In first order, it really does not matter (though both of your PDFs pull down, don't they?).
I heart that with traditional TTL logic, pulling an input to LOW took more current than to pull it to HIGH, hence pullups are more common. (Maybe someone can explain, are there usually PNP transistor bases on inputs?)
You get them for free in the Arduino, since they are integrated and switchable in the AVR. So with a pullup, you can just attach a switch und do the rest in software.
Also, ground may be better available at the switch than 5V. Usually, everything is filled with a ground plane anyway, you don't have to break it with a 5/3.3V trace to one switch terminal.