TE: So how much access to the Dreamcast hardware did this community manage to figure out? Can you guys pretty much program a game that is as technically good as the kind that commercial developers made for the system?
DP: In some ways we are more advanced than the commercial development kits, but given my limited knowledge of what they can do, I can't answer too definitively. Someone inside Sega once told me some of the Dreamcast engineers had looked at the [KallistiOS] code and were pretty impressed with it.
I think there's certainly enough low-level tools to do a decent game now. But it comes down to the normal bottlenecks for developers on any platform: access to resources and just using the tools to push something out.
TE: What tools do you guys use in homebrewing for the Dreamcast?
DP: These days, we actually make a lot of use of standard PC development tools - Visual Studio on the PC side of things, Xcode on the Mac. Most of the code work is done with a toolkit called Tiki, which is the result of factoring out a lot of pieces of Feet of Fury and porting them to be more modern C++ and focused more on a PC-ish environment. I had written this thing to port Feet of Fury actually, and someone ported it back to the Dreamcast, so now we use that for most things. It's much easier to develop and debug on a PC.
When we actually build for the Dreamcast, we use a free software tool chain from GNU along with KallistiOS, which includes a ton of ported extras, like Xiph's Ogg Vorbis libraries.
Tiki has actually been picked up and ported to a lot of platforms now, including the Dreamcast, GP32, Nintendo DS, Windows, Linux, Mac. It's worth looking at if you're looking for something simple to get going, because it serves the purpose of something like SDL, but without a lot of the bureaucratic overhead I think of when I think of SDL.
TE: What do you think is the biggest challenge about programming the Dreamcast?

DP: The biggest challenge these days is, oddly enough, acquiring parts. The best tools to work with are a brand-new Dreamcast and a broadband adapter. Both are nearly impossible to obtain these days. Thankfully, I have a big stack of Dreamcasts to cannibalize as pieces if others die. [The Dreamcast] was a neat little machine, but - I hate to say it - not very durable in the long run.
At one time, I had built a "Frankenstein Dreamcast" with flash ROM, an IDE interface, and hooks for an ISA network adapter, though I never quite got around to building all the logic for that. It was a really neat project, but it just got to be too much of a time sink for me to really justify it for its own sake, especially as how I doubt anyone would've picked it up and mass-produced it for other developers.
From a technical standpoint, it's a pretty powerful machine with some great tools out there now, but if you're not up to par on your debugging and other hackery, you won't get very far.