The preprint that was cited has a lot of issues. My biggest issue is the complete lack of data presented. Yes, there are numbers but actual data relating to the tests, how many were IgG, IgM, and IgG and IgM positive are missing, plus, the most important part, how many had previously shown symptoms. The information from this was collected but not reported in the manuscript. There were also two false positives (0.5%) in the negative control samples. Another important factor is where were the positives identified. Were they geographically clustered or evenly distributed. If they're clustered it will invalidate the projection to an entire population. This is the sort of science I hate; first past the post. It's very typical of Stanford.
This preprint also demonstrates the importance of peer review of which I do a considerable amount. If this came across my desk for review I would haul it over the coals. The conclusions a very week and over interpreted from the small sample size (3,330), poor data analysis and oversimplified population modelling. Sadly, the scientific rigor of a lot of studies during this pandemic has been allowed to slide because some folk are desperate to get their work published first.
To a final point, if this paper does hold and there are 80k cases then this further reinforces why SIP orders are needed not the opposite.