Whilst Bayesian epistemology is widely regarded nowadays as our best theory of knowledge, there are still a relatively large number of incompatible and competing approaches falling under that umbrella. Very recently, Wallmann and Williamson wrote an interesting article that aims at showing that a subjective Bayesian who accepts the principal principle and uses a known physical chance as her degree of belief for an event A could end up having incoherent or very implausible beliefs if she subjectively chooses the probability of an event F for which she has much poorer evidence. They also argued that their own version of objective Bayesianism is completely immune to that challenge. In this article, after having presented the strongest version of Wallmann’s and Williamson’s argument, I will show that if successful, it has far-reaching consequences and would not only invalidate moderate subjective Bayesianism and imprecise probalism but also a form of objective Bayesianism that relies on conditionalisation, the principal principle, reference classes, and the principle of indifference applied to the most basic partitions. I then argue that their argument can be defeated by adding the rule that it is always irrational to choose a probability that can be computed from the known probabilities associated to one’s other beliefs. I finally argue that the authors’ main intuition that probabilities have different degrees of reliability favours imprecise Bayesianism over precise Bayesianism.

尽管贝叶斯认识论在当今被广泛认为是我们最好的知识理论，但是在这种保护伞下仍然存在着相对大量的不相容和相互竞争的方法。最近，Wallmann和Williamson撰写了一篇有趣的文章，旨在表明接受主原理并使用已知的物理机会作为事件A的信念程度的主观贝叶斯主义者，如果她主观地认为可能会导致不连贯或非常难以置信的信念选择事件F的证据不足的概率。他们还争辩说，他们自己的客观贝叶斯主义版本完全可以应对这一挑战。在本文中，在介绍了Wallmann和Williamson的论点的最强版本之后，我将证明如果成功，它具有深远的影响，不仅会使温和的主观贝叶斯主义和不精确的权衡主义无效，而且会使客观贝叶斯主义成为一种形式，它依赖于条件化，主要原理，参考类别以及对最基本划分适用的冷漠原则。然后，我认为可以通过添加以下规则来打败他们的论点：选择一个可以从与其他信念相关的已知概率中计算出来的概率总是不合理的。我最后认为，作者的主要直觉是概率具有不同程度的可靠性，这主张精确的贝叶斯主义胜于精确的贝叶斯主义。主要原理，参考类别和冷漠原则适用于最基本的分区。然后，我认为可以通过添加以下规则来打败他们的论点：选择一个可以从与其他信念相关的已知概率中计算出来的概率总是不合理的。我最后认为，作者的主要直觉是概率具有不同程度的可靠性，这主张精确的贝叶斯主义胜于精确的贝叶斯主义。主要原理，参考类别和冷漠原则适用于最基本的分区。然后，我认为可以通过添加以下规则来打败他们的论点：选择一个可以从与其他信念相关的已知概率中计算出来的概率总是不合理的。我最后认为，作者的主要直觉是概率具有不同程度的可靠性，这主张精确的贝叶斯主义胜于精确的贝叶斯主义。