HOW INTELLIGENCE WORKS · VALUE OF INFORMATION · eduKateSG
How a Mind Decides Which Missing Fact Is Worth Obtaining Next
Value of Information is the intelligence process that compares the expected benefit of reducing a particular uncertainty against the cost, delay and risk of obtaining the information before deciding what to investigate next.
Decision → uncertainty → candidate information → possible result → expected decision change → benefit of knowing → cost and delay → gather, defer or act now.
This article belongs to the How Intelligence Works series. Question Generation owns turning missing structure into questions; Hypothesis Testing owns discriminating tests. Value of Information owns the resource-allocation question: which uncertainty is worth paying to reduce now?
The We-Could-Learn-More-but-Should-We Problem
More information sounds automatically useful. It is not. A test can be expensive, slow, distracting or incapable of changing the decision.
Intelligence therefore asks not merely “What do I not know?” but “Which unknown could actually improve what I do next?”
Information has value when knowing it can change a consequential choice.
1. Information Value Depends on a Decision
A fact can be interesting without being decision-relevant. Value appears when the information could change which action, plan or belief is preferable.
The same fact can therefore have high value for one receiver and almost none for another because their decisions differ.
2. Ask What Different Results Would Make You Do
Before gathering information, imagine its possible outcomes. If every plausible result leaves the decision unchanged, the information has low immediate action value.
If one result would reverse the choice, justify escalation or prevent irreversible harm, the information may be highly valuable.
3. High Uncertainty Does Not Always Mean High Information Value
A person can be very uncertain about something that barely affects the decision. Another uncertainty may be smaller but sit directly on the boundary between two actions.
Value depends on the combination of uncertainty, consequence and the probability that new information changes the preferred route.
The most uncertain question is not always the most important question.
4. Information Has Costs
| Cost | Why it matters |
|---|---|
| Time | The decision may expire while waiting |
| Money | Testing or research consumes resources |
| Attention | More investigation can distract from execution |
| Risk | Some tests expose people or systems to harm |
| Opportunity | Gathering one signal prevents gathering another |
| Privacy or authority | Useful information may not be appropriate to obtain |
5. Value of Information in Mathematics
In a problem, one missing quantity may unlock the whole structure while another can be derived later. Strong solvers ask which fact reduces the solution space most sharply.
This is a cognitive version of choosing the next useful measurement rather than calculating everything available.
6. Value of Information in Learning Diagnosis
A teacher rarely needs to test every possible weakness. One carefully chosen question may distinguish a language problem from a mathematics problem or a retrieval problem from a conceptual one.
The highest-value diagnostic is often the smallest test that changes the teaching decision.
This is why first-weak-link diagnosis should seek discriminating evidence rather than produce a giant inventory of everything a learner could possibly do.
7. Value of Information in Operations
Before shutting down a system, ordering parts or escalating an incident, operators may choose among inspections, sensor readings and tests.
The best next observation is the one whose result most changes the safe action relative to its cost and delay.
8. Sometimes the Right Decision Is to Stop Gathering Information
Perfect information is usually impossible and often unnecessary. If remaining uncertainty is unlikely to change the choice, further search can become procrastination.
Intelligence therefore needs stopping rules: enough evidence for this decision, under this consequence, by this deadline.
9. Value-of-Information Failure Atlas
| Failure | What happens | Repair |
|---|---|---|
| Information hunger | More data is gathered without decision purpose | State the decision first |
| Cheap-data bias | Easy facts crowd out decisive ones | Rank by decision impact |
| Certainty chasing | Search continues after action is already clear | Use stopping rules |
| Cost blindness | Information benefit is counted, acquisition cost ignored | Price time, money and risk |
| Wrong uncertainty | Effort reduces a question unrelated to the decision | Map uncertainty to action boundary |
| Delayed-information failure | Answer arrives after decision window closes | Include timing in value |
| Authority blindness | Useful information is gathered without proper right or consent | Respect permissions and privacy |
10. Value of Information and Search Are Different
Search and Exploration finds useful territory. Value of Information decides which search deserves scarce time before a decision.
Search asks where to look. Value asks whether looking there is worth the cost now.
11. Teams Need a Shared Information Budget
Teams waste effort when every function asks for additional analysis without comparing its effect on the final decision.
A strong decision process names the open uncertainties, estimates which ones could change the action and assigns research effort accordingly.
The best question is not the one that produces the most interesting answer. It is the one that most improves the next choice.
12. Institutions Need Decision-Centred Data Collection
Institutions can collect enormous volumes simply because measurement is available. Data becomes intelligence only when it connects to a purpose, action threshold and return path.
High-value institutional information often comes from signals that alter allocation, safety, access, maintenance or policy—not from metrics that are easiest to display.
13. Artificial Intelligence and Value of Information
Tool-using AI must decide when to search the web, query a database, run a calculation, ask a discriminating question or act from current information.
Unbounded tool use increases latency, cost and complexity. Reliable agents choose calls whose expected information gain can change the planned action or materially reduce risk.
When uncertainty is low-impact, the system should often proceed rather than gather information for its own sake.
14. The Value of Information Audit
- Decision: What choice must be made?
- Uncertainty: Which unknowns could matter?
- Candidate information: What could we learn?
- Result branches: What different answers could arrive?
- Decision impact: Would any result change the preferred action?
- Consequence: How important is that change?
- Cost: What time, money, attention or risk does the information require?
- Delay: Will the answer arrive in time?
- Authority: May we appropriately obtain it?
- Stopping rule: Is current information already sufficient?
15. Practical Application: Seek Information That Changes the Decision
Every extra question, search or evidence request costs time and effort. Value of Information asks which new fact would most change our understanding of the situation, choice of approach or final action.
Do not search simply because more information is available. Search because one missing fact can change the decision.
16. Return to the Question Worth Asking
Intelligence cannot investigate everything.
Value of Information turns ignorance into a prioritised research budget. It asks which uncertainty is close enough to the decision boundary, consequential enough and cheap enough to deserve attention now.
Sometimes the smartest next move is a question. Sometimes it is to stop asking and act.