value filtering.

W01W02W03W04W05W06W07W08

Frame only. Complete-case counts will be connected from the weekly dataset build.

Period Trend

Period-based trend agreement

Frame Only

A frame for showing agreement rates across short, medium, and longer windows.

7D
30D
90D
180D
M0M1
7D30D90D180D

Preview only. The dual bars represent a display frame for period-based agreement.

Confidence

Confidence-band distribution

Frame Only

A frame for how outputs distribute across confidence bands in a way general users can still read.

0-.25
.25-.5
.5-.75
.75-1.0
0-.25.25-.5.5-.75.75-1.0

Preview frame for confidence-band distribution.

All charts below are UI frames only. They are format previews, not live data.

05 / Boundary

What this page does not claim

Having a research surface does not mean astrology has already been scientifically validated. The claim is narrower: GenesisCore is being built so the question can be tested under fixed, falsifiable rules and monitored with weekly reruns and public metrics.

A null result can still count as a successful experimental operation if it is reached under the fixed preregistered rules.
Next Route

After research, move to pricing or runtime.

Once the methodology and the research posture are clear, the remaining next steps are price review and the actual runtime entry.