Analysis: There Is No Best Way to Think
The better thinkers know which method a problem demands, and when to combine them.
Table 1: Strengths and Weaknesses of Different Analytic Traditions
Most organizations do not train to that standard. They teach a familiar playbook, then expect people to apply it across a wide range of problems. The intelligence community is one example. Its structured analytic techniques can help analysts manage uncertainty and challenge assumptions. But when every problem is approached through the same playbook, analysts can be poorly prepared for tasks that require stronger causal reasoning, forecasting, measurement, or operational design.
The Silos Are a Problem in the Age of AI
These problems become more important in the age of AI. Generative systems can produce large volumes of plausible-sounding analysis quickly. That output can look like deep thinking, but it still requires critical evaluation, analytic supervision, and skepticism. Using AI well requires more than prompt fluency. It requires the ability to judge evidence, break problems into their component parts, identify assumptions, structure tasks, test causal claims, and decide where human judgment should remain in control.
No single discipline does all of that equally well. A course on forecasting can make someone a better forecaster. A course on intelligence analysis can make someone better at handling uncertainty and alternative explanations. But organizations working with AI need broader analytic capability: people who can identify what kind of question they are facing and select the method the question demands.
Good thinking is not the monopoly of any one discipline. The question is not, “what is the best method?” The question is, “what method does this question and challenge demand?”
Every profession has its own way of thinking. They each tend to describe their own methods as the legitimate way to reason through hard problems. Academics often emphasize the scientific method, even though scholars disagree about what that phrase means in practice. Business and military operations analysts tend to structure and quantify problems so they can compare alternatives. Intelligence analysts also structure analysis, often with an emphasis on uncertainty, deception, fragmentary evidence, and cognitive bias.
The development of these methods that can’t be applied elsewhere was not a mistake or shortsighted. Disciplines face recurring challenges that need to be solved repeatedly. These specialized methods are powerful inside their intended domain, but they can become brittle when the problem changes.
Start With the Problem, Not the Method
Practitioners are often very good at applying their own methods to familiar challenges. The difficulty begins when a method is treated as universal. A social scientist, an intelligence analyst, a forecaster, a business strategist, and an operations researcher may all see something different in the same problem because they have been trained to notice different things.
That is a weakness of disciplinary silos, but it is also an opportunity for those of us that don’t need to live in those silos. Each tradition has developed techniques that can be selected and combined according to the question at hand. Military affairs offers a useful analogy. Since the beginning of organized warfare, militaries have built tools and formations for specific tasks. The strongest forces are rarely those that rely on a single perfect weapon. They are the ones that know how to combine different capabilities, sequence them, and use each where it fits.
The same logic applies to analytic methods. Strong analysts do not begin by asking which favored method they want to apply. They begin by asking what the problem requires. Some questions require evidence evaluation. Others require causal inference, structured comparison, probabilistic forecasting, decision analysis, or a sequence of methods that answer different parts of a larger question. The better standard is not methodological loyalty. It is methodological fit.