Wired.com surveyed 2,636 iPhone users in the United States, Europe, Australia and Canada. The results showed variation by country, major metropolitan areas and carrier.

"In our view, this data is a strong indicator that performance of the mobile carrier's network is affecting the iPhone 3G more than the handset itself," wrote Brian X. Chen of Wired.com. "[I]t's highly unlikely that Apple is going to wave a magic wand and say, '3G problems, be gone' with a software update. Before Apple can make such a claim, it needs to wait for all of its carriers to optimize 3G network behavior -- in terms of number of towers, how they're positioned and how much bandwidth each tower can handle."

AT&T in the Middle

The survey ranked U.S. carrier AT&T third, tied with Telstra, Telia and SoftBank. AT&T users averaged 990 Kbps, half of the 1,822 Kbps averaged on T-Mobile's European networks but miles beyond the 390 Kbps offered by Australian carriers Virgin and Optus. The average in Australia was 759 Kbps.

But the vast majority of participants reporting "0" speeds, meaning they dropped off of 3G entirely, were AT&T customers in the U.S. Out of more than 1,600 survey participants from the U.S., 63 reported speeds of 0. Only 80 participants worldwide reported speeds of 0.

In the American cities that should be the most built-out for 3G, many users are getting no better than EDGE speeds, the survey found. In San Francisco, one-third of respondents (10 out of 30) reported speeds just barely better than EDGE. This is probably because the number of iPhone 3G users are overwhelming the available 3G towers, the report said.

More Anecdotal Evidence

The survey results comport with new data from the Swedish site Teknik & Trender, which tested the iPhone's antenna in a test chamber by Bluetest. The test equipment measures how a mobile device sends and receives data under different conditions.

The results? The iPhone performed slightly worse at receiving signals than a Sony Ericsson P1 and slightly worse at sending than a Nokia N73, but the difference was minimal, Bluetest CEO Mats Andersson said.

New York-based Yahoo blogger Ben Patterson contradicted Wired's results by running an iPhone 3G and a Palm Treo Pro through DSL Reports' mobile-phone test suite. The iPhone was substantially slower, averaging download speeds of 173 Kbps compared to 293 Kbps for the Treo. And the iPhone's performance dipped as low as 72 Kbps, while the Treo never fell below 155 Kbps.

Finger-Pointing

Patterson's test is admittedly "just one guy in one place," but his results match what users have been anecdotally reporting for some time. Wired's poll is anecdotal, as well.

Reports show that problems have been rising since the iPhone 3G's launch, the Gizmodo blog notes, which supports the idea that carriers' towers are simply being overwhelmed.

In the final analysis, though, customers will not be satisfied with finger-pointing among Apple, its chip manufacturers or the carriers. As one commenter to Gizmodo wrote, "I guess they (Apple) figure that if they turn this into a giant game of CYA that they've got a 50/50 chance that all of the problems will be blamed on their partner and not them."