A while ago I've been interviewed for a forthcoming article in the LinuxUser magazine on the topic of why there are so many Debian derivatives and on the potential impact, on them, of the (potential) advent of the so called "Debian Rolling" experiment.
Below you can find my take on those topics. Questions are by Ferdinand Thommes, who has been involved in various Debian derivatives, including the forthcoming Siduction.
1.) Why do you think people base distributions on Debian? Looking at how many there is, since Corel and Libranet set sail in 1999, there must be a reason why, for one, people fork Debian in the first place, and secondly, why they choose Debian to do so and not some other distribution.
There are always advantages in creating a so called "derivative" distribution with respect to creating one from scratch. The main one is the possibility of reusing the packaging work which is already done in the "upstream" distro, focusing derivatives' efforts on customization, usually with clear target public in mind. Debian pushes most of those advantages (for derivatives) at their extremes.
The Debian archive is huge (about 30'000 packages with Squeeze), meaning that the amount of packaging work that can be reused is similarly huge.
Debian has a long history and a well established community, which reduces the risk of "betting" the efforts of derivatives on a distro that might disappear anytime soon.
Debian is well known for its obsession with package quality, meaning that the risk of trusting pre-existing packages is greatly reduced, and that package behavior is on average very dependable as well as documented in standard documents such as the Debian Policy
Debian is also well known for its obsession with licensing, at the point of being considered by many one of the authoritative sources for deciding what constitutes Free Software and what does not. For derivatives that implies that they do not need to worry---as long as they trust Debian, of course---neither about legal redistribution risks nor about software freedom claims on their distro.
But there are also a couple of "political" reasons for basing derivatives on Debian. One is quite subtle and applies mostly to commercial distributions. If you are designing one such commercial distro, you have to be based on an independent distro with no commercial interests, lest risking that petty (technical or otherwise) choices might be made just to undermine your business. Among "popular" GNU/Linux distros, Debian is essentially the only one which is both volunteer-based and not ascribable to any specific company.
The last reason why I think many derivatives are based on Debian is our vocation to universality. Since we are not targeting any specific public, but rather trying to provide an operating system suitable for several different use cases (e.g. the tasks supported by the Debian installer or the choice of Debian Pure Blends), we offer an ideal starting point for those who are going to have customization as their main business.
2.) What could be the benefits for Debian and for its Derivates, if the discussions about a second Debian Release based on the Testing branch really come to be reality?
We are considering the possibility of adding a new Debian suite which is "rolling" like Testing, in the sense that new software releases flow in regularly rather than only at periodic bumps corresponding to major distribution releases. There seem to be user demand for that and I personally believe that such a scheme could address the needs of both advanced desktop users and of developers, which are now strand to mix and match Debian suites ending up using package combinations which benefit, as a whole, of very little quality check. The main issues to face to get there is how to make sure that adding such a suite would not get in the way of preparing high quality Stable releases, in terms of quality control, developer focus, etc.
Note that such a hypothetical "rolling" suite needs does not necessarily need to be based on Debian Testing, although that is a good starting point. The reason why it's not a good endpoint is that Testing has been created (circa 2000), and is still used, as an internal tool to prepare the forthcoming Debian Stable release. As such, it is not entirely suitable for final user consumption. For instance, during the Debian "freeze" period the flow of new software releases that reach Testing is greatly reduced up to stopping completely just before the release of a new Debian Stable. I've been told that users of testing-based (derivative) rolling releases regularly complain about this aspect during Debian freezes, which is not surprising.
The main advantage for derivatives which are already offering a Debian-based "rolling" release will then be the possibility of basing their work on a suite less bound to the life cycle of Debian Stable than Testing is at present. On the converse side, that would also mean that derivatives will need to differentiate more from Debian than by just saying they are "rolling", which is a good thing: differences among distros are healthy and drive innovation. With those derivatives, on the other hand, which will recognize their goals as being aligned with Debian's goals---Debian Stable today, maybe Debian "rolling" tomorrow---we will be happy to join efforts.