paper: too much detail about the implementation and API

[rmflight]

Per my own understanding, JOSS papers should have a:

[x] summary
[x] statement of need
[ ] almost nothing else

The current draft includes

• Details about the speedy approximation implementation that seem out of place besides mentioning it in passing
• Examples of usage and differences with respect to Pearson that should be in a vignette

[muellsen]

Thanks for the comment! We agree that some JOSS papers do have just this basic structure but there are certainly many other examples, e.g., here, a paper that is similar in spirit, but structured differently, a recent contribution of my colleagues and myself, or this one. For this contribution here, we would like to follow a similar route. Also, if I understand @corybrunson's comment correctly, we may embed code and text a bit better by expanding the mtcars example (and connect the ternary example better to the code).

In fact, one question here is whether it would be OK to add a speed-up plot into the paper? This would bolster the performance claims (in addition to the reduced memory footprint).

[rmflight]

You've shown that others have gone this route, that's fine. I'm OK with it.