You can't watch TV, listen to the radio or read a newspaper today without being bombarded by ads for broadband Internet service. Cable and telephone companies are in a ferocious battle for market share, and they're all bragging about how fast they are. Unfortunately, they're usually exaggerating.
In fact, the gap between advertised and actual performance may be so great that it doesn't matter whether you choose cable, DSL or fiber-optic service. If your online life is mainly limited to Web browsing and e-mail, any broadband provider will deliver acceptable speed. But it's likely to be well below the advertised speed for the slowest category of broadband - basic DSL.
On the other hand, raw speed may be important if you frequently download large files, such as full-length movies, or if you want to watch streaming TV shows online. Even then, the performance of your Internet service provider (ISP) is likely to be well below its advertised speed.
Don't believe me? Check it out yourself. The Web is full of free broadband speed testing programs. The problem is that different tests are likely to produce widely differing results - because they may not test speed in quite the same way.
We'll look at two examples today and help you decide which one is more useful. Or you can try both; there's no charge.
Before you start testing, let's talk about how Internet speed is measured, advertised, and abbreviated.
Typically you'll see ads that say something like this: "Blazing 8 Mbps downloads," or "Rip-roaring 768Kbps downloads." These generally refer to how many bits per second the service can transmit to your computer.
A bit is a digital one or a zero - the basic unit of data - and it's abbreviated with a small "b."
The "K" or "M" at the beginning stands for "kilo," which is Greek meaning 1,000, or "mega," which means 1 million. So a service with an advertised speed of 768 Kbps can move 768,000 bits of data per second from its server to your computer.
An ISP that offers 4 Mbps downloads is claiming 4 million bits per second. Typically, upload speeds from your computer to another one are slower, but that's another story.
It takes roughly 10 bits to transmit a single character of data (known as a byte). The text in this column requires about 6,000 bytes, or 60,000 bits. So basic DSL (digital subscriber line) service operating at the advertised speed of 768 Kbps should be able to transmit this golden prose in less than 1/10th of a second.
To put this in perspective, the first modem I bought in 1983 transmitted data at just 300 bits per second, so sending a column to The Sun took almost 3 1/2 minutes, assuming the connection held up that long. That's why today's communication speeds seem like magic to old-timers like me.
Why aren't you likely to get the speed your ISP advertises? First, that speed is the maximum the network is set up to deliver - and then only on a point-to-point delivery of a single file from a server on the same network. If you venture no further than your ISP's Web site or servers, you might get close to the advertised performance.
But the real-world Internet is an unthinkably huge and complex highway system, with millions of byways, many of them slower than your ISP's network. The data that flows through them is pumped by millions of servers handling requests from millions of users every second. The data you send or receive may take dozens of hops from network to network, from server to server on the way to its destination. And it will travel only as fast as the slowest link in the system.
The lesson: Even if you pay for 8 Mbps of download capacity, it won't help if you're waiting for a server that can only pump out 200 Kbps.
Now consider Web browsing, which is far more complex than receiving e-mail or downloading a song from iTunes. A Web page is a simple text file that tells your Web browser how to display the text it contains, as well as the location of the photos, graphics, and other multimedia components on the page.
Your computer has to retrieve each of those files over that same labyrinth. The ads on commercial pages often come from faraway, third-party servers. By the time your computer completes dozens of these transactions for each page, your ISP's advertised download speed looks like wishful thinking indeed.
True, your Web browser stores some downloaded graphics on your hard drive so it won't have to reach out to a remote server if they're needed again. But there's still plenty of overhead built into every complex Web page.
How much overhead? The best speed test I've seen is a free download from PC Magazine called SurfSpeed. It duplicates requests your Web browser would generate from 10 popular Web sites, including Google, Yahoo, MapQuest, AOL, MySpace, Microsoft and Apple.
It measures your average, real-world surfing speed, then uploads the data - including your general location and ISP - to the magazine's servers. So PC has compiled the results of millions of speed tests from users and ISPs around the world. That lets you can compare how well you do with users of other providers, surfers in other states and ZIP codes, and so on.
For a custom test, you can add your favorite Web sites, although not all sites are compatible with the test protocol.
My Comcast service is theoretically rated at 5 Mbps, but SurfSpeed's browsing test showed I was getting 281 Kbps in real world browsing - less than 6 percent of the advertised speed.
As slow as it sounds, that was still still better than the average for Comcast users in Maryland (250 Kbps) or for Verizon DSL (230 Kbps). I won't belabor the details here. Try it yourself. You'll come to the same conclusion: As far as Web browsing is concerned, your ISP's maximum speed is virtually meaningless.
Other factors may also affect your perception of Web browsing speed - including the speed of your processor and graphics adapter in decoding and displaying Web images.
When you download large files, including music, high-resolution photos and video, your ISP is likely to deliver a lot faster performance. To test this kind of raw speed, visit BroadbandReports .com, a Web site with a wealth of information about broadband providers around the country - including reviews and message boards for users.
In its Tools section, BroadbandReports.com offers a choice of browser-based plug-ins (one written in Java, one in Flash) that test your ISP's speed by downloading and uploading a couple of medium-to-large files from a handful of servers around the country.
Here Comcast fared much better on my home computer: a 1.5 Mbps download from a New York server; 2 Mbps from Palo Alto, Calif.; a blistering 5 Mbps from Fort Worth, Texas; and even a respectable 444 Kbps from Perth, Australia. That's another thing I learned - distance does matter.
Another interesting large-file test is available at Speedmatters.org, a site sponsored by the Communications Workers of America (which promotes investment in broadband infrastructure that provides jobs for its members).
The site includes a report showing average transmission speeds in all 50 states. (Maryland ranks 10th in download speed and 20th in upload speed.)
Give it a try yourself. Here's where to find the speed tests:
SurfSpeed: www.pcmag.com/article2/0,1895,1960250,00.asp (This one may not work on corporate systems protected by heavy-duty firewalls.)