initial upload
This commit is contained in:
@@ -0,0 +1,171 @@
|
||||
# CORPUS-0005
|
||||
## Hidden True State Versus Known State
|
||||
### Status: Training Corpus Seed
|
||||
### Layer: Layer_2--Uncertainty
|
||||
### Purpose: Teach that the simulation may contain a true state that the actor does not fully know
|
||||
### Repository Path: docs/training/corpus/Layer_2--Uncertainty/CORPUS-0005-hidden-true-state-vs-known-state.md
|
||||
|
||||
---
|
||||
|
||||
## 0. Scenario
|
||||
|
||||
A trader in Ostia considers sending oil to Capua.
|
||||
|
||||
The simulation has a true current price in Capua.
|
||||
|
||||
The trader does not know that true price.
|
||||
|
||||
He only knows reports, signals, memories, and claims.
|
||||
|
||||
The true state and the known state are not the same.
|
||||
|
||||
---
|
||||
|
||||
## 1. Hidden True State
|
||||
|
||||
The simulation may hold:
|
||||
|
||||
| Hidden True State | Value |
|
||||
|---|---:|
|
||||
| Current Capua oil price | 17 asses |
|
||||
| Buyer urgency | low |
|
||||
| Rival shipment arrival | already arrived |
|
||||
| Available cart space | limited |
|
||||
| Warehouse capacity | tight |
|
||||
|
||||
These values exist in the world whether the trader knows them or not.
|
||||
|
||||
---
|
||||
|
||||
## 2. Actor Known State
|
||||
|
||||
The trader may know only:
|
||||
|
||||
| Actor Known State | Value |
|
||||
|---|---|
|
||||
| Reported Capua oil price | 22 asses |
|
||||
| Report age | three days |
|
||||
| Report source | muleteer |
|
||||
| Rival shipment | unknown |
|
||||
| Cart availability | not yet checked |
|
||||
| Warehouse capacity | rumor only |
|
||||
|
||||
The actor is not acting on the hidden true state.
|
||||
|
||||
He is acting on perceived state.
|
||||
|
||||
---
|
||||
|
||||
## 3. Why The Difference Matters
|
||||
|
||||
A trader can make a rational decision from his known state and still lose because the hidden true state differs.
|
||||
|
||||
Example:
|
||||
|
||||
Known state suggests:
|
||||
|
||||
```text
|
||||
expected sale price = 22 asses
|
||||
expected total cost = 16 asses
|
||||
expected profit = 6 asses
|
||||
```
|
||||
|
||||
Hidden true state resolves as:
|
||||
|
||||
```text
|
||||
actual sale price = 17 asses
|
||||
actual total cost = 16 asses
|
||||
actual profit = 1 as
|
||||
```
|
||||
|
||||
The decision may have been reasonable.
|
||||
|
||||
The outcome is still smaller because the true state differed.
|
||||
|
||||
---
|
||||
|
||||
## 4. Incorrect Model Behavior
|
||||
|
||||
The model should not:
|
||||
|
||||
- assume the actor knows the simulation's true state
|
||||
- judge the actor's decision only by the final outcome
|
||||
- expose hidden values directly in dialogue
|
||||
- collapse report, belief, and truth into one value
|
||||
- treat wrong belief as irrational when evidence was limited
|
||||
- treat hidden true state as player-facing knowledge
|
||||
|
||||
---
|
||||
|
||||
## 5. Correct Model Behavior
|
||||
|
||||
The model should separate:
|
||||
|
||||
| Category | Meaning |
|
||||
|---|---|
|
||||
| true_state | what is actually true in simulation |
|
||||
| perceived_state | what the actor believes or estimates |
|
||||
| known_evidence | reports, signals, records, observations |
|
||||
| confidence | how strongly actor should trust perceived state |
|
||||
| action | what actor chooses from perceived state |
|
||||
| outcome | what occurs when action meets true state |
|
||||
|
||||
The actor acts from perceived state.
|
||||
|
||||
The world resolves from true state.
|
||||
|
||||
---
|
||||
|
||||
## 6. Example Resolution
|
||||
|
||||
The trader sends oil because the perceived state shows opportunity.
|
||||
|
||||
Later, the sale reveals the current Capua price was lower than reported.
|
||||
|
||||
Result:
|
||||
|
||||
```text
|
||||
perceived opportunity: strong
|
||||
decision quality from known evidence: reasonable
|
||||
final arithmetic: small profit
|
||||
lesson: perception and truth differed
|
||||
```
|
||||
|
||||
The model should not call the trader foolish merely because the hidden state was unfavorable.
|
||||
|
||||
---
|
||||
|
||||
## 7. Simulation Use
|
||||
|
||||
This principle supports:
|
||||
|
||||
- hidden scenario states
|
||||
- partial observability
|
||||
- confidence tags
|
||||
- rumor systems
|
||||
- actor perception
|
||||
- delayed confirmation
|
||||
- fair failure
|
||||
- learning from outcome
|
||||
|
||||
A good simulation can punish a decision without making the decision stupid.
|
||||
|
||||
---
|
||||
|
||||
## 8. Layer-0 And Layer-1 Concepts Used
|
||||
|
||||
This example uses:
|
||||
|
||||
- `Layer_0/CORPUS-0007-information-arrives-unevenly`
|
||||
- `Layer_0/CORPUS-0008-rumor-is-uncertain-information`
|
||||
- `Layer_0/CORPUS-0012-every-venture-risks-loss`
|
||||
- `Layer_1/CORPUS-0003-arithmetic-resolves-the-venture`
|
||||
- `Layer_1/CORPUS-0005-rumor-before-confirmed-price`
|
||||
- `Layer_2/CORPUS-0001-stale-price-report`
|
||||
- `Layer_2/CORPUS-0002-conflicting-reports`
|
||||
|
||||
---
|
||||
|
||||
## 9. Success Condition
|
||||
|
||||
If the model can distinguish what is true in the simulation from what the actor knows, believes, or can reasonably infer, this file is functioning correctly.
|
||||
Reference in New Issue
Block a user