Friday, January 1, 2010

Problem Solving Reference

Project Schema 
Define:
If you can't explain it clearly, you don't understand it yourself. Talk about it with other people. Engaging your speech in your thinking will help you understand it better.

* Describe specifically what we are trying to solve.
* How many are affected?
* How should we expect it to work?
    |_Why should we expect it to work like that?
* Do I have all the information?
* How can I restate the problem?
* Restate the problem as a question.
* What does everything have in common?
* What are the differences?
* What is the most important thing?
* Where does it come from?
* Where does it go to?


Brainstorm:
Write things down, make lists, draw pictures. This "extends" your brain and lets other people read your mind.

* What are the assumptions?
    |_Verify assumptions
* What is known about it?
    |_What do I know about it?
    |_What do my peers know about it?
    |_What do experts say about it?
    |_Who is an expert?
    |_Why are they an expert?
    |_What does the documentation say about it?
    |_Is the documentation reliable?
    |_What is the source of the documentation?
* Have I ever seen anything like it?
* What confirms it?
* What disconfirms it?
* What else could cause this?
    |_How would it cause it?

Create Timeline
* When was the last time it happened?
* Is it happening now?
* When was the last time it worked?
* When was the last time it failed?

Diagrams:
Lists, Functional Block, Flow Chart, Hierarchy, Timeline, Venn, Causal, Hypothesis Matrix

* What does it depend on?
    |_How much of what it needs is it getting?
    |_What is the hold up?
* What depends on it?
    |_How much is it producing?
    |_How good is what its producing?
* What relationships does it have, what does it connect to?
* How many "connections" does it have to which "neighbor"? How "tightly coupled" is it to anything?

Divide and conquer
* Can I solve part of the problem or improve it?
* Does it work somewhere else?
* Pick a point in your diagram and start there. The point that most things have in common is a good place to start.

Test
* Figure out a way to test it.

No comments: