Don't just base your conclusions on one page load, though, as this article seems to do. I wish there was a tool for Web Inspector/Firebug to select various elements over a series of page loads and give me statistics after e.g. 100 refreshes. Page load is highly variable and dependent on things like caching server-side, client-side, within the database, etc. as well as network conditions so repeating the tests is important.
None of those conclusions are based on the two screenshots shown in the article. I just set those up and included them for illustrative purposes. The article's points themselves are based on my experience routinely using those optimizations in production.
