pg was just wrong here. Or rather, it's a position that could've made sense at the time, but only if the thing you're most interested in experimenting with were

pg was just wrong here. Or rather, it's a position that could've made sense at the time, but only if the thing you're most interested in experimenting with were mocking up page layouts to e.g. present to users for UX testing; layouting in 2008-era CSS was not something you necessarily wanted to do up-front, whereas with tables you could at least copy some boilerplate and have at it. (It makes no sense in context; it doesn't make sense in most contexts. But it is a position which could've been defensible at the time in a different context. Problems with tables include that (a) you don't want to write into a table layout; it's at best tolerable if you paste text into a template, and (b) there is no natural transition away from a table layout other than just redoing the layout wholesale. So even if you're experimenting with it, you probably should not ever let it go into production, because once it's there you will get convinced to leave it there for a long, long time.) It's been a somewhat less sensible position since flexbox support became widespread, and is frankly just wrong in 2017.