This critique is just a bit unfair to Scheme, given his first couple examples being used to show how cumbersome it is compared to Miranda. Normally you'd have a fold function defined and use that instead:
(define (sum x) (fold + 0 x))
Writing it this way is more idiomatic Scheme, and isn't really cumbersome.
See chapter 1.3 of SICP. It talks about the motivation and usage of higher order abstractions in general, and also about fold specifically, which they call 'accumulate'.
I was asking more about code in the wild that uses accumulate instead of the naive recursive formulation. That's typically what we mean by idiomatic.
I see foldl often in haskell, but rarely in scheme. I'm not disagreeing that it's a great idea, just that this is how it 'would have been written'. Personally I find the naive recursive formulation plenty clear.
Maybe you've never seen it in Scheme because Scheme's standard library is so anemic. Of the usual FP higher order list manipulation functions it only provides map! And people are understandably reluctant to have everyone write their own differently named library functions. The situation has probably improved since the list manipulation SRFI (I think it is SRFI-1), as those are quite popular and available in most implementations.
