News for hackers. Stuff that matters.
Re:Timeline for Logic, λ-Calculus, and Programming Language ... by jcora on Tuesday September 03, 2019 @01:42PM (#20869022)
But this is clearly not a property of the particular notation, but lambda-calculus itself (and the way beta-reduction is defined). If you were to make it a constant-time operation in length it literally wouldn't be lambda-calculus.

Which actually makes me curious: what sort of change did you have in mind?

Re:Timeline for Logic, λ-Calculus, and Programming Language ... by whatshisface on Tuesday September 03, 2019 @02:30PM (#20869600)
If you could quickly translate a lambda-calculus expression into a program written in some other model of computation (that was easier to simulate on a Turing machine) and back, then you could do the complete reduction in the other model, only using lambda-calculus as your input and output. To perform the entire reduction you would still have to scan the expression in O(n), but it wouldn't necessarily be O(n * number of reduction steps). There could potentially be a relationship between the number of beta reductions and the number of steps required to execute the reduction in the other model, in which case beta reduction count would be justified as a measure of time.