# We Must Seek Good Explanations The central theme of both *The [Fabric of Reality](Fabric%20of%20Reality.md)* and *The [Beginning of Infinity](Beginning%20of%20Infinity.md)* is that **we must seek good [Explanations](Explanations.md)**. The arguments are explained in detail in [Fabric of Reality](Fabric%20of%20Reality.md) and [Beginning of Infinity](Beginning%20of%20Infinity.md), however, there is an interesting counterpoint that is worth addressing. DD effectively treats seeking good explanations as a foundational assumption. He emphasizes that they are driving force to progress and they are better than alternative approaches (such as [Induction](Induction.md) or [Instrumentalism](Instrumentalism.md)). At first glance it appears that he is using an inductive argument here, namely: "good explanations have been good for progress and worked in the past, so we should expect this to continue". DD continually criticizes inductive arguments, so it seems unlikely that he would have allowed one of his core ideas to fall prey to their shortcomings. And indeed, he did not. His argument is actually a bit more nuanced. It hinges on the idea that the success of the scientific method, driven by the pursuit of good explanations, is a fact itself that requires explanation. So we can understand his position more effectively as: 1. The Success of Seeking Good Explanations is a Fact: Throughout history, the scientific method, characterized by its emphasis on seeking good explanations, has led to an unprecedented increase in knowledge and understanding of the world. This is an undeniable fact about the progress of science. 2. This Success Needs an Explanation: Like any other observed regularity or pattern in nature, the success of the scientific method itself demands an explanation. It's not enough to simply accept it as a fortunate accident. 3. The Nature of Reality as the Explanation: Deutsch suggests that the success of seeking good explanations points to a fundamental feature of reality itself: the universe operates according to "universal laws of nature" that are inherently comprehensible. The fact that we can uncover and understand these laws, he argues, explains why the pursuit of good explanations has been so fruitful. It's not merely a human invention; it's a method attuned to the underlying structure of reality. Now at this point you may start to wonder if this is a [Circular Argument](Circular%20Argument.md): is he not using the success of seeking good explanations to explain why we should seek good explanations? Not quite. Rather, this is a form of [Abduction](Abduction.md) (inference to the best explanation). Deutsch is essentially claiming that the best explanation for the success of the scientific method is that it's grounded in a fundamental principle of reality—the comprehensibility of the universe. This is highly related to [Rational Inquiry Requires Pursuing Good Explanations](Rational%20Inquiry%20Requires%20Pursuing%20Good%20Explanations.md). ## Good Explanations Force Taking a Position In order to generate good explanations, frequently one is [Forced To Take a Position](Forced%20To%20Take%20a%20Position.md). This could be because [Logical Consistency Forces Taking a Position](Logical%20Consistency%20Forces%20Taking%20a%20Position.md). But more generally it occurs when you encounter a [Problem](Problem.md) with two conflicting ideas. In order to achieve a good explanation, you must pick one. Consider [Logic](Logic.md) for a moment. Logic deals with [Propositions](Proposition.md) and how they relate to each other. In other words it is the study of the [Structure](Structure.md) between propositions. Given a set of axioms and rules of inference, conclusions are instantly implied by the system. It does not matter if we enact the rules of inference, the link is already there. The same type of [Structure](Structure.md) is present in good explanations! There is an inherent [Logic](Logic.md) to [Explanations](Explanations.md). They are both *hard to vary* and often *forced to take a position*. Put another way: good explanations have a specific type of *structure*. [Critical Rationalism](Critical%20Rationalism.md) attempts to find explanations that satisfy this structure. ## Why Should We Seek Good Explanations *Why* should we seek good explanations? Is this just a core assumption, a self evident axiom about the world? Or is it in fact independently explained? Is Seeking Good Explanations Circular? Initially one may view seeking good explanations as a core assumption. It serves as a fundamental principle that dictates how we gain knowledge, understand reality, and even determine moral values. It is a basic principle from which many other concepts are derived, rather than being a conclusion itself. But is that really so? Is there no independent argument for why we ought to seek good explanations? Let's start by treating it as starting assumption. In that case the logical flow will be: * Premise: We must seek good explanations * Conclusion: Theory A is preferable to theory B because it is a better explanation This is completely acceptable because the conclusion is a logical consequence of the premise. Now, say we had: * Premise: We must seek good explanations * Premise: A good explanation is one where theory A is preferable to B * Conclusion: Theory A is preferable to theory B because it is a good explanation This is a straightforward example of [Circular Reasoning](Circular%20Reasoning.md). The concept of a 'good explanation' is entirely dependent on the conclusion that theory A is preferable. This creates a circular dependency where the definition of "good explanation" already assumes the conclusion to be true. Now lets walk through a far more realistic example: * Premise: We must seek good explanations * Conclusion: Theory A is preferable to B because it is a better explanation * Conclusion: Because theory A then helps us make progress, it is "evidence" that we should seek good explanations Here, the first two statements, taken alone, form a non-circular argument similar to what we've discussed previously. The premise "We must seek good explanations" establishes a principle, and the second statement, "Theory A is preferable to B because it is a better explanation," applies that principle, using an implied independent criterion for "better explanation". The third statement, "Because theory A then helps us make progress, it is 'evidence' that we should seek good explanations," is where the circularity emerges. The problem is not that progress is made, but how this progress is interpreted as evidence. The argument implies: * We seek good explanations because doing so leads to progress. * We know something is a good explanation when it leads to progress. So our initial premise isn't really an independent assumption or supported by independent evidence or argument. It is supported by our *conclusion*. The argument becomes circular because the "evidence" (progress) is dependent on the prior application of the principle ("seek good explanations"). Essentially, the argument implies: * We should seek good explanations because seeking good explanations leads to progress. * That progress is defined in terms of what was achieved by seeking good explanations. Put as simply as possible we have: * We should seek good explanations because seeking good explanations leads to progress. * Progress is roughly defined as "improved explanations" * So we could just as well have written "we should seek good explanations because it leads to improved explanations". And depending on how we are arguing this is either [Circular](Circular%20Reasoning.md) or a [Tautology](Tautology.md). At this point we are starting to see two things. Firstly, the concept of seeking good explanations *does* seem to have a supporting argument, albeit one that is vague right now. Secondly, this vagueness is leading us towards circularity and tautology, so we must tread carefully. The reality is that seeking good explanations is not an independent assumption or axiom. We *should* seek good explanations because it leads to [Progress](Progress.md). But - and here is how we avoid the circularity - progress is defined as moving from [Problems](Problem.md) to better [Problems](Problem.md). It is in this way that we clean up all confusion. We still have our tautology of course - after all, seeking good explanations will lead to good explanations. But in no way are we defending our search for good explanations *with* this tautology. We are saying that we should seek good explanations because they lead to progress, and progress is independently defined, making no reference to explanations[^2]. Now, seeking good explanations can be seen as a [Means To An End](Means%20To%20An%20End.md) of progress, but it is not the definition of progress[^1]. We can now see that there is really one core assumption underlying all of this: that we should strive for [Progress](Progress.md), where we move from [Problems](Problem.md) to better [Problems](Problem.md). ## Self-Reference Is an Explanation for Seeking Good Explanations an example of a [Self-Referential](Self-Referential.md) argument? It is, but that is not inherently problematic. Not all self reference is [Bad Self-Reference](Self-Referential.md#Bad%20Self-Reference). For instance think about language. Language can be used in a [Meta](Meta.md) sense - we can use language to talk about language (consider grammar text books, or dictionaries). The explanation of seeking good explanations that we came up with doesn't suffer from any of the flaws of bad self reference. An example of bad self reference is [that which shows up in the hierarchy of theories](Hierarchy%20of%20Theories.md#Self-Reference). The problem there is that we have a closed off self-reference that prevents criticism and thus [Error Correction](Error%20Correction.md). There is something deeper at work here. The explanation of why we must seek good explanations operates within a "closed system" of rational thought. However, this closed system is *open* in a critical way: namely, it is open to criticism and revision. The system is [Universal](Universality.md) in the sense that it can accommodate any idea, criticism, or proposed alternative. We can say that is exhibits [Critical Universality](Critical%20Universality.md). The [Hierarchy of Theories](Hierarchy%20of%20Theories.md), on the other hand, does not display [Critical Universality](Critical%20Universality.md). We can "criticize" it by moving the elements of the hierarchy around, but we can never criticize it via external means. Put another way, there are criticisms that exist outside the hierarchy that we cannot employ. ## There Are No Exceptions Imagine we are in a conversation with [Mr Witt](Mr%20Witt.md). He states: "I seek good explanations 99.9% of the time. But a small fraction of the time I do not seek good explanations. I specifically do this in cases where good explanations do not exist, so I only worry about a reductionist prediction. This is fine, and it does not mean I lose any of the benefits of good explanations that you just mentioned, for I seek them 99.9% of the time—when it matters. I am pro explanation! To say otherwise is to misclassify me." How ought we respond to [Mr Witt](Mr%20Witt.md) in this case? To start, remember that the theory "we must seek good explanations" is a *explanatory theory* itself. It is a theory of epistemology and knowledge creation. It is currently our *best theory* for how to make progress and move from problems to better problems. However, we are [fallible](Fallibilism.md) and [All Knowledge is Conjectural](All%20Knowledge%20is%20Conjectural.md). It could be that eventually we will find a better way to achieve progress than simply seeking good explanations. But at the moment that is far and away our best theory for how to achieve progress. It is superior to all of it's rivals at achieving progress. Thus, there is no "proof" that seeking good explanations is optimal approach for achieving progress. But as of today it is our best theory, our best explanation, and we are right to rely on it for that reason. [Explanations Are Justified By Their Superior Ability to Solve Problems They Address](Explanations%20Are%20Justified%20By%20Their%20Superior%20Ability%20to%20Solve%20Problems%20They%20Address.md), and generating continual progress is a real problem we face. Whatever approach solves that problem the most effectively should be that which we rely on. Of course if another theory comes along and more effectively solves the problem of achieving progress, we would be right to reconsider (reject, update, etc) our theory of seeking good explanations. So we cannot rely on some formal proof to argue against Mr Witt. However, we can simply ask him: "You are proposing a new qualification to our existing best theory. This in effect has created a new theory. What problem does your theory solve? What is your argument in favor of this new theory? Have you criticized it? What does it explain? How does it fit with existing explanations? How, when and why does explanation not hold?". The fact of the matter is that his proposed theory solves *no problems* and spoils our best theory by raising all sorts of new problems. It's additional qualification is left entirely unexplained. Put simply, Mr Witt has *removed explanatory power* from our best existing theory by asserting that it has arbitrary exceptions, but he has not *explained how, when or why* those exceptions occur. A great example of this is explored in [A Conversation About Justification](7%20-%20A%20Conversation%20About%20Justification.md). As stated in the [Principles of Rationality](Principles%20of%20Rationality.md), theories are postulated in order to solve problems. Any postulate which solves no problem is to be rejected. Mr Witt will most certainly *not* have a compelling example of where explanation could not apply. However, he will most likely take one of two tacts. He will either start trying to [Vaguemanning his Argument](Vagueman%20Argument.md), making it ever more vague and abstract—a classic way to *avoid criticism*, or he will try and make it concrete via a terribly contrived example that is a distraction. At this point we have already provided a strong, compelling argument against Mr Witt's claim. However, we are just getting started. %%TODO: arbitrary boundary%% * This is an arbitrary boundary, is it not? Why does problem solving and explanation apply inside the boundary, but not outside? * Specify the boundary you are referring to. Without clarify the boundary, you are effectively vague-manning the theory and making it impossible to critique. * That we "should seek good explanations" is a theory/explanation. It is supported by argument. It is currently our best theory (see [Logical Consequences of Striving for Progress](Principles%20of%20Reason.md#Logical%20Consequences%20of%20Striving%20for%20Progress)). To add an arbitrary region where it works and where it doesn't must be argued for—there must be an explanation for this! * It is arbitrary. How do you know when there should be a boundary and when there shouldn't? Why should some small domain be cut off from explanation? This is reminiscent of [Defend Science by Arguing Against Arbitrary Boundaries](Defend%20Science%20by%20Arguing%20Against%20Arbitrary%20Boundaries.md). DD effectively had to argue this exact thing with many worlds theory %%TODO: bad philosophy entrenches existing errors%% * This is classic bad philosophy, allowing for some arbitrary boundary that is outside the range of criticism. * This tiny area is where the rot beings—we risk entrenching errors. If we allow areas to be exempt from explanation, then we are saying: there are places where criticism, reason, and improvement must not enter * Bad philosophy is not only false, but actively prevents the growth of knowledge - BOI pg 308 * Thus the theory claimed to stand outside all normal modes of criticism—a hallmark of bad philosophy - BOI pg 310 %%TODO: The argument really boils down to "but we can't prove that explanation always holds"%% * One may say "but *couldn't* it logically be the case that a boundary exists. Do we have *proof* that it doesn't?" Well certainly it *could* be the case. But due to [Fallibilism](Fallibilism.md) that is always true. --- Date: 20241011 Links to: [Fabric of Reality](Fabric%20of%20Reality.md) Tags: References: * []() [^1]: This is analogous to how a spaceship's purpose is not merely to burn fuel, but to transport its payload. Fuel burning is a means to that end. [^2]: But wait, do we use an explanation to evaluate our problems? Specifically, we say we want to move from problems to "better" problems. How do we define "better" problems? Wouldn't that require an explanation? And at that point, wouldn't we be saying: "we should seek good explanations because they move us from problems to better problems, where better problems are based on some explanation"? This is indeed correct, but it is not circular! Remember, a circularity would be "we should seek good explanations because it creates good explanations"—in other words, just restating the premise in the conclusion. Structurally, we are saying that our premises are "we should strive for progress" and "seeking good explanations leads us from problems to better problems, i.e. progress" and the conclusion is "we should seek good explanations"