This (pack dynamics/speed) has been an interesting topic of late. Search the forums for pack dynamics 4.1 and 4.1.1 threads, if you’re interested (do so at your own risk 
).
Currently, we are in PD 4.1.1. What the case was before was that, due to a quirk in Zwift physics, we had the “churn” or “washing machine” effect. Essentially, this means that if everyone is putting out roughly equal power, riders not in the front row would be able to get a boost from the draft and gradually come shooting out the front. Once they hit the wind, their speed slows, but they are now providing draft to the people behind them, who come shooting through, who now provide draft… and so on. This drives up pack speeds, without anyone really having to do any extra work. Thus breakaways were very doomed unless they were really good. Even low-power packs could overtake single riders fairly easily.
The goal of PD 4.1 was to reduce this churn, which it did. This slowed down pack speeds, giving breakaways a chance of success. But this made things feel “unnatural” for many (though it theoretically was more consistent with the IRL idea of not being able to physically ride through the person in front of you). So Zwift chose to find a “middle ground” with PD 4.1.1, though some argue it is very similar to PD 4.1. Zwift Insider’s tests report that it is about middle ground, but from what I’ve seen people are experiencing a lot of the churn now.
I personally haven’t been riding or racing on Zwift a lot the past few weeks so my understanding may not be perfect, but this is what I think is the case.
For further reading and potential banging your head against the wall (plz don’t 
):