I've been looking closely at the route, re-plotting it (in RWGPS) to reduce the inaccuracies due to canyon effect (it looks like someone's recorded track, but subsequently simplified, so a long way off in places). A couple of illuminating facts fell out of this, which will become apparent to all finishers, when they finish (I won't spoil the surprise

then again, RWGPS could be wrong). I've also written a routesheet (audax-stylee), although without riding it to check it then it's worth not much, except to give an indication of what's coming up.
One of the things I did was to try to work out the surface on the ground and tot them all up to give an indication of what to expect. Road is pretty clear; hike-a-bike is also pretty obvious. However, gravel vs. singletrack isn't always cut'n'dried — what's gravel to one might be ST to another, and StreetView satellites haven't spied on Wales for a while. I think I've probably erred a couple of field-crossings too far to gravel, but we'll find out when we get there. Ultimately, though, this is what I think it looks like:
The longest road section is the descent to the finish at 15.8km. The longest gravel section is that final climb to Dylife at 14.8km. The longest singletrack is one of the forestry sections at 2.3km. And the longest hike-a-bike is the Carnau traverse after nightfall, which is 4.6km,
IF you manage to keep to the trail in the dark — thanks, Stu
There are nearly 70 surface changes, so it's not like you get all of one surface in one go.
I'm not particularly looking forward to either the push up nor the descent down to the reservoirs — that just looks insane. In fact there are several descents around Elan valley that look tricky just in terms of off-road gradient at >10% and a couple >20%!
