Skip to content
The Measurement Corpus

The lab is not the product. The data is.

FORGE sells qualification: measurement under vacuum, at temperature, with a traceable data package. Every job it runs also leaves something behind that never ships: a calibrated record of how real hardware actually behaves. That record compounds. This page explains the flywheel we are building the company around, counted from zero, in public.

0
Articles qualified
Hardware articles through the full qualification workflow with a delivered data package.
0
Measurement campaigns
Completed test campaigns with calibrated, traceable results in the corpus.
0
Instruments in the chain
Instruments on an active calibration schedule with traceable records.
0
Corpus records
Individual calibrated measurements retained against future campaigns.

Counters are real and start at zero. They move when the floor does. Last reviewed 2026-08-01.

Interface artifact · Computed 2026-09-16

Twenty-two interfaces have a value on both sides. Three agree.

In a public community dataset covering one lunar process, both the providing and the receiving organisation stated a value for the same interface twenty-two times. The two values were compared as stated.

22
Interfaces with a value stated on both sides
3
Where the two values agree
19
Where they do not
6
Where the receiving system needs more than the providing system plans to deliver

All three power interfaces disagree.

Power is the case with the least room for a difference of opinion, because the consumer either draws what it needs or stops working. It is also the case where every matched pair in this dataset disagrees.

InterfaceWhat the matched pair shows
Power interface 1The provider plans to deliver 40 percent of what the consumer states it will draw.
Power interface 2The stated values disagree. The magnitude is held in the computation and is not published here.
Power interface 3The stated values disagree. The magnitude is held in the computation and is not published here.

Method: every interface in the dataset where both sides stated a value was matched pair by pair, and the two values compared as written, with no normalisation, no unit conversion beyond the one the dataset already used, and no judgement about which side is right.

The contributing organisations are anonymised. Where this artifact needs to distinguish them it uses Contributor A, B and C.

Twenty-two matched pairs from seven organisations describing one process indicate that interface disagreement is measurable and worth measuring at scale. This is an indication, not a proof. Quantifying the problem across the lunar architecture requires a larger supporting count before this page can report its scale.

The flywheel

Four turns, each one paid for by a customer.

01

Measure

Every article FORGE qualifies is measured against acceptance criteria on calibrated, traceable instruments. The customer gets the data package; the corpus keeps the measurement.

02

Accumulate

Measurements are retained in a structured corpus: component behavior under vacuum at temperature, screening outcomes, instrument drift, uncertainty actuals. Data nobody else is positioned to collect, because nobody sells this measurement to U.S. defense customers today.

03

Compound

Each campaign makes the next one better: tighter uncertainty budgets, faster test design, screening criteria backed by observed populations instead of handbook values.

04

Return

The corpus flows back out as faster qualification cycles, priced screening data, and eventually reference datasets the industry checks against. The lab is how we acquire it; the corpus is what appreciates.

The record

What a corpus record carries.

The shape is fixed before the first record exists, because a corpus is only as good as the discipline of its first entry. Contents accumulate as the floor runs; the shape is the commitment.

article
Part identity, lot, and configuration under test
campaign
Test campaign, procedure revision, and acceptance criteria
environment
Vacuum level, temperature profile, and fixture configuration
measurement
Measured values with units and timestamps
uncertainty
Stated uncertainty budget for the measurement chain
traceability
Calibration chain for every instrument in the setup
disposition
Pass, fail, or nonconformance reference

The capability behind the corpus.

The corpus exists because FORGE is being built to sell the measurement nobody else in the U.S. offers commercially. See the qualification gap map for the four-way comparison, or bring us the part you need proven.

The Qualification Gap Map Request a Briefing