You're right! How frustrating. The most frustrating is that I was just about to share this with some friends and know that I now risk either: 1.) Them smugly pointing that inaccuracy out and failing to enjoy how cool this is otherwise, or 2.) The website author correcting it by the time my friends see the link so any "btw there's a typo" comment I make being confusing. I'm going to share it anyway!

EDIT: I fear this may be worse than we had initially thought. The diameters of the Moon and Mars suffer the same problem and the pixel distances appear to be based on those wrong numbers so actually all the "apparent" distances are twice as long as they should be. (My working was to check that the Earth was indeed 100 pixels on my screen, calculate that 1 pixel = 127.42km, multiply that by the claimed "6033 pixels" to the moon to get 768724.86 which is twice as large as it should be...)

I tweeted the error to the creator (David Paliwoda) and it's now been corrected.

Shouldn't it be something more like 20,000 km, since the equator is about 40,000 km long and we're looking at half of it? Anyway 6,371 can't be right though.