Meta-Modeling

These are frames that I can use for generating and improving models of systems or topics. There are plenty of examples in Modeled Systems.

1. Model - Technique framing -
   • Take techniques that work, and let the model be what must be true of the system for that technique to work. Then generate more techniques from that understanding.
2. Inverse Abstraction (Deconstruction, or Instantiation)
   • Break into component pieces, in multiple directions
     • Ex. Machine learning becomes Linear Algebra + Calculus + Probability Theory + Computer Science, which break into their own subregions
     • Ex. Scientific Field becomes Major Papers + Categories of the topic + Conferences + Major Researchers + Quality Sites
3. Simplify Recursively
4. First Principles
   • Ask what the underlying goal is for a space, and for what solutions would serve that goal. For each solution, think of the components necessary to make that solution happen.
5. List
   • Create a list of examples of the item being modeled.
6. Idea list technique
   • Set 10m timer, write down 10 creative ideas in a given domain in time given.
7. Problem - Solution Framing -
   • List important problems in / with the domain. For each problem, suggest a solution. Then criticise the solutions.
8. Abstraction
   • Get a list of examples of the thing being modeled, and look at what’s similar behind those many examples
9. Book
   • To construct a model from a book, skim through the book and write down every idea that appears multiple times.
     • Also works across a set of videos
     • Also works across a set of academic papers
   • Take the hierarchy from the table of contents
   • Take 3-4 books / papers with quality models, and look to find overlap and compress them.
10. People Based Framings
    • Visionaries / Leaders / Creators in a domain and their models for the space
    • Characteristics of the people who spend time with the domain and what that says about how it’s been conceived
    • Relationships between people
11. Organization Based Framings
    • Major organizations in a space, their road to prominence, their model of the domain and plans for action and progress
    • Problem Solving - cut out all the skills / resources necessary to solve problems in a space. Then break down the components that make up those skills or resources, and systematically identify failure modes for problem solving in that space.

Extension of working Model

1. Internal - External [Extends Model-Technique Framing]
   • For a model of a personal process, create a model for what’s happening and then generalize to group, or generalize the technique directly.
2. Synthesis
   • Take 3-4 models built by other people (often in the form of the above, or in the form of books, or an online course). Then compare them side by side and compress them into a single strong model.
3. Recursive Extension - Take a model, and create a model of each subcomponent of the model.
   • Ex. For a list, take each listed item and create a model of it.

Systematizing Systematizing

Examples: Systematized Predictive Modeling Systematizing Creativity

1. (Abstraction & Generalization) What kinds of x exist? Go from an object to the category the object is in.
   • (Idea List) For each kind of x, idea list ways to do x
2. (Decomposition) How can x be broken into sub-parts?
   • (Abstraction) What are the functions of each of the sub-parts?
3. What are the goals of the system and what are paths that lead to the accomplishment of those goals?
4. Optimize over sub-parts of x
5. Recombine sub-parts of x
   • Take the functional representation of sub-parts of x and generate new sub-parts or combinations of sub-parts that perform the same function
6. List the goals of the system
   • Generate feedback processes for each goal, using its associated metrics
7. Blow the system up and cut the system down
   • Add every conceivable valuable module to the system, under time pressure for generating those hypotheses.
   • Triage, cutting the system down to its most valuable parts
   • Create long and short term / large and small versions of the system
8. Look at the way the system grounds out in low level behaviors
9. Think about the way information needs to flow between different parts of the system, and at which points in time, and with which dependencies

Systematize(x) <- write this function

def Decompose(x):
* What kinds of x exist?
* In what ways can you do x?
* How can x be broken into sub-parts?
* What are the functions of each of the sub-parts?
* What are the goals of the system and what are paths that lead to the accomplishment of those goals?

def Systematize(x):
* Decompose(x)
* Optimize over sub-parts of (x)
* Recombine sub-parts of (x)
* Take the functional representation of sub-parts of x and generate new sub-parts or combinations of sub-parts that perform the same function

Decompose the problem into sub-problems. Write down algorithms that correctly identify the sub-problems. Write down high-level solutions. Decompose the high-level solutions into lower level, specific algorithms that solve the sub-problems. Compose those algorithms into a system that addresses the overall problem. Feedback on each part of your system using metrics that direct valuable adaptations.

Generating Generators

So.

There’s an algorithm for generating systematizing creativity that I like to describe as to carefully watch a brilliant mind generate creative ideas, and then to 1. through seeing patterns and similarities among the generated ideas, categorize them and 2. create categories for the thought processes or mental motions that generated the ideas.

Those are two generators of generators.
What else exists?
(Note, these are general enough to be meta-generators)

1. Categorize ideas using some pattern or notion of similarity. Use the examples to install the pattern, and look for the pattern elsewhere.
   • Ex., you see attempts to establish control (over an idea, or a situation), attempts to find the optimal solution, attempts to break an idea down, etc.
   • You turn these into generators by asking
2. Decompose a given pattern, seeing its variants, properties and types. Explicitly try to turn the pattern itself into a generator, as well as all of its decomposed variants.
3. Recombine your generators with one another
   • Ex., what is the most general version of recombination?
   • If I look at differing types of recombination, how do they differ? In the process, in the inputs, in the outputs?
4. Read creative authors, subject them to a deep analysis of their conceptual style.
5. Listen abstractively to people speaking. Name / categorize the mental motions that lead to creative ideas.
6. Gather powerful solutions to important problems. Ask how / why they work. Turn their source of strength into a property of solutions / ideas that its worth incorporating or founding new solutions on. (Ex., data structures as re-representation / abstraction)
7. Watch your own thought, especially when you’ve solved something or generated something new. Ask after the prompts that lead to that path of thought.
8. List the most powerful ideas.
   • Ask, how could I have come up with this, or seen this pattern for the first time?
     • Ask, how can I see this pattern more often when it comes up, and how does that process of pattern recognition work?
       • Often, by seeing many examples of a pattern, you mind starts to pick it up much more easily. So to see patterns that others haven’t seen, look at data that others haven’t looked at.
9. Learn to see the incompleteness of things (frameworks, conceptual schemes, say checking for collective exhaustiveness, or searching for datapoints that don’t fit) and complete them (or rework the ontology entirely)
   • Apply this to systematizing creativity, taking creative ideas that can’t be explained as a call to enlargen the technical toolset, or make existing techniques more flexible
10. Create your own concepts, elucidate them in some detail, name them, and then see what patterns exist in your concept creation.
    • This is general to all creative work. Start creating, and watch yourself closely.