100% agree with Smoove. That is a very nice workout but lower in volume (18:00 of quality running), which speaks well of 5k fitness but is a far cry from running 3:00:00 at a set pace.
There are speed-oriented runners and strength-oriented runners and neutral runners...you could take that workout ability and get 3 vastly different marathon outputs depending on the rest of a given runner’s training schedule and ability.
One way I like to look at it is this...you need at least 3 reliable data points to tell you that you are ready to run a marathon at a given pace.
1) A data point like the one mentioned (3 x 1mi at ~6:00/mi off 3:00-3:30 recovery jogs). This tells you that you have the basic speed and leg turnover to run sub 3.
2) A medium data point - something like 7 mile continuous tempo at 6:30-6:35/mi, or 2 x 5mi at 6:35-6:40/mi off a 5:00 jog recovery. This tells you that you have the basic strength to run a sub 3.
3) A long moderately hard run, like 17-18mi @ 7:15-7:30/mi. This tells you that you have the strength endurance to run sub 3.
Obviously the more data points in between you have, the more reliable the prediction. Which is why I favor the spectrum of paces approach by Canova. It is like having 12-15 different data points from the basic speed side of the spectrum to the strength endurance side of the spectrum.
Is there one single indicator workout that is a guarantee sub 3 predictor? Yes...1 x 26.2mi at 6:52/mi off zero recovery jogs. :)
But seriously, a single speed workout indicates nothing by itself. A single long endurance workout indicates nothing by itself. 10 workouts along the spectrum that, when plotted like an X/Y graph, line up like a soldier of data points all pointing to sub 3? Much more reliable.
But at the heart of it, 3 solid data points in the speed/strength/endurance zones is much more reliable than one data point, though not as reliable as 10 or more data points, obviously...