[mettamage]

Advocating for programmer ergonomics is always a good thing. And I think more people should advocate and try to design languages in such a way that the way of programming more closely resembles the actual real thing [1, 2].

As you might recall in cognitive psychology there's a specific idea that translating a problem to a more recognizable problem (or simply changing the symbols) is a good thing [3]. Having less working memory is a good thing.

Reading your post, I believe those principles are behind it.

[1] http://worrydream.com/LearnableProgramming/

[2] Sublime's feature of showing a color when you give a hex value.

[3] Chapter 12 - Cognitive Psychology (3rd edition) by Bruce Goldstein

[Ygg2]

While I love Bret's work I don't think that's scalable. If you can simulate time axis, it means it can simulate only systems that can are hundreds of times smaller than system resource. E.g. can you simulate a Kubernetes swarm with thousands of different settings?

It's a great learning tool, but not much else.

[jononor]

Depends on the desired level of detail, and how "simulate-able" the system is designed to be. For example if everything uses a central event queue, one can to Discrete Event Simulation by just jumping to the next event. However an exiting Kubernetes swarm is not very simulate, as with most other existing software. And until (or if) simulation becomes a priority, this will continue to be the case.

[mettamage]

Bret's work is an example of the principle. Sublime its feature, for example, is scalable.