The Best Effort Science
At the onset of the COVID-19 pandemic, numerous nations and regions experienced significant turmoil. As this was a global pandemic occurring in a modern society, there were no readily available data or precedents to guide responses, resulting in decision-making through trial and error.
Across media platforms, social networks, and everyday conversations, calls for policy decisions based on evidence and scientific grounds were ubiquitous. But what does evidence mean in the context of an unprecedented event? If policy decisions are not permissible without sufficient scientific evidence, does that mean we must wait idly until data accumulates?
Here lies a clear tension between science and reality.
The Best Effort Science
The essence of what I propose as “The Best Effort Science” lies in the scientific spirit of doing our utmost to understand phenomena rationally and objectively, even when evidence is scarce or difficult to obtain, or when time and resources are limited.
Science is the intellectual pursuit of rationally and objectively understanding facts. To achieve this, it is crucial to gather an adequate amount of evidence that ensures objectivity and to verify results through experiments or simulations that yield the same outcomes regardless of who conducts them. If these rigorous procedures are satisfied, the understanding is deemed rational and objective.
This scientific spirit has enabled progress and contributed significantly to society. By applying the understanding accumulated through science, technological advancements have soared, and through societal implementation of these technologies, our quality of life has dramatically improved.
On the other hand, the traditional approach of science has proven insufficient for addressing complex and unprecedented crises like pandemics. This is because acquiring sufficient data or verifying results through experimentation is often impractical, and there is no time to wait in such urgent situations.
Moreover, science struggles not only in real-time crises but also in domains where significant future risks are foreseen.
In the case of global warming, even when the risks were evident with high certainty and became more apparent over time, the scientific community was compelled to invest enormous time and effort into validation. Meanwhile, societal decision-making was delayed until the precision of the scientific community’s findings improved enough to justify strong measures or ambitious goals.
Additional risks are also coming into view, such as those posed by gene-editing and AI technologies. The scientific community has likewise struggled to address these risks adequately. When such risks manifest, it is evident that their adverse effects will spread far faster than those of global warming, reaching irreparable levels, yet science remains powerless.
AI technology, in particular, lacks any historical precedent unlike global warming or gene-editing issues. It is an entirely novel concept with no evidence or possibility for experimental validation. As such, science has yet to even begin addressing the risks posed by AI, instead passively observing.
If science is faltering or doing nothing in the face of such issues, what can we rely on? Relying on individual intuition or democratic elections for decision-making will not eliminate the problems. For these issues, too, we must use the best available data and knowledge and act on rational and objective understanding to take the best possible measures.
Who then bears the responsibility of approaching these critical domains, beyond the scope of traditional science, with rationality and objectivity?
Is it philosophy, politics, or economics? I believe it is not the responsibility of these sectors. Only science can assume this responsibility.
The spirit of science lies in rationally and objectively understanding facts. Evidence and verification are not the essence of science. Considering the ease of obtaining evidence, the feasibility of verification, and the constraints of time and resources in real-world conditions, engaging in intellectual activities to rationally and objectively understand facts aligns with the spirit of science.
In other words, science is possible even when evidence is scarce, verification is difficult, or time and resources are limited. The current scientific community merely lacks an established methodology that can be appropriately applied to different subjects or situations.
The methodologies of the current scientific community do not define science. Science has the potential to expand its methodologies and broaden its scope. If science does not assume its responsibility in addressing unprecedented or anticipated crises, our society will have no choice but to rely on irrational and subjective judgments to navigate its survival.
In an era of emerging global or humanity-wide crises, clinging exclusively to traditional scientific methods is illogical and irrational. Science is not driven by societal demands, but it must address its own irrationalities.
Recognizing the necessity of expanding scientific methodologies to properly address problems like COVID-19 or global warming — issues that traditional methods have failed to address effectively — is vital for the advancement of science itself. For science to accurately recognize its own limitations and adopt new approaches is ultimately the best choice for both science and society.
This is the spirit of “The Best Effort Science.”
Binary Risk Theory
To understand the application and necessity of The Best Effort Science, I will introduce the Binary Risk Theory. This is a theory I devised, but it is not particularly complex; anyone with the right insight can easily arrive at the same conclusion. It is an extremely objective and rational theory.
Binary risks refer to risks that impose intolerable impacts on the individuals or groups affected. For an individual, this might be the risk of death or losing what gives life meaning. For a society, it could mean societal collapse or total annihilation.
The term “binary” signifies the two possible states: 0 or 1. This derives from the fact that the probability of a sustained binary risk materializing will eventually converge to either 0% or 100%.
For instance, even if the probability of a risk materializing within a year is 1%, if the risk persists over the long term, the probability of its occurrence across all periods increases. In the case of perpetual risks, the probability of occurrence ultimately reaches 100%.
When considering risks, one typically weighs the benefits of accepting the risk against its potential downsides. However, for binary risks, such a comparison is meaningless. This is because accepting a binary risk guarantees its eventual occurrence. When the risk materializes, all benefits enjoyed up to that point will be nullified.
Furthermore, binary risks cannot be compared or balanced against non-binary risks. The benefits of reducing non-binary risks are lost when a binary risk materializes. Thus, the only meaningful comparison for a binary risk is another binary risk.
This is the essence of the Binary Risk Theory.
Binary Risk Theory in The Best Effort Science
The Binary Risk Theory is not particularly complex; as previously mentioned, anyone can arrive at the same conclusion given the insight. In this sense, it is rational and objective. From the perspective of The Best Effort Science, it is clearly a scientific theory that can be acknowledged as established even without additional information.
However, if The Best Effort Science is not embraced, the Binary Risk Theory may be regarded as an unestablished idea. Proposing this theory would likely invite demands for prior research citations, evidence, or experimental plans. This is consistent with traditional scientific methodology, which remains the prevailing approach in the scientific community.
This assertion is speculative, but there is circumstantial evidence.
Risks that could lead to human extinction, such as existential risks or X-risks, are well-known concepts. Environmental degradation, artificial pandemics, and AI overreach fall under this category. Various researchers have investigated potential scenarios and probabilities of these risks, with AI risks gaining particular attention due to recent technological advancements.
However, as far as I am aware, a theoretical framework like Binary Risk Theory has not emerged. Despite its clarity and intuitive nature, its absence might be attributed to a lack of recognition of its utility in research outcomes or to the belief among researchers that developing such theories is outside the purview of science.
Yet, under the spirit of The Best Effort Science, this theory is the natural starting point for addressing X-risks, forming the basis for all subsequent discussions. When examined through this framework, efforts to scrutinize specific X-risk scenarios or estimate probabilities with precision seem misplaced.
If Binary Risk Theory is correctly understood, any entity that identifies a binary risk should allocate all available time and resources to minimizing the probability of that risk as close to zero as possible.
When standing on a railway track with a train approaching, the focus should not be on predicting the train’s arrival time or calculating survival odds upon collision. The priority is determining how to move to safety and acting immediately upon that determination.
In reality, however, X-risk research rarely takes such an approach.
In this way, Binary Risk Theory serves as both an application of The Best Effort Science and a compelling demonstration of its necessity.
Relative Fermi Estimation
Another application of The Best Effort Science is Relative Fermi Estimation.
Fermi estimation generally refers to approximating target values using available numbers and rough formulas. While this approach includes errors from imprecise input data and calculations, the estimated values can often be useful despite deviations of several orders of magnitude from the actual figures.
For example, knowing whether a value exceeds 100 is sufficient in many cases. If the estimated value is 10,000, an error of tenfold magnitude is inconsequential.
Relative Fermi Estimation applies this concept to compare scientific hypotheses. Traditional science deems hypotheses valid as long as they do not overtly conflict with available information. However, validating a hypothesis typically requires substantial evidence and verification through experiments or simulations. Without sufficient evidence, a hypothesis remains provisional and subject to ranking based on the number of supporting scientists or the opinions of established authorities.
In fields where no hypothesis has attained recognized validity, they are treated as equivalent. New hypotheses introduced to such fields are also considered equivalent or inferior until they garner sufficient support over time.
This situation persists, especially when the subject inherently lacks decisive evidence and experimentation or simulation is impractical due to complexity or scale. Hypotheses remain indefinitely unvalidated and thus equivalently speculative.
Relative Fermi Estimation objectively evaluates the validity of such hypotheses lacking evidence. It identifies key indicators and their rationales, then approximates relative differences between hypotheses using available data.
For instance, if two hypotheses yield estimates differing by two orders of magnitude while the underlying data and calculations have similar error margins, the difference cannot be conclusively attributed to validity.
However, if the estimate difference spans five or more orders of magnitude against a two-order margin of error, a clear relative validity difference emerges between the hypotheses.
While this estimation involves subjectivity and bias, explicitly stating the basis for calculations allows reevaluation by other researchers. If multiple researchers consistently identify significant differences, the relative validity of one hypothesis over another can be objectively recognized.
Although it remains inferential and relative due to the lack of direct evidence or verification, it constitutes a rational and objective scientific methodology. When one hypothesis demonstrates clear superiority under this method, that distinction can be scientifically acknowledged.
In the spirit of The Best Effort Science, the most favorable hypothesis should be supported as the best available understanding within the constraints of time and resources. This influences decisions on resource allocation and prioritization of further investigation.
Conclusion: Transparent Intelligence
Judgments based on intuition are often dismissed in science for being subjective and irrational.
However, intuition often aligns with rational and objective evaluations. Binary Risk Theory corresponds with our intuitive response to standing on a railway track. Similarly, Relative Fermi Estimation frequently resonates with intuitive decision-making. These methodologies formalize and systematize our intuitive sense of correctness into rational and objective evaluations.
These are not part of traditional scientific methods and might be dismissed as unscientific. Nevertheless, they adhere to the rational and objective spirit of science, complementing rather than contradicting established scientific methodologies. They extend science into areas where traditional methods are challenging to apply.
This extension is often guided by cases where intuitive judgment proves accurate. In such cases, the lack of existing scientific methodologies suggests potential for extending The Best Effort Science to include new approaches.
I refer to the domain where intuition aligns with rationality and objectivity as “Transparent Intelligence.” Transparent Intelligence provides reasonable understanding and judgment, even when not formally recognized by current scientific methodologies. To deny it risks narrowing or distorting intellectual capacity.
The question of whether something conforms to Transparent Intelligence should precede concerns about its scientific status. Its incompatibility with existing scientific methods does not disprove its validity. Intuitively acceptable judgments often reveal the latent potential for undiscovered scientific methodologies.
Recognizing the importance of Transparent Intelligence helps us identify gaps between it and established scientific methods. Bridging these gaps fosters new methodologies, expanding science and realizing the spirit of The Best Effort Science.
