initial upload
This commit is contained in:
@@ -0,0 +1,174 @@
|
|||||||
|
# CORPUS-0001
|
||||||
|
## Stale Price Report
|
||||||
|
### Status: Training Corpus Seed
|
||||||
|
### Layer: Layer_2--Uncertainty
|
||||||
|
### Purpose: Teach that a report may be true when spoken but unreliable when acted upon because time has passed
|
||||||
|
### Repository Path: docs/training/corpus/Layer_2--Uncertainty/CORPUS-0001-stale-price-report.md
|
||||||
|
|
||||||
|
---
|
||||||
|
|
||||||
|
## 0. Scenario
|
||||||
|
|
||||||
|
A trader in Ostia hears that oil sold for a high price in Capua.
|
||||||
|
|
||||||
|
The report may be accurate.
|
||||||
|
|
||||||
|
But the report is three days old.
|
||||||
|
|
||||||
|
The trader must decide whether the old price still helps him.
|
||||||
|
|
||||||
|
A true report can become dangerous when stale.
|
||||||
|
|
||||||
|
---
|
||||||
|
|
||||||
|
## 1. Report Received
|
||||||
|
|
||||||
|
A muleteer says:
|
||||||
|
|
||||||
|
> Oil sold in Capua for 22 asses.
|
||||||
|
|
||||||
|
The trader asks when the muleteer left Capua.
|
||||||
|
|
||||||
|
The answer:
|
||||||
|
|
||||||
|
> Three days ago.
|
||||||
|
|
||||||
|
The reported price may have been real then.
|
||||||
|
|
||||||
|
It may not be real now.
|
||||||
|
|
||||||
|
---
|
||||||
|
|
||||||
|
## 2. Known Facts
|
||||||
|
|
||||||
|
| Fact | Value |
|
||||||
|
|---|---|
|
||||||
|
| Origin | Ostia |
|
||||||
|
| Destination | Capua |
|
||||||
|
| Good | oil |
|
||||||
|
| Reported Capua sale price | 22 asses |
|
||||||
|
| Report age | 3 days |
|
||||||
|
| Current Capua price | unknown |
|
||||||
|
| Travel time from Ostia to Capua | not yet resolved |
|
||||||
|
| Rival shipments | unknown |
|
||||||
|
|
||||||
|
---
|
||||||
|
|
||||||
|
## 3. Why Staleness Matters
|
||||||
|
|
||||||
|
During three days:
|
||||||
|
|
||||||
|
- a rival shipment may have arrived
|
||||||
|
- the urgent buyer may have already purchased
|
||||||
|
- local supply may have changed
|
||||||
|
- rumor may have spread
|
||||||
|
- sellers may have raised prices at origin
|
||||||
|
- transport space may have been reserved
|
||||||
|
- the buyer's need may have disappeared
|
||||||
|
|
||||||
|
The report can be true and still no longer useful.
|
||||||
|
|
||||||
|
---
|
||||||
|
|
||||||
|
## 4. Possible Outcomes
|
||||||
|
|
||||||
|
### Outcome A — Report Still Useful
|
||||||
|
|
||||||
|
The price remains high.
|
||||||
|
|
||||||
|
```text
|
||||||
|
reported price = 22 asses
|
||||||
|
actual sale price = 22 asses
|
||||||
|
```
|
||||||
|
|
||||||
|
The trader benefits from acting on the report.
|
||||||
|
|
||||||
|
### Outcome B — Report Partly Stale
|
||||||
|
|
||||||
|
Some demand remains, but the best buyer is gone.
|
||||||
|
|
||||||
|
```text
|
||||||
|
reported price = 22 asses
|
||||||
|
actual sale price = 18 asses
|
||||||
|
```
|
||||||
|
|
||||||
|
The venture may still work, but margin narrows.
|
||||||
|
|
||||||
|
### Outcome C — Report Fully Stale
|
||||||
|
|
||||||
|
A rival shipment arrived first.
|
||||||
|
|
||||||
|
```text
|
||||||
|
reported price = 22 asses
|
||||||
|
actual sale price = 14 asses
|
||||||
|
```
|
||||||
|
|
||||||
|
The report was true when spoken, but harmful when acted upon as current.
|
||||||
|
|
||||||
|
---
|
||||||
|
|
||||||
|
## 5. Correct Model Behavior
|
||||||
|
|
||||||
|
The model should separate:
|
||||||
|
|
||||||
|
| Category | Meaning |
|
||||||
|
|---|---|
|
||||||
|
| report_content | what was claimed |
|
||||||
|
| report_age | how old the report is |
|
||||||
|
| report_truth_at_origin | whether it was true when observed |
|
||||||
|
| current_truth | whether it remains true now |
|
||||||
|
| action_delay | time before the trader can act |
|
||||||
|
| price_decay_risk | chance that the reported price no longer holds |
|
||||||
|
|
||||||
|
A report is not only true or false.
|
||||||
|
|
||||||
|
It is also fresh or stale.
|
||||||
|
|
||||||
|
---
|
||||||
|
|
||||||
|
## 6. Incorrect Model Behavior
|
||||||
|
|
||||||
|
The model should not:
|
||||||
|
|
||||||
|
- treat a true old report as current certainty
|
||||||
|
- ignore report age
|
||||||
|
- ignore travel time after receiving the report
|
||||||
|
- assume demand remains unchanged
|
||||||
|
- assume no rival acted during the delay
|
||||||
|
- treat stale information as useless in all cases
|
||||||
|
- collapse freshness and truth into one value
|
||||||
|
|
||||||
|
---
|
||||||
|
|
||||||
|
## 7. Decision Questions
|
||||||
|
|
||||||
|
The trader must ask:
|
||||||
|
|
||||||
|
- who gave the report?
|
||||||
|
- when did they leave Capua?
|
||||||
|
- did they witness a sale or repeat talk?
|
||||||
|
- how fast can I act?
|
||||||
|
- who else has heard the same report?
|
||||||
|
- how quickly can the price change?
|
||||||
|
- is the reported price tied to one buyer or general demand?
|
||||||
|
- what happens if the price falls before arrival?
|
||||||
|
|
||||||
|
---
|
||||||
|
|
||||||
|
## 8. Layer-0 And Layer-1 Concepts Used
|
||||||
|
|
||||||
|
This example uses:
|
||||||
|
|
||||||
|
- `Layer_0/CORPUS-0002-goods-have-local-prices`
|
||||||
|
- `Layer_0/CORPUS-0006-delay-is-economic-cost`
|
||||||
|
- `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-0005-rumor-before-confirmed-price`
|
||||||
|
- `Layer_1/CORPUS-0007-rival-buys-the-cart-space`
|
||||||
|
|
||||||
|
---
|
||||||
|
|
||||||
|
## 9. Success Condition
|
||||||
|
|
||||||
|
If the model can recognize that a report may be true, useful, stale, or dangerous depending on age and action delay, this file is functioning correctly.
|
||||||
Reference in New Issue
Block a user