I think Tunnel is right on. Or more accurately, Renato is. As someone who trains mostly on hilly routes at high altitude, I can tell you it is misleading to take Mosop's run 4 weeks out from 2011 Boston, for example, do the math the way Canova does, and says it was _% of the pace he ended up at in Boston
The altitude and hills - and in many cases, dirt roads - combine to slow one down significantly compared to running at the same heart rate, blood lactate level, and subjective effort on a flat, sea level, paved road. There was a thread where Canova suggested that heavy trainers (this was pre- anything 'super', so he means as opposed to traditional road flats), hills, altitude, and dirt combined were worth 6 seconds per km or 10 per mile.
I won't look up any of those day-to-day threads right now, but what Tunnel suggests is exactly right - there are long runs close to 95% of goal pace about a month out, but maybe 92 or 93. Easily 95% of effort considering the informal conversion Renato believes to be reasonably accurate.