<-- btw, anon b/c i don't feel like registering right now.
i found the author's comments mostly on the mark. while not everybody is dogmatic about these myths, large portions of the (especially inexperienced) o-s community typically preach/practice these myths.
-- and a comment about documentation/reading code. i'd be interested if somebody sampled projects on sourceforge to see how many had poor/adequate/good levels of project documentation. often without project doc, you have no alternative to dive in and read code. and when, in reality, alot of code doesn't get the cleanup it deserves, nor is the code itself well-commented, someone new to the project can spend hours trying to gain a 'larger view' of the project, hours wasted when 15 minutes of project-documentation could've told the story about how large pieces of the project go together.
yes, reading code is good practice, and is inevitably needed to learn the details of a project, but when reading code is your only source of info, then something is terribly wrong.
how long might you spend trying to create spec for apache (or some library) if all you had was the source code???