I wanted to see which pages were performing best. I used speedIndex as a measure for this, to test for page's visual parts being displayed.
So I started by limiting my scope to top Alexa rank of 1000, and within this scope find the top 10 pages with best speedIndex. This would become my Wall of Fame. Let's see this for the run of 1 Feb 2014:
select rank, url, renderStart, speedIndex, onLoad, PageSpeed from httparchive:runs.2014_02_01_pages
where rank < 100 order by speedIndex asc limit 10
The interesting bit is to observe other parameters:
renderStart: if the page could be optimized better, to begin
rendering sooner, could the performance be improved? Almost all the
pages begin rendering > 1s.
PageSpeed: pages with low page speed score have high scope of improvement. Can these pages analyse using page speed, and improve performance even more?
onLoad: compare this to the onload times. It's interesting to see that onload doesn't correlate to speedIndex.
As properties keep optimizing their pages constantly, it would be interesting to see the patterns of each of the above pages over time, and see if performance has improved or degraded over time.