Thoughts on How to Abstract Subjectivity from Reductionist Analysis Processes
In my extensive understanding of cellular automata model applications, I have observed a phenomenon: cellular automata models (mainly heterogeneous models, which have broader expressive capabilities) are programmed with different rules—these are often related to their specific application contexts, such as traffic flow simulation automata, urban expansion simulation automata, biological pattern formation automata, etc.
Among the factors influencing rule selection, subjectivity plays an important role in choosing different rules. The subject defines the research scope and observation objects (although such selection may not be reasonably delineated and could be fragmentary in terms of actual problem solving, especially in complex biological systems), and the subject also limits the research purpose and understanding of rules (same as above, this also constrains certain things, though I cannot provide a good explanation for this).
Thoughts on Eliminating the Influence of Subjective Selection
So how do we eliminate the influence of subjective selectivity? I don't think this is a massive engineering project; rather, I believe it requires a complete overthrow in thinking.
Massive engineering projects, in the industrial and scientific history I have observed, have only appeared in those "mega-projects" that are saturated with reductionist blood. Intuitively speaking, massive and complex analysis "perhaps" and reductionism are twin brothers, while the proposition of eliminating subjective selectivity itself "should" be at odds with piles and piles of specific reductionist content.
Based on these two intuitions, I embark on the pursuit of designs for eliminating subjective selectivity.
How Subjectivity Affects Reductionist Modeling
To eliminate subjectivity, we must first understand how subjectivity affects our reductionist modeling.
The subjective selectivity of cellular automata mentioned above has two layers:
The first is choosing the cellular automata model itself
The second is choosing a specific rule that the subject believes can correspond to a particular discipline, containing many predetermined parameters
The second layer, due to its directedness toward specific problems, has particularly obvious subjective selectivity. From a more fundamental perspective, we should mainly examine the relatively vague first layer of subjective selectivity.
Analysis of Cellular Automata Model Characteristics
The characteristics of cellular automata models lie in a two-dimensional spatiotemporal discrete grid structure, where each cell has the same attribute table and its own attribute state, while applying the same set of rules.
Some previous studies have utilized this model structure, understanding this global rule evolution as "time"—this is a subjective choice; similarly, some other studies have researched surface environments (urban or natural terrain), utilizing this two-dimensional spatial structure—this is also a subjective choice.
The greater characteristic of these two types of studies lies in how they select specific rules and assign parameters to conduct effective simulation, but this doesn't mean that this layer of basic structure selection can be ignored.
Broader Model Selection
In a larger sense, this basic structure selection is vast:
Mathematical equation models
Multi-agent system models
Structural diagram flowchart models
Knowledge representation logical reasoning models, etc.
They each have distinctive inherent selection advantages. We can enumerate some typical selection advantage elements:
What the subject believes conforms to the nature and structure of the problem being addressed
What the subject believes will produce results that better align with the subject's purposes
What the subject believes makes problem-solving easier to understand and more consistent with their thinking and logical processes
Deep Analysis of Subjective Selection Elements
The above three selection elements actually point to different aspects of the deep subjective selection problem. Why a subject adopts this model rather than that one, why they abandon this analytical method rather than that one—all can be traced back to these factors to varying degrees.
Requirements for Design Schemes to Eliminate Subjective Selection
If a model design scheme for eliminating subjective selectivity is to succeed, then designing independently of these selection elements is necessary. That is to say, this design scheme, regardless of what the subject is (it could be humans, %&7cni18!bb%&b, two-headed monkeys, or the thinking species from distant parts of the universe), regardless of the subject's modeling or the subject living in a world with certain movements and structures, and within an environment that developed through certain logic to enable its existence—none of these should limit the structure of this design scheme.
This allows for a kind of structural independence: that is, if the scheme is to be implemented, its specific content might be filled with the logic and detailed content of the subject's environment and world, but the theoretical structure before implementation must be independent of the various a priori structural designs and detailed content when the subject conducts self-modeling.
This can serve as a basis for testing whether a scheme for eliminating subjectivity is effective.
Comparative Research Method for Finding "Invariance"
In fact, if the introduction of a priori design for subject self-modeling becomes the standard for judging whether a scheme for eliminating subjectivity can be effective, this suggests that such an effective scheme can be researched on any specific model or through comparative studies of multiple specific models, seeking a kind of "invariance" or some "invariants" or a certain "constant structure" in the comparison, which can help us design such an effective scheme.
My Research Practice
I personally select cellular automata models, flowchart-like models, and network models as comparison objects that I often use to consider the problem of eliminating subjectivity.
Taking cellular automata as an example, I often use the Game of Life rules as a special case. As mentioned above, choosing which rule is as independent to the elimination scheme as choosing which model—I just select some models I'm relatively familiar with to tackle this scheme-finding problem.
Current Situation
I am also still searching for specific elimination scheme designs, and there are still many fundamental problems that await development.