I have to give props to Wilson Rothman and the features team at Gizmodo. They offer some excellent information from a multi-city test where they pitted 3G networks against each other. Contending were the big-three: Sprint, Verizon and AT&T. T-Mobile was likely left out since they’re relatively new to the 3G market here in the U.S. You should definitely read what they found out: long story short, Sprint was the big winner in terms of download throughput, which is more important to most folks than uploading. AT&T offered better upload speeds in the tests, while Verizon was generally near the bottom in both. Now the Giz folks did this testing in eight metropolitan areas coast-to-coast, so why don’t I believe that this test is “definitive”?
Definitive is a pretty strong word because in this case, it would mean that you’ve removed all of the variables in testing so you could therefore draw a definitive conclusion. Eight cities is a great effort and you can’t take that away from the team. In fact, I applaud them for spending the time and effort. It’s those darn variables that come into play and here’s why.
Wireless broadband is a finicky beast. I’ve sat down with my Verizon Wireless EV-DO card and witnessed killer download speeds over 2.3 Megabits per second. Fifteen minutes later in the same exact location, I’ve seen those speeds drop to between 700- and 800 Kilobits per second. I didn’t move any farther or closer to the towers providing the service and yet the throughput dropped by over 60%.
I attribute this to at least two variables: increased wireless traffic congestion and the backhaul. Wireless traffic congestion is just what it sounds like: there are more users of the wireless service sharing the same “cell” at a given time. You can’t easily account for that in a test like this, so you really can’t remove that variable, nor can you tell how much it’s affecting your results. The wireless spectrum is pretty fixed and shared, so if you’re the only one on a wireless network, you’re in great shape. Saturate that same network with supported wireless devices and your data will feel like a cow in a cattle chute.
Backhaul is related and in simple terms, it’s the connection from the cellular tower to the web. Wireless users connect to the cell tower and their web traffic is sent back and forth over the fixed backhaul pipe to and from the Internet. Essentially, we’re all sharing the backhaul from a tower to the web and presumably, we’re all enjoying different activities. If I start watching videos on Hulu over EV-DO, it’s sure to take up more bandwidth on that pipe. Add in tens or hundreds of other users and their different uses and you’ve got another variable.
My point here isn’t to criticize the Definitive Coast-to-Coast 3G Data Test, because I think the Gizmodo team has done an admirable thing. I’m sure it wasn’t easy and they accounted for everything within their power to provide this comprehensive test. No, my point is to make sure current and potential 3G customers understand the different variables at play here. It can be a huge disappointment to purchase a 3G data plan that advertises speeds at 1.8 Mbps, for example, and then wonder why your speed tests show the expensive plan running at a crawl where there’s coverage. Bear in mind that you can often use software to tweak your network settings for a 3G speed boost. Oh wait… that’s another variable!