n=2 is only very slightly better than n=1, and still nothing close to a "debunking". Neither the original claim in the tweet nor the alleged debunking provide anything substantial, and nobody should be claiming anything until they do something at least semi-rigorous (n>10, control group, actual sampling of Uber prices and then statistical analysis).
I couldn't do this, as none of the people in my friend circle have the Uber app, but I'm not in SV - seems like there should be a few HN denizens who could pull off something like this relatively easily and put the question to rest.