Recorded history / identity / chemistry
Same molecule.
Different history.
An exact product identifier tells us what the record says was obtained. It does not tell us which reactants, catalyst, base, additive, or reaction record got there.
01 / Same target, different records
Choose an exact product identifier
The five controls are the complete product groups in the pinned 4,312-record dataset. Each comparison uses deterministic measured-yield extrema, not hand-picked winners.
loading- Same target identifier?
- --
- Same ORD record?
- --
- Same route identity?
- --
- Recorded yield gap
- --
02 / Stronger continuity evidence
When records actually point to each other
The islatravir cohort is different: later ORD inputs carry native reaction_id references to earlier records. That supports material continuity; matching compound text alone does not.
Why the arrows matter. They represent native source references, not a route inferred from similar chemical strings. The duplicated first reference is preserved exactly.
03 / Declared analysis space
Four routes, one admissible chain
Counterfactual shortcuts make the comparison finite. They are graph hypotheses only; none is promoted to actual chemistry.
04 / Bounded Onto2D result
What Historical Load means here
3 - 1 = +2Read the result at the right level
Source record
ORD IDs, source strings, inputs, conditions, workups, outcomes, measurements, DOI provenance, and cross-references.
Derived comparison
Exact-string groups, route identities, extrema pairs, and the two +2 Historical Load calculations.
Counterfactual
Three declared shortcuts used to bound the calculation. Their chemical feasibility is unknown.
Not claimed
No canonical chemistry equivalence, mechanism, physical batch without a source reference, safety, economic cost, or universal route score.