But if you have an external hardware MPG2 encoder and then burn to DVD, aren't you re-encoding again?
Not unless your DVD Authoring application is extremely stupid. There is a difference between encoding and muxing. When you import a DVD-compliant MPEG-2 file into most DVD Authoring applications I've used, it detects it and doesn't re-encode the video, it just muxes it into the proper BUP/IFO/VOB structure needed for DVD playback.
Also, a quality hardware MPEG-2 encoder will generally do a MUCH, MUCH better job at encoding the video than most cheap software encoders would. Now, if you spend $4k on your software encoder, then you might have a different story, but...
MiniDV camcorders certainly
do not transfer the video untouched (or even "mostly" untouched). MiniDV camcorders encode the video to DV, which is a lossy codec (a fairly high-bitrate one, but a fairly lossy one), and when you make the DVD then it is transcoded to MPEG-2. More damaging however is that using a DV --> DVD workflow also requires a color-space conversion, which causes pretty severe artifacts and loss of color fidelity and resolution. DVD uses the 4:2:0 color space, but DV uses 4:1:1. That means you end up with 1/8 of the original color "resolution" in the end when you go through an intermediary DV codec on the way to MPEG-2. Going directly to MPEG-2 at least preserves the color space end-to-end, so you're only limited to 1/4 of the "color resolution", and there are no rounding errors introduced (plus you don't introduce any unnecessary DV artifacts).
If you don't know what I mean about color spaces, do some research on video YUV color spaces. DV --> DVD is generally a no-no. HDV is a much better option, even if you are going to downsample it to Standard Def for DVD, simply because it doesn't go through a color space conversion then.
To put it in terms of audio files, which might be easier to understand... Going through a MiniDV camcorder is like doing this with an audio file:
Say you start out with an uncompressed 48kHz, stereo wav file, and your goal is to convert it to a 48 kHz, stereo MP3. However, instead of going directly to MP3, you first compress it to AAC at 44.1 kHz in mono, but you do it at 400 kbps (very high for an AAC). And only then do you take that AAC file and compress it to MP3, resampling it back to 48 kHz and "faking" the stereo signal. Even though the bitrate is very high on the AAC file, you've still thrown out all kinds of audio data in the process.
The same thing happens with running through a DV intermediary step. It does allow you to more easily edit the video though, as MPEG-2 is a terrible editing format (the NLEs are getting better at it now though because HDV is really just fancy High-Def MPEG-2).