布尔巴基基于《数学原本》的真理定义是否已过时？

Great question—this cuts to a core debate in mathematical logic that’s evolved a lot since Bourbaki’s foundational work in the mid-20th century. Let’s unpack whether their "truth = provability" definition is outdated, and why.

First, remember Bourbaki’s grand project: to unify all of mathematics under a single formal set-theoretic framework (their theory of sets, distinct from ZFC). For them, the goal wasn’t to explore philosophical notions of "truth"—it was to eliminate ambiguity and ground every mathematical statement in a rigorous, derivable system.

In their view, if a formula could be proven within their set theory, that’s exactly what it meant for it to be "true" in mathematics. This was a pragmatic choice tied to their mission: they wanted to show all existing math could be reduced to theorems of their system, so equating truth with provability streamlined that project.

Three key developments have shifted the mainstream view away from Bourbaki’s equivalence:

• Gödel’s Incompleteness Theorems: The big one. Gödel proved that any consistent formal system powerful enough to model basic arithmetic will contain statements that are true in the standard model but cannot be proven (or disproven) within the system. Bourbaki’s set theory is certainly strong enough for this to apply—so there are mathematical truths (by modern standards) that aren’t theorems of their system. This directly breaks the "truth = provability" link.
• The Rise of Model Theory: Modern logic draws a sharp line between syntax (provability, formal derivations) and semantics (truth in a model). We now standardly define "truth" as holding in a specific structure (e.g., the natural numbers, the real numbers) rather than being derivable. For example, the Continuum Hypothesis is neither provable nor disprovable in ZFC, but it can be true or false depending on which model of set theory you’re working in. Bourbaki’s definition collapses this critical distinction.
• Pluralism in Mathematical Foundations: Today, mathematicians often work with multiple axiomatic systems (e.g., ZFC vs. ZF without choice, or constructive set theories). A statement might be provable in one system but not another—if truth were just provability, its "truth value" would be system-dependent, which clashes with many mathematicians’ intuition that some statements have objective truth status.

Not entirely. It still has niche relevance:

• For Bourbaki’s own project, the definition is self-consistent. Their goal was to formalize existing math, not explore unprovable statements, so within their narrow scope, "truth = provability" worked.
• In weak formal systems (like propositional logic), completeness theorems do guarantee that all tautologies (statements true in every model) are provable. Here, the equivalence holds—but these systems are too weak to model most interesting math.
• Some constructivist mathematicians still tie truth to constructive provability, though their frameworks are very different from Bourbaki’s classical set theory.

In mainstream mathematical logic and philosophy today, Bourbaki’s "truth = provability" definition is no longer the standard. We’ve moved to a more nuanced view that separates semantic truth from syntactic provability, thanks to Gödel’s results and the development of model theory. That said, it was a reasonable choice for its time and purpose—so it’s less "outdated" than it is superseded by later, more precise frameworks.

内容的提问来源于stack exchange，提问作者Randy Randerson