I'll just throw in there, that these kind of % off numbers feel very old school..... anytime you have to do actual math in dnd beyond adding or subtracting a few numbers you should question if that is the best mechanic.
Aka....is there just a better way to put someone off track than having to pull out a calculator and actually determine the distance between A and B?
Well, like my question, this is the same thing the current 5E book does for determining how far off you are. (Just looked it up after Morrus's comment.)
In particular, this heavily weights the distance "off" you are towards the low end. On a d100, a 19 and a 91 both give you a 9% distance off. 50% of your rolls will give a total of under 25% off from your target distance. 80% of your rolls will be off by 50% or less.
Now, you could just take a single d10 roll and multiply that by 10% to get the distance off. That's be much simpler. The fact that it's now a ritual means repeating the spell isn't a huge ordeal of having to wait to the next day for another chance to cast a 7th level spell. Since the original wasn't a ritual, weighting any miscasts such that you're most likely to be closer than not makes sense.
On the other
other hand, a minimum of 10% off from a 1000 mile teleport means a minimum of 100 miles. Using 2d10s, though, makes the minimum distance 10 miles.
Eh, mathematically, I like the d10*d10 method, but it can certainly be annoying to calculate. Roll a 2 and 7, and try to get 14% of a 900 mile teleport, and you're generally not going to figure it out without a calculator. That's an extra bit of gameplay friction that probably doesn't need to be there.
If I were building it from scratch, I'd probably make it a 1d4 roll. 1 = somewhere within 10% of your distance to the goal; 2 = somewhere within 25% of your distance to the goal; 3 = somewhere within 50% of your distance to the goal; 4 = completely random. Then I'd leave the exact positioning up to the GM (who may use the d8 for direction). Those percentages are roughly equivalent to the chances of those ranges using the d10*d10 method.
