Does it matter? If the estimate produces an insignificant error, it doesn't matter. If it matters totally depends on the degree of precision you wish to obtain. Since the subject is climate science, a one percent error is roughly equal to the impact of a doubling of CO2, so one percent error is my initial "baseline" for significance.
So what would this mean? That TSI would need an ~70 Wm-2 (65 to 67 Wm-2 appears to be that actual value) DC component due to annual orbit variation. That DC component is nearly 50% of the "Greenhouse Effect" impact and has absolutely nothing to do with atmospheric gases, only orbital variations.
I could of course be wrong, but averaging averages is prone to produce errors. Since the solar "constant" has been revised downward and the estimated "average" surface temperature based on the 1951-1980 also revised downward, it would be very easy to confuse the causes and effects.