Introduction
The Lean proof assistant has quickly gained popularity within the mathematical community. Is this fame due to its intrinsic merits or the influence of prominent figures? This article explores Lean's strengths and evaluates whether alternatives like Metamath or Isabelle/ZF could dethrone it.
Why Lean?
Lean has caught attention through notable projects such as Kevin Buzzard's Xena Project and Peter Scholze's Liquid Tensor Experiment. These initiatives showcased Lean's potential to validate complex proofs, whether human-generated or AI-generated. As of 2023, Lean version 4 is praised for its robustness and active community.
A major advantage of Lean lies in its use of the propositions-as-types philosophy. This approach allows for formalizing complex mathematics in a manner that is more intuitive for some users. However, it is not without criticism, especially when compared to more traditional set theory-based approaches.
Lean's Weaknesses
Despite its successes, Lean is not without flaws. Soundness bugs have been identified, raising questions about its total reliability for critical applications. AI is particularly adept at uncovering these flaws, but this does not solve the underlying issue.
Viable Alternatives
Metamath
Metamath, based on set theory, offers a higher assurance of correctness. Thanks to Mario Carneiro's work on Metamath Zero, the risk of soundness bugs can be significantly reduced. However, Metamath suffers from a lack of visibility and institutional support compared to Lean.
Mizar and Isabelle/ZF
Other alternatives include Mizar and Isabelle/ZF, which also rely on set theory. While they offer solid foundations, their adoption is limited compared to Lean, mainly due to a less dynamic ecosystem.
The Power of Influence
Lean's popularity can be partly attributed to the influence of renowned mathematicians who have chosen to adopt it. This dynamic, while effective at drawing attention, does not guarantee that Lean is the best technical solution. It is crucial for the community to continue objectively evaluating alternatives.
Conclusion
Lean has certainly made its mark in the field of proof assistants, but it is essential not to overlook alternatives that might offer more robust and reliable solutions. So, are we really stuck with Lean? The future of proof assistants may well depend on our ability to diversify our approaches.
Let's discuss your project in 15 minutes.