Very nice example of "testing hypotheses suggested by the data"...

This is a plot of the world's most famous pyramids and ancient sights. There is a documentary which "reveals" that a lot of ancient sights are lined up perfectly on a line around the globe that is defined by the position of the Giza Pyramids and Machu Pichu.

I found it interesting, so I plotted those sights and the line on a 3d globe. Indeed they line up very well. Though it's still far easier to explain this phenomena with a common mistake when it comes to statistics: "testing hypotheses suggested by the data". Essentially it says, if this line wouldn't work, we could come up with many other simple patterns, such as other lines, triangles, rectangles, circles, etc, that describes the data as well as this line does... So given enough space for possible models, you will find a random model that describes your data perfectly.

Since most people are really impressed by such a pattern, it is important to deliver good explanations for this faulty reasoning. How would you explain this to a five-year-old?