This is why "the fold" should never be a hard number. You can still design for a larger portion of the initial viewport with percentage based "folds" (along with content).
Thanks for the share guys! I am the creator of the site.
I had some issues to begin with, where some fine folks were abusing the site. I think they're all pretty much tied up now though.
I've got some ideas about posts I could write about this project (from technical details to how the data stacks up) — is there anything in particular you guys think I should write about?
Interesting. There are several clusters: 360, 650, 960, 1555.
If it only shows unique heights, without information on how common it is, how is it possible that there are so many variations around a single value?
Are there so many devices with different size?
I would have expected to see fewer lines.
Well the data sent is "window.innerHeight" so if 1920x1080 is a popular screen resolution, some people will be using windows XP which will have 30 px taskbar size, and some people will be using 40 px taskbars with windows 7. Then some people will have browsers "not maximized" so they will be missing a few pixels too, and so on.
So from single height value, we have many variations differing by few pixels. Then, I think, we can gather some popularity info from this, if there is many variation around 1080 and less around 1200, this would mean the former is more popular.
So we don't need devices with different sizes but just a few different software choices.
Without knowing the specific data item collected, browser sizes can vary given constant screen resolution.
Just a few guesses off the top of my head:
Font size for the window title-bar, for the bookmark bar, for the tabs. Browser toolbars, whether or not title bar exists, whether or not bookmark bar is enabled. Whether or not window title-bar is enabled.
Status bar, task bar (could be at the side). Two windows stacked vertically on a portrait monitor.
