A few days ago I said there was a large variance in the levels of attention of readers. The same is true for software users and the audience of talks. For example, Kieran Healy just happily discovered the OptiPNG hook in knitr, which was first introduced about five years ago, but I rarely mentioned it in the public (other than in the knitr book, which is not free, so not many people will notice it).
PNG plots produced from R can usually be (heavily) optimized via OptiPNG, as I mentioned in the Appendix B.4 in the blogdown book. I provided the chunk hook hook_optipng() in knitr to automate the optimization if OptiPNG has been installed.
From time to time, users keep discovering hidden features that I have rarely promoted, which often surprises me (such as the new JS macro feature in xaringan). The main reason that I don’t actively promote them is because I prefer deep wins over wide wins, i.e., compared to a large number of users moderately excited by certain features, I prefer a handful of users who are truly excited. Those who care about a software package will eventually discover all its features. For those who don’t care so much (which is totally fine), I try not to preach, and save the time for doing more exciting work.
When I give talks, I also notice that some people listen really carefully to what I say. For example, Rahul bought the book “Principles” since I mentioned it in my rstudio::conf 2018 talk. I just casually mentioned this book during the talk, without even putting its title or cover image in my slides, but he heard and remembered it.