Mental models exist and are quite simple. (Algebras? Graphs?)
Write your hypotheses explicitly when thinking!
Memory is generally similar to a tree. (Making cross-branch links is hard)
People react on presented options, and often do not consider the non-given. This is an error.
People do not understand how much (amount) data is representative of the matter (especially if it is consistent).
People are often blind to deliberate lying.
However, people often think that others are planning their actions very well.
People over-value their own actions in success but not failure.
People like logic an consistency. (People look for patterns.)
People like similarity between cause and effect both in nature and in size. People feel that correlation equals causality.
Draw contingency tables! (Yes, I mean you, the reader.)
Give explicit probabilities to events, and prefer probabilities to grades. (How likely are you to buy a game/film, not “how good it is”.)
The very thinking about something makes is seem more likely. (This may not be true)
Assessing something post-factum is always biased.
Trick to assess quality of information. If the opposite had occurred, would I have been surprised?
- Define the problem
- Generate hypotheses
- Collect information
- Evaluate hypotheses
- Select the most likely hypothesis
- Monitor quality in hindsight
Disprove yourself all the time! And read your enemies more than your friends.
- Sensitivity testing. What would change the assumption output?
- How do you find that this is no longer true?
- Are you attributing your own characteristics to other people?
Externalise. Write notes and draw pictures. Assimilate data into the schemata.
Make a hypothesis/evidence matrix.
Ask your friends/experts, but do not forget about data.
Read your old documents!
Use scientific method.
Perception is an active process of polling sensors, rather than receiving data.
Memory can be crudely modelled as a 3-level cache. Fun, cos computer memory is too.
Types of thinking (influenced by my programming language experience)
- Situational logic
- similar to imperative programming. Cause and effect are unique and concrete. Predict forward by extrapolation, or backward by reverse execution.
- Theoretical logic
- use generalisation to obtain for the same goals. (Similar to functional programming?)
- Logic by analogy
- machine learning? pattern transfer?
- Logic by data
- kinda same?
Quality of data is generally more important than quantity of data.
How the data may discribe the model:
- Data on variable value in model instances.
- Additional variables.
- Trustworthiness of variable data metadata.
- Variable coupling medatada.
Read aloud your thoughts!
Keep a list of unexpected events.
Summary for myself
Intelligence people’s work consists of writing reports. This is one of the reasons why literature is considered such an important subject in school. How many books do these people create each year? How much is declassified? How much is destroyed? How can this be compared to the FSB, GRU, and SVR? Who gives them assignments?
It is impressive how much the way a human society works resembles how a living organism works, consisting of different organs.
It is hard to make yourself think about problem solving directly.