I was reading this story about Lompoc, Calif.'s soon-to-emerge wireless network: The town is planning full fiber-to-the-home but deploying a Tropos-based Wi-Fi network to fill the gap and offer some price competition and fuller service areas.
This paragraph piqued my interest:
"Even though no subscription prices were talked about, the magic number the city has repeatedly mentioned is $20 a month, which is less than what telecommunication titans such as Verizon and Comcast charge for high-speed Internet services. That price is also less than what America Online charges for dial-up, which is considerably slower than broadband and DSL."
Of course, the "cost" of dial-up is hard to analyze from any one angle. America Online certainly makes money from its dial-up operations on average, but the costs of the system are very interesting. Because dial-up service requires circuit-switched telephone lines, the more dial-up users, the more a central office of a phone company is filled with in-use circuits.
Many people get a second line for dial-up if they use it often, and while phone companies make money on those lines over the long haul, sometimes bringing the wire out requires a multi-year recovery of costs. With broadband in ascendance, it's likely that many of these additional lines will only be in place for a year or two before they're not used. But the copper stays on the pole and in the ground. (I once had a T-1 line (four wires) and three phone lines to my house. I now have one that, with line sharing, brings in DSL and dial tone; I use VoIP with my DSL provider for a second line.)
At the America Online end of things, their modem pool providers have to maintain a certain overage on the number of circuit-switched lines they maintain in each calling area--although I may be missing the boat and it's possible this is all back-hauled via ATM to centralized pools reducing local point of presence (POP) expense. But they are still maintaining thousands of POPs and millions of modems and phone lines, all of which involved circuit switching.
Fundamentally, subsidizing DSL service through competitive pricing allows the phone company to reduce their circuit-switched burden. Line-sharing has meant that less copper needs to be brought out and maintained. Overall, a dial-up line costs everyone more money than DSL: it's not peculiar that DSL and cable service cost less than unlimited monthly dial-up and a separate phone line.
Cable companies have a different problem, however. They've got this fat wire in your home and they need to fully load it with as many logical services as possible: phone, video, HD, video on demand, channels, broadband, what have you. The more money they can extract from that fat wire the better, meaning that subsidizing broadband service to keep you a subscriber--instead of switching to satellite and DSL for TV and broadband--makes it worthwhile for them, too.
The big switchover will happen in the next few years. By my reading, cable companies should be able to offer affordable 10 to 20 Mbps service in most areas that they already serve using the DOCSIS 3.0 standard. This doesn't require more infrastructure for them, but it does require swapping out cable modems already in use and upgrading head-end equipment. Likewise, DSL is now readily available in 6 Mbps down and 1 Mbps flavors around the country for under $100 per month. DSL speeds should start approaching what's standard in Asia as well.
Eventually, dial-up will be an expensive luxury. I see projects like Lompoc's setting the stage for more users giving up dial-up in favor of higher speeds, and then many of them migrating--as economic and infrastructure realities allow--to even higher speed services.