Alexa is a terrible way to compare traffic on websites. Their data is biased towards people who run Windows and are willing to run their toolbar.

I know for sure that it has always massively undercounted reddit traffic for example but I'll bet it is fairly accurate for Yahoo, since "runs Windows and random toolbars" is their target audience.