I am yet to see any evidence to support this. On the contrary, there are plenty of articles that show the calculation is surprisingly accurate compared to real life (or at least, industry standard calculators such as BestBikeSplit).
New individual outdoor 24-hour record is not that dissimilar to the current Zwift record.
I think the issue is not the accuracy of solo ride speed. I find the issue to be the blob speed in Zwift. Somehow the riders at the front of the pack still receives a draft benefit in Zwift, leading to ridiculous speed differences between solo/small pack riders and a bigger blob. As a result, even if you output the same power as the guy who is pulling the pack, the pack ends up going faster than you do in Zwift. IRL, it takes coordination in the pack to catch up a breakaway, take turns and what not. None of this needs to happen in Zwift to catch a breakaway because of the “blob effect”.
OK, I will match your hundreds of races (at 1059 today).
Breakaways are reasonably common. Although it can depend on the race setup. For example my last two races were small 3R Classique x 5 laps. Both races I started in the A group with about a dozen other people, similar numbers in B and C. Within the first two laps in both events the race evolved into a break of 3-4 riders (mixed A/B), first chase about 8 (mixed A/B), then everyone else (small groups of C and D).
Both of the WTRL autocat test events, large (40ish) starts of mostly high B, some low A (C3) and both races the field split into the breakaway group (dozen plus) and the rest in the first or second lap.
I have seen numerous races where a strong rider will make an early jump at roughly 800-1000m to go, to get out front and win (if they don’t get caught.)
In Zwift the Blob (usually) wins. But not always.
You mean…like in real life?
Zwift’s physics, other than drafting which has some fundamental issues (pack position determined by client rather than server), are fairly good. In real life, W/kg is what determines speed on climbs, while W/CdA is more important on flats, where CdA is proportional to cross-sectional area, and cross-sectional area has a component from the bike (bike size depends on rider height) and the rider (where there’s height, but also a contribution from weight since width depends on weight). While it’s true people some people lie about their dimensions, it’s more realistic this way. And it’s also true some people obsess about weight, but it’s not really any different IRL, since real physics also depend on weight.