Last week, Newsweek ran a brief "Catch Up on Hot Blogs" column. They suggested that readers check out Perez Hilton's celeb-watching website because "the snarky one gets 105 million page views a month. That's hot." Other articles over the past few months have given estimates for his traffic ranging anywhere from 2 million page views daily (Entertainment Weekly) to 4.5 million page views a day (Globe and Mail) to a whopping 4 million unique visitors a day (Stuff, pictured). He's gotten lots of ink from the mainstream media lately; in addition to reporting on his ongoing legal drama (he's being sued by a number of photo agencies for using their photos without permission), everyone seems to be in awe of his seemingly astronomical traffic numbers. And the kids love him! He's being lauded as, of all things, a musical tastemaker in addition to a gossipmonger. But how much traffic does he really get?

ComScore, one of the two major internet tracking and market research firms, calculated that PerezHilton.com had nearly 1.7 million U.S.-based unique visitors in May. The other, Nielsen/NetRatings, comes in almost exactly the same—they say that just over 1.7 million people in the U.S. visited his site in May.

Where Perez really blows up in this data is in the number of page views.

Nielsen counted around 33 million U.S.-only page views for Perez in May. According to (notoriously unreliable) online traffic-counter Alexa, 60% of Perez's audience comes from the U.S., which would give him, by those wacky numbers, total page views for May of 55 million. (Alexa says something similar for Gawker about international traffic; we think instead that 90% of our traffic is from the U.S., so we take these country distribution numbers with a grain of salt. Also: Nielsen says that Perez had just 98,000 unique visitors in the U.K. in May.)

ComScore gives Perez 48 million U.S.-only page views for May. With additional international page views, hey, that's a lot! But it's still not 105 million—and sure not 4 million uniques a day.

One sort-of measure of traffic is Blogads, which is a clearinghouse for blog advertising—if you want to advertise on, say, a bunch of gossip blogs, the site lets you do it with a few clicks, providing the ad rates and the promised number of impressions in a week that the site can offer. This site is the only one whose statistics seem to align with Perez's—the ad at the top right-hand column of his site goes for an $9,000 for a week, with a promised 26 million impressions each week.

But just how does Perez manage to rack up even that many page views anyway? By the ComScore numbers, each visitor would have to be looking at about 26 pages, or, obviously more likely, returning that many times as "visitors." Here's to retention of visitors! The kids like it, the kids come back.

(To understand the numbers for other sites: Nielsen calculated that Gawker had 851,000 unique U.S. visitors and 12.8 million U.S. page views in May; this definitely differs from other estimates; Google Analytics said we had 9.7 million page views in June, with 3.2 million "visits" and 1.4 million of what they call "absolute unique visitors.")

Perez also had a big jump in traffic over the winter. This graph is of uniques on top and page views on the bottom:

Some sites—Drudge, New York magazine, and Slate among them—have a sneaky automatic refresh, which forces an unattended page to reload on the user's computer, thereby delivering free additional page views. If you look at his traffic from over the winter, as in the graph above, you might assume that was something he'd installed to account for those big jumps in November and December. But as near as we can tell, Perez doesn't have anything like that. In fact, this looks like an insane growth in popularity. Go figger.

He currently lists at number 21 on the Technorati list of 100 most linked-to blogs. (It's surprising he ranks this high, as others have pointed out.) Michelle Malkin comes in on that list at #11; her traffic, as per Sitemeter, is 4.5 million page views, with 3.6 million visitors last month. Boing Boing ranks at #2 on that list; that site's internal stats say that it got 2.8 million unique visitors and 24.5 million page views last month.

Of the outfits that measure traffic, Nielsen/NetRatings uses a method similar to how it counts television viewers: It measures audience data based on a recruited audience of 30,000 and extrapolates from that. Nielsen claims to be more accurate than most websites' internal traffic monitors because it excludes bots and other web crawlers that could artificially inflate traffic.

ComScore claims to have a roster of more than 2 million people who have agreed to allow comScore to "confidentially capture" their browsing behavior. ComScore has its own calculation issues.

Alexa, which calculates traffic based on the behavior of people who have downloaded its toolbar, is also suspect, since the sites that download its measuring toolbar aren't audited.

Because of these different metrics, all sites, Perez (and Gawker) included, have inconsistencies in their traffic numbers. (For example, while Nielsen reported that we had 12,771,000 page views in May, SiteMeter counted only 10,319,781 page views that month. So it's probably safe to say that we had somewhere in that range.)

But what's head-scratching is that the numbers that Perez reports, the numbers that end up in print, are so much higher than almost any traffic stats seem to indicate. We know the internet is all confusing and crazy to the journalists of the world—but why do his traffic stats get repeated, without apparent question or research, by so many news outlets?

Soon it might not matter anyway. Apparently it'll be all about how much time people spend on your site in the future. Page views? That's so last year.