OmniRayn · Foundry Classroom · Intro module

Reading
Scientific
Certainty

Confidence tiers vs. validation state.

Students Instructors
01
OmniRayn/Reading Certainty
StudentsInstructors
The two axes

Two questions every result must answer.

Question one

How strong is the signal?

The quality of the corroboration — what the computation actually found, and how much to trust it. This is confidence.

Question two

How far up the ladder?

How much real-world testing has happened — from an idea to a lab-confirmed result. This is validation state.

02
OmniRayn/Reading Certainty
StudentsInstructors
Question two · validation state

The evidence ladder.

01
Hypothesis
A computational prediction — an idea worth testing.
→
02
Community-reported
A lab has run it and shared a result.
→
03
In-vitro-validated
Independent labs confirm it; reviewer signs off.
→
04
Lead
Strong enough to take toward development.
03
OmniRayn/Reading Certainty
StudentsInstructors
Question one · confidence tier

How strong is the corroboration?

Grounded

Lines up with a lot of established biology. Strong corroboration.

Class-level

Looks like a known family — borrow some confidence, not all of it.

Novel

Little to anchor it. The most exciting — and the least certain.

04
OmniRayn/Reading Certainty
StudentsInstructors
Worked example · urolithin A × Parkinson's
0.861

PINK1 ↔ Parkinson's association

The only "strong" verdict in the catalog — ~1,700 literature references.

A textbook-strong hypothesis.

  • Mechanism: mitophagy — clearing damaged mitochondria.
  • Target: PINK1, the canonical Parkinson's mitophagy gene.
  • Confidence tier: grounded — about as strong as corroboration gets.
05
OmniRayn/Reading Certainty
The load-bearing distinction
The trap
0

clinical trials in this use

Strong computational corroboration is not efficacy.

A grounded signal, zero rungs up the ladder. However good the corroboration, it is not evidence the molecule treats the disease.

06
OmniRayn/Reading Certainty
Non-commercial
The license

"Computational predictions, pending wet-lab; structure-function framing only; not experimental evidence and not a disease, treatment, or efficacy claim."

Confidence tier is not efficacy. We say "computationally corroborated," never "validated," for anything not wet-lab confirmed.

07