Hacker News Reader

>Further, it has been exacerbated by recent computing mechanisms, such as multithreading, NUMA, GPGPU, and distributed computing, which have moved away from the "single tape" Turing Machine model of computation that most programming languages are founded upon.

Stopped reading here. No programming language with arrays resembles single tape Turing machines. The correct model would be random-access machine (RAM).

View on HN

That's unfortunate, because they weren't talking about the physical execution, they were referring to the determinism, decideability, and serial-sequential nature of program execution.

I am curious in which way programming languages with arrays are not equivalent to a Turing Machine model. It's not clear to me what you mean with "resembles" and why that would be the same thing as "founded upon".

All Turing-complete programming languages/models of computation can simulate each other with polynomial overhead. What I think is weird is that the article specifically mentions that programming languages are built on single tape Turing machines (as opposed to Turing machines in general, or random access machines), because single tape Turing machines are highly impractical and programming them would feel completely different than programming with any real programming language. For example as basic operation as array access takes a linear number of operations in the size of the array in the single tape model.

Reply on news.ycombinator.com