PROPRIETARY RESEARCH SOFTWARE IN DEVELOPMENT

ASSAY

CHEAP CURIOSITY. EXPENSIVE BELIEF.

Most research systems make it easy to generate explanations. Far fewer make those explanations compete fairly, confront contradictory evidence, and earn the right to become conclusions.

Assay is a proprietary research software product being developed to investigate that difference.

It is designed to turn competing explanations into structured research objects, challenge them against shared evidence, expose relationships and dependencies, identify what prevents resolution, and govern how findings advance.

The ambition is not another machine that produces convincing answers. It is a system designed to make explanations earn them.

FROM RESEARCH METHODOLOGY TO SOFTWARE SYSTEM

Assay begins with a research discipline: define the question, map competing explanations, establish shared evidence challenges, and prevent unsupported conclusions from acquiring authority.

The software ambition is to make that discipline operational.

Rather than relying on disconnected documents, informal debates, or an AI model’s most persuasive narrative, Assay is being designed around structured research objects, explicit evidence requirements, dependency tracking, reviewable decisions, and controlled progression.

The methodology defines what responsible investigation requires. The proprietary software is intended to make those requirements repeatable, inspectable, and scalable.

THE ASSAY RESEARCH ENGINE

EIGHT CONNECTED STAGES OF INVESTIGATION

01

EXPLANATORY UNIVERSE

Build a broad field of plausible explanations before prematurely selecting a favorite.

02

ACTIVE FIELD SELECTION

Identify which explanations deserve immediate investigation and which should remain visible but inactive.

03

RELATEDNESS MAP

Identify where explanations overlap, depend on one another, compete, or may describe different parts of the same phenomenon.

04

SHARED EVIDENCE CHALLENGES

Design tests that multiple explanations must confront, rather than allowing each hypothesis to select only its most favorable evidence.

05

RESPONSE MATRIX

Record how each explanation responds to each evidence challenge, including support, contradiction, uncertainty, and missing information.

06

ROW AND COLUMN ANALYSIS

Examine both individual explanations across the evidence and individual evidence challenges across the competing explanations.

07

BLOCKER MAP

Identify the specific evidence gaps, unresolved dependencies, and research limitations preventing stronger conclusions.

08

EVIDENCE-EARNED PROGRESSION

Allow conclusions, new research branches, and stronger claims to emerge only when the evidence justifies them.

These stages describe the intended software workflow and governing methodology, not a claim that every component is already implemented or validated.

WHAT MAKES THE SOFTWARE AMBITION DIFFERENT

RESEARCH THAT CAN BE INSPECTED, CHALLENGED, AND REVISITED

COMPETING EXPLANATIONS, NOT A SINGLE NARRATIVE

Preserve alternatives long enough to test them instead of converging immediately on one persuasive story.

SHARED EVIDENCE, CONSISTENT CHALLENGES

Require competing explanations to confront common tests and contradictory findings.

DEPENDENCY AND RELATEDNESS TRACKING

Make the relationships between hypotheses, evidence, assumptions, and conclusions explicit.

EXPLICIT INDETERMINACY

Recognize when the evidence does not support a confident answer, rather than manufacturing certainty.

RESEARCH BLOCKER INTELLIGENCE

Identify which missing evidence matters most, what decisions it prevents, and what new information could unlock.

GOVERNED CONCLUSIONS

Separate observations, provisional interpretations, and evidence-supported findings.

EVIDENCE-EARNED DISCOVERY

Permit new questions and investigative branches to emerge from research findings rather than uncontrolled speculation.

Assay is intended to make the process of reaching a conclusion as inspectable as the conclusion itself.

TEN QUESTIONS WORTH INVESTIGATING

Assay’s proposed research program begins with consequential questions where competing explanations, conflicting evidence, or unresolved causal relationships make disciplined investigation especially valuable.

These are candidate investigations—not completed studies or established findings.

01

Why did homicide, traffic deaths, and overdose deaths all fall sharply and together after 2022?

02

What is actually driving retail electricity-price increases, and is the data-center explanation wrong?

03

Why did lethal violence fall dramatically while aggravated assault barely moved?

04

Why did U.S. traffic deaths fall from 2021–2025, and why are cyclist deaths still rising?

05

Is AI actually removing the entry-level rung of the labor market?

06

Did housing supply actually bring rents down, and why did metro rent paths diverge so sharply after 2020?

07

Why do working-age Americans die at much higher rates than peers in other rich countries?

08

Did American obesity actually turn, and did GLP-1 drugs cause it?

09

Why did U.S. fertility fall so far, so fast?

10

Did smartphones and social media cause the adolescent mental-health decline?

WHERE THE FIRST PROGRAM MAY BEGIN

Two potential starting points illustrate the kind of investigation Assay is being developed to support.

CONVERGING NATIONAL DECLINES

Why did homicide, traffic deaths, and overdose deaths decline over a similar period? Are these separate trends, or do they share underlying drivers?

DIVERGING ROAD SAFETY OUTCOMES

Why have overall U.S. traffic fatalities declined while cyclist fatalities have followed a different trajectory? What might explain the divergence?

Neither investigation has been selected as the definitive first program. No completed Assay findings are being claimed.

DEVELOPMENT STATUS

EARLY SOFTWARE DEVELOPMENT · METHODOLOGY UNDER EVALUATION

Assay has an articulated research methodology, an intended software architecture, and a defined field of candidate investigations.

The project is being developed toward proprietary software that can operationalize structured inquiry, shared evidence challenges, dependency analysis, research blockers, and governed conclusions.

The full system is not yet presented as operational, independently validated, or commercially available. The next work is to translate the research architecture into working software and evaluate it through actual investigations.

Assay’s ambition is not to make artificial intelligence more convincing. It is to build software that makes its conclusions more accountable.