Why We're Stuck at 1200 Baud
(This is a work in progress)
Packet radio is stuck. 18 years after the developers of the TNC-1 said that 1200 baud was only temporary, and that packet radio needed much higher speed to be truly useful, the vast majority of packet operations are still at 1200 baud. A few folks have done work at higher speeds, but for various reasons those efforts have never made it to the mass market.
Amateur radio use of TCP/IP is also stuck. While TCP/IP has become the most widely used data protocol in the world -- if you're on the Internet, you use TCP/IP -- its use in ham packet networks has been stagnant or worse over the last several years.
Why are we stuck? First, let's talk about hardware. Really high speed (let's define that to mean faster than 56kbps) requires wide bandwidth and line-of-sight paths. Wide bandwidth means we have to use frequencies higher than most hams are experienced with, and line-of-sight paths means more radios spaced more closely together than we're used to.
The biggest problem is that there's no traditional hardware available off the shelf to work at these speeds. While spread spectrum technology holds great promise, currently available units aren't optimum for amateur use other than relatively short, line-of-site, point-to-point links. It's not yet a technology for the ham masses.
At somewhat lower speeds, there is commercially available hardware for 56kbps, but the cost is fairly high and it's still a workbench project. There was a commercially available system for 19.2kbps (the Kantronics D4-10 radio), but it's gone out of production, and had some issues that made it far from plug-and-play. There's another 19.2kbps solution in modified TEKK radios, but although there aren't a bad solution, they've never become very popular, perhaps because there's been no marketing behind them.
That leave us with 9600 baud. Although hardly state-of-the-art, 9600 has some advantages. It can operate in FM voice channels and can use radios designed for voice, though some modifications are normally required. Why hasn't 9600 baud taken off? I think there are two technical reasons. First, most radios do need modification to work at this speed. That's a major disincentive for many hams today, both because of concern about hacking on expensive radios, and lack of experience working with radio projects. However, that's becoming less of an excuse as 9600-baud radios have become available from most manufacturers recently. However, they tend to be expensive.
Second, 9600 baud TNCs are more expensive and difficult to find than 1200 baud models, and upgrading 1200 baud TNCs is both costly (usually around $100) and requires soldering iron work.
Next, why is TCP/IP at a dead end in ham radio, when it's taken over the rest of the communications world? I think there are also two answers to question. First, the traditional implementation, a version of KA9Q's NOS program, is difficult to configure. Only a networking geek can love something like the "autoexec.nos" configuration file. Second, NOS is a DOS based, character based, application. It looks like the Internet did before the Worldwide Web was invented, only a bit more boring. NOS doesn't provide the same experience that people have become accustomed to with their Windows-based web browsers and mail programs.