If you want to talk about things with certainty, you need to appeal to science, where attempts are made to isolate, and measure the effects.
With respect to EPO, if you want to appeal to science, you will find very little that can be reliably extrapolated to elite athletes, or for any running distance much longer than 5K.
If we look to science for the answer, we see they have only studied "well trained", or "amateur" or even worse, "untrained" subjects, over a short period of a few weeks or months, over short distances, like 3K or 5K time trials, or even worse, on a stationary bicycle in a lab. Many studies measured VO2max, or time to exhaustion at sub-VO2max, rather than looking at real-world time trial performances, further insulating the study's results from real-world performance benefits. From these types of limited experiments, we cannot reliably conclude, for example, if EPO potentially enables "super human" results, like steroids can for power events.
If we want to think, as we see often in running tabloids, that the *only* benefit of EPO is increased oxygen delivery to the muscles, due to increased RBC, then logically it should favor VO2max duration events (which for running Jack Daniels defines indirectly as a continuous 10:58 effort), or shorter (like the 1500/mile). Longer/slower sub-VO2max events would not be limited by restricted maximal oxygen delivery. The marathon distance is further complicated by non-oxygen related factors like optimal glycogen management. One obvious anomaly, that suggests EPO must have other benefits, is attempting to explain why EPO would benefit a 3 week long cycling tour that last much longer than 11 minutes, with efforts much lower than VO2max efforts. If we still want to believe EPO benefits longer events, there must be other benefits to EPO, like increased recovery from hard training, and placebo or other psychological effects, and protective effects.
If we think further about why EPO benefited the "grand tour" cyclists, we shouldn't forget that the great cycling EPO experiment did not control for effects due to steroids, testosterone, HGH, blood transfusions, cortisone, and placebo. If we want to conclude, for example, that EPO helps recovery, then we need to account for, and factor out, improved recovery benefits from other sources, like steroids.
With respect to the marathon distance, any influence, positive or negative, due to EPO has never been reliably isolated, observed or measured. Renato is often crucified for his repeated insistence that it cannot help some narrow class of elite, altitude dwelling class of runners. But the fact is that today, any statement about anyone benefiting at the marathon distance from EPO, either in training, or in a race, is pure hypothesis, speculation, or faith. It remains to be observed, measured, in order to be confirmed or denied.
In my non-expert, non-educated opinion, any conclusion connected to EPO impacting performance, must carry with it a high degree of uncertainty, when straying too far from 11:00 3K runners, or 20 minute stationary cyclists in a lab.