In local schools systems, we see the pattern that some schools will have a relatively large and dramatic outbreak. In one of these large outbreaks that make global news, fifteen staff are sick, 50 students are sick, 200 kids are in quarantine. https://twitter.com/a_haema/status/1344634423879340035

In a significantly larger number of schools (maybe it's power law distributed) you just have small clusters of 2 students, plus a teacher maybe, requiring the quarantine of a class but the spread is limited or maybe happened in the community anyway if attended a mutual event.

Suppose this is all happening randomly, due to the large k-value for covid (overdispersion/superspreading), and we can consider how we would predict this would look for a new variant.

The small clusters happen frequently in all schools. Small clusters with the new variant happen as well, at whatever background rate we have for the variant, but go unnoticed.

If a superspreading event were to occur in a network with the new variant, it would appear that the new variant just 'exploded' relative to all the other spread that is occurring. After that, the new variant is seeded all throughout that network, creating a founder effect.

Which we probably underestimate the effect of. Some cities don't have large outbreaks until weeks or months after similar "sister" cities for which you'd predict similar timelines.

The new variant is likely more transmissible (based on the increase in Kent, and the in vitro data especially) but our degree of confidence and magnitude estimates also need to fold in how weird we know covid is.