This is a really cool small scale experiment. More of this type of work is need so we can have fairer and more transparent search options.

I'd be curious to see the sensitivity of the ratings to things like rater composition (is it a quirk of Redditors they like Bing better?) and search topics.

Also makes me wonder how much of the ratings rank to do with a decline in quality (or diversification/drift away from the Redditor vector) of Google's own search raters (the index is heavily influenced by manual rating).