Investigation trail
Goal → prediction → experiment → observation → revision
Investigation intelligence / 01
Evidence Atlas
Follow the chain from goal to action to result to revision. Compare what Tend expected with what the evidence actually earned, and see which findings changed the next move.
◈
Evidence discipline is the interface. Observations, interpretations, and unknowns stay separate. Curated snapshot assembled November 9, 2026. Coverage: selected PR375, PR381, PR383 and PR384 results only. It is not Tend’s latest status, a live feed, or a complete research log.
Atlas entries
06
4 PRs · plus boundaries and costs
Obligations proved
8 / 12
Z3 screen · PR381
Unknown / unsupported
4
3 unsupported · 1 solver unknown
Transfer result
Negative
Later investigation cost increased
4 PR records
Decision trace
How the evaluation rule was revised
G
Goal
Determine whether retained learning improves a later investigation.
P
Prediction
Retained history might improve decision quality or reduce evidence and total cost.
O
Observation · PR375
Both guides chose candidate 6. Diversity did not produce a different consequential test.
R
Revision
Different choices are one diagnostic. Equal choices still require a fair cost and outcome comparison.
↗
Unresolved in this comparison
The cost of reaching candidate 6 under each guide is not reported here. No efficiency claim follows from the shared choice.
Evaluation rule
Measure outcome and total cost under fair conditions. A different action can explain an advantage; the same correct action reached with less evidence or cost can also be an advantage.
Current progress vs retained history · PR375
Guide comparison on one task; keep cost and outcome claims bounded
| Measure | Current progress | Retained history | Interpretation |
|---|---|---|---|
| Candidate selected | Candidate 6 | Candidate 6 | Same selected candidate; no action divergence shown |
| Evidence and total cost | Not reported here | Not reported here | Efficiency advantage remains unknown |
| Verified outcome | Not reported here | Not reported here | Do not infer equal quality from choice alone |
A different choice is one diagnostic, not a transfer requirement. A fair comparison could still find a benefit from reaching the same correct choice with less evidence or lower total cost. PR384 is a separate negative-transfer result, shown in the comparison view.
Evaluation rules
Inferences for assessing future transfer claims
01
Compare verified outcome and total cost under fair conditions; different actions are a useful diagnostic, not a requirement.Inference from PR375 · same candidate does not establish or rule out an efficiency gain
02
Report transfer cost and outcome. Consuming prior outcomes alone is not evidence of improvement.Supported by negative transfer result · PR384
03
Preserve uncertainty classes: unsupported semantics and solver unknown are different failure states.Applied to PR381 · not a generalized policy yet
Cost & unresolved work
Unknown is a first-class result
PR384 absolute investigation costUNKNOWN
PR384 baseline cost definitionUNKNOWN
PR381 unsupported legacy semantics3 cases
PR381 solver result1 unknown
Guide comparison · PR375
Current progress versus retained history on the tested task
| Measure | Current progress | Retained history | What is supported |
|---|---|---|---|
| Selected candidate | Candidate 6 | Candidate 6 | No consequential choice difference demonstrated |
| Evidence used / total cost | Not reported here | Not reported here | Relative efficiency unknown |
| Verified outcome | Not reported here | Not reported here | Quality difference unknown |
Choice divergence can explain a benefit, but it is not necessary. The same correct choice with less evidence or lower total cost would also count if a fair comparison established it.
Later transfer outcome · PR384
A separate investigation; do not combine its cost with the PR375 guide comparison
Prior outcomes were consumed, but the later investigation became more expensive. The numeric delta, cost definition and cause are not supplied in this snapshot. This is evidence against efficiency in that reported transfer, not a measured cost for PR375.
Separate proof and repair screens
Different tasks and measures; neither is a retained-history comparison
| Screen | Observed | Boundary |
|---|---|---|
| Invariant proof · PR381 | 8 of 12 proved; 3 unsupported legacy semantics; 1 solver unknown | Partial proof coverage |
| Repair loop · PR383 | 12 queries reduced 150 predicates to 5; 3 obligations passed | Bounded benchmark result, not transfer efficiency |
Repair outcomes
Preservation evidence is an implication, not reachability
+
Initiation positives
Recorded as constraints that must remain included.
→
Preservation implication pairs
If predecessor is included, successor must be included.
−
Sufficiency negatives
A failed preservation check can instead exclude its predecessor on restart.
Failed preservation & transfer
Two ways a model can mislead
FP
Failed preservation is a constraint signal, not proof that the successor is reachable.PR383 records implication pairs and restart exclusion behavior
NT
Negative transfer means prior experience made the later investigation more expensive.PR384 · mechanism and numeric cost delta unknown
Learning ledger
What changed in the system’s future choices?
Evidence classes
Read the epistemic status before the claim
Observed
Directly reported outcome or count.Example: 8 of 12 inputs proved in PR381.
Inference
A conclusion drawn from observations; scope stays attached.Example: tested guides did not demonstrate consequential divergence.
Unknown
Not supplied or not resolved. No guess substituted.Example: absolute PR384 cost delta.
Negative
Evidence against the expected improvement.Example: retained outcomes increased later investigation cost.