General reflections - read first.

[RobertASmith]

Hi Nathan,

I think a paper on this topic would be super useful and is long overdue.

I think areas that are missing are:

• how outputs are displayed - lets converge on a set of plotting functions that have been verified.
• When looking at modelling code I'm also interested in following the code through, can I quickly see which functions are nested where. Is the code messy, well commented, are there places where errors are likely to occur (e.g. data hard coded in functions, or processes replicated in code that should have used a custom function).
• Also interesting to a lot of people is the use of C++ to speed up R code, within the packages section it may be worth mentioning some of the benefits of packages where groups have spent a long time making them very efficient. But also trade-off the potential 'black-box-ness' of health economists just using others' packages without understanding them.

I've put some other comments on your individual sections - more than happy to discuss if useful.

Cheers,

Rob

[giabaio]

Thanks, Rob - this is all very helpful! Let's keep the discussion going!! :-)