Suppose you went to the supermarket to buy a pound of steak for dinner and when you got home you noticed that the package seemed very light. \n\nSo you went back and complained to the manager, only to be told that the label says "up to 1 pound," and you're stuck with it. You'd be furious, of course. But that seemingly ridiculous stratagem is used every day by broadband providers across the country. Don't believe \n\nme? Check your agreement. In my case, AT&T tells me that I'm entitled to upload speeds of "up to" 3 Mbps and download speeds of "up to" 384 \n\nkbps. What do I have? Download speeds that average about 15 percent slower depending on the time of day, and upload speeds that are more or less \n\nas promised. You can do the math as well I can. A big file, such as a backup or a photo album that takes 120 minutes to download at 3 Mbps, takes an extra 17 \n\nminutes at 2.5 Mbps, my actual download speed. OK, so maybe that's not the biggest deal in the world, but why should I burn up an extra 17 minutes when I thought I was paying to avoid that? And \n\nas I found out, I'm lucky. Many consumers get just 50 percent of the speed they thought they purchased. If we were talking about almost any other consumer service, you'd have the choice of taking your business elsewhere. But not in broadband. \n\n"Ninety-six percent of the country has two or fewer choices for broadband -- the cable provider and the phone company," says Chris Riley, policy \n\ncounsel for Freepress, a non-partisan advocacy group.How Fast Is Fast?Not only do we lack a standard for acceptable broadband service, there's not even a standard definition of what constitutes broadband. The FCC, \n\nas part of the National Broadband Plan, has just begun collecting data on connection speeds across the country via a test posted on its website. Sure, that's a good idea, but hello, this is 2010! Why don't we know this already? Why, as Eric London \n\nof the Open Internet Coalition points out, are ISPs allowed to claim that 90 kbps is broadband? That 90 kbps figure may sound like a relic of the analog modem days, but in fact it's what you get on your iPhone when forced to connect to \n\nAT&T's sclerotic and nearly useless Edge wireless network. Authorized by Congress, the Broadband Plan lays out numerous goals for this country's data infrastructure, most notably assuring 100 Mbps access \n\nfor most of the country by 2020. To reach that goal, average download speeds in the United States will have to increase five fold. Although its data is incomplete, the FCC cites private research indicating that broadband customers aren't getting what they pay for. In fact, \n\nhouseholds are, on average, experiencing download speeds that are approximately 40 percent to 50 percent of the advertised "up to" speed they paid \n\nfor, the report states. Upload speeds average about \n\n45 percent of what's advertised. All in all, it appears that the average actual download speed around the country is about 4.1 Mbps. "There's no excuse for the industry to use 'up to' speeds," says Joe Ridout, of Consumer Action in San Francisco. "It's useless to consumers. The \n\nFCC should require the use of average speeds."Number GamesSadly, the slippery marketing of broadband service is hardly the only numbers game designed to fool technology consumers. Most egregiously, \n\nprinters are marketed as "up to" a certain number of pages per minute, a speed not often realized in the real world. Similarly, expensive ink cartridges are \n\nmarketed the same way, and if you think it's difficult to note how fast your printer prints, just try and keep track of how many pages you get from an ink \n\ncartridge. PCs and cameras are a bit different. If Intel says a chip runs at 1.63 GHz, it probably does; and if your camera is designed for a resolution of 10 \n\nMegapixels, you'll likely get it. However, it's taken consumers a long time to look behind those numbers and understand what they really mean. As many \n\nof us have learned, CPU speed and megapixels by themselves tell you little about the real-world performance of a PC or the quality of a digital photo. \n\nBroken SpeedometersUnfortunately, there are real questions about the reliability of download\/upload tests. Part of the problem is the very real issue of network \n\ncongestion. Connection speeds drop off sharply around dinner time in much of the country, as consumers jump on the Net. What's more, the tests themselves are often inconsistent. The FCC, for example, is using two providers: Ookla and M-Lab. In tests of my DSL \n\nconnection, I saw relatively minor variations in results between the two services, but others have found huge discrepancies. Writing in the Los Angeles Times recently, journalist David Lazarus said his cable connection was clocked by one of the services on the \n\nFCC page at 18 Mbps; and a few seconds later the other reported speeds of 6 Mbps. And if you use one of the numerous speed tests on privately \n\nowned sites, the numbers can be all over the map. My 2.5 Mbps connection (according to Ookla) plummeted to 1.16 Mbps when clocked by CNET's \n\nShopper.com.And you thought that butcher had a heavy thumb. San Francisco journalist Bill Snyder writes frequently about business and technology. He welcomes your comments and suggestions. Reach \n\nhim at firstname.lastname@example.org. Follow everything from CIO.com on Twitter @CIOonline.