From their website:

For the first time ever, see who is browsing on any Web page. Chat, get tips, share links, and meet new people with similar interests. Use the Web like you always do, only now, see what happens when you can see Who is Live!

Well…sort of.

One major difference between Whoislive and me.dium, again, from their website:

Can I also see people who are browsing on other pages of the same website?

No. You can only see who is live with you on one page at a time.

Me.dium had a map of the local neighborhood of sites your friends were surfing. More or less creepy?  You decide.

Anyway, this reminded me that, while the business model of me.dium didn’t get anywhere, there was some interesting technology under the covers. It was just getting off the ground, but there were hints at some possibly fertile directions for research and development.  It also reminded me that I started to write this up a few years ago and still had a draft around.  Most of the ideas are covered under patent applications made by me.dium (and so are public information already, but nearly unreadable) and some are fairly “obvious” (though that seems to be mostly neglected as criteria for patentability (see, e.g. this patent for a wooden dog toy you throw)). The explorations of how the system might work and attempts to create some toy examples to assist with explanations, as well as, the section on proposed new signals is my work. If you are at all interested in the details of how we were trying to create surf-with-friends, take a peak. (Fair warning, this draft hasn’t been edited much, so some of it might be a little cumbersome.)

Interest in the stock market starts to take off in early September, just as the NASDAQ starts the dramatic decline. Curiosity and research on the bailout don’t seem to be closely correlated with stock prices or interest in the stock market.

Interest in recession started back in June when the NASDAQ showed signs of weakness then declined.  More recently, interest in recession has continued to since late October.

Interest in Obama continues to accelerate, while query counts for McCain have grown only slightly.  Obama continues to widen the gap in the world of query attention.

Looking once again at the interest in McCain and Palin shows that they are running even.

Data

I added a few misspellings to the queries so, while the patterns are similar, the absolute numbers have likely changed a bit. Recent data is from pool of about 90K unique users per day searching on all major search engines and some little known engines as well. Searches on e-commerce sites are included as well.

Obama continues to outpace McCain in internet queries:

Gov. Palin continues to garner approximately as many query counts as McCain, but the trends for each are in the opposite direction.

The plot below compares the fraction of queries containing the terms "McCain" and "Obama" since May, 2008. The search term "Obama" appears in about twice the queries as "McCain."  Without digging further, there is not way to tell if this is favorable or negative interest.

Senator McCain’s campaign got more attention from Web surfers after the announcement of Governor Palin as his running mate.  The plot below shows the fraction of query occurrences of the term "Palin" compared to "McCain for the same period.

It is difficult to get a sense of scale on the interest in either presidential candidate. Comparing to a term that many Internet searchers are interested–yes, sex–gives some sense of relative interest. Below I plotted the fraction of searches containing the term "Sex" against the fraction of searches containing "Obama."

Inexplicably, "sex" lost favor over the summer up until the Democratic convention, when the frequency of the "Obama" term peaked. Not that neither candidate ever come close to matching our interest in sex. What does this mean for our democracy?

Data

The data is taken from click stream data for about 95K Web surfers each day. For each plot, I look for occurrences of the terms in all search queries each day. I don’t try to sort out all the misspellings or synonyms, so these results are most likely under counts. The data covers period well before the Democratic nominee was chosen and before the conventions. Numbers are reported as a fraction of the total search queries counted for each day (usually around 500K). Queries are representative of major search engines and shopping sites.