Sleeping time is a site with an interesting idea. It takes the time of a Twitter user's Tweets, and it makes a guess at their sleep schedule. The idea is that if you're sleeping, you won't Tweet (obviously this only applies to amateur Tweeters). Once you enter a person's nickname, the site draws out a visualization of the sleep schedule. For example here is my sleep schedule.
Well, the first impression one gets is that I sleep as much as a brown bat (19.9 hours per day apparently). The red area is much larger than half the clock, so it looks like I sleep the whole day away.
Of course the clock only has 12 hours on it, so while it looks like I sleep 2/3 of the day, I actually only sleep half of that, or 1/3 of the day, according to Sleeping Time.
So, if the visualization doesn't give you a quick visual idea of what portion of a day you spend sleeping, what is it good for? It tells you what portion of half a day you spend sleeping, but that's not really a metric that's ingrained in any of our brains.
This is the problem with data visualization. A graph often looks clear cut and simple, but it is often simply not showing you an answer to the question your are asking.
If you found this post even mildly interesting, then you should know that a great book about graphs, charts, and other ways of showing data is The Visual Display of Quantitative Information.