A Case In Point For Missing The Boat (So Keep Up The Interstate Sledging All Of You Princesses)
Victoria has recorded 191 new cases of coronavirus overnight, the largest single-day increase so far in the state. Of the new cases, 37 are linked to known outbreaks and the remaining 154 are under investigation — meaning health officials are still tracing the source of the infection. The record number of new infections brings the number of active cases of COVID-19 to 772.
The possibility of further lockdowns is a real prospect, with Premier Daniel Andrews previously warning more postcodes could have restrictions reimposed on them if the virus continues to spread. Currently, 12 postcodes covering 39 Melbourne suburbs are under stage three stay-at-home orders.
https://www.abc.net.au/news/2020-07-07/victoria-coronavirus-cases-rise/12428704
What case in point? Try this.
Early on, we already recognised the doubling time was around 3 days. That puts growth rate per day at 1.26, meaning the number of new infections is 0.26 or approximately 1/4 of the total infections. And that’s only for KNOWN cases.
Consistency check: 191 new cases; active cases 772. Not all of those are the recent outbreak, so perhaps 1% less, like maybe 764.28. As you all know, 191 * 4 = 764 so we’re not just in the order-of-magnitude-engineering-ballpark with the estimate here, we’re a fucking albatross in the cup.
How long does it take to resolve a case that is “under investigation”? 1 working day? 3 infection days over a weekend because obviously viruses don’t take holidays? 1 week which is just over 6 days? Let’s call it 3 to 6 days, which is 1 to 2 doubling periods. That’s fucked. Instead of ringfencing “all possible contacts”, should we stop movement only where the case currently is, and wait for completion of investigation to find the other contacts elsewhere? That could take 2 doubling periods. That means 4 times as many cases. 764 * 4 = 3056 in case we forgot.
“That’s just theoretical!” Fine, bring in some more pseudoreality then. Let’s say we just work with the 154 “under investigation” cases, assuming that the known outbreaks have been controlled (yeah right, look at human behaviour in all our favourite countries). The trouble is that “under investigation” means “no fucking idea”. People could be incubating virus for up to 2 weeks (more if we’re talking the tail 2.5% of cases). In a better-case scenario, that includes the investigation time. In an abject failure, that’s an additional 4, almost 5 doubling periods because of UNKNOWN cases.
So we’re talking minimum 4 to 5 doubling periods (multiply by 16 to 32), or 6 to 7 (multiply by 64 to 128) if the authorities are getting it wrong. We put that at 764 * 138 = 97792 cases in Victoria before the end of July. It really will be New South China. Daily counts as follows.
| 20200707 | 150 |
| 20200708 | 189 |
| 20200709 | 238 |
| 20200710 | 300 |
| 20200711 | 378 |
| 20200712 | 476 |
| 20200713 | 600 |
| 20200714 | 756 |
| 20200715 | 952 |
| 20200716 | 1200 |
| 20200717 | 1512 |
| 20200718 | 1905 |
| 20200719 | 2400 |
| 20200720 | 3024 |
| 20200721 | 3810 |
| 20200722 | 4800 |
| 20200723 | 6048 |
| 20200724 | 7620 |
| 20200725 | 9600 |
| 20200726 | 12095 |
| 20200727 | 15239 |
| 20200728 | 19200 |
Total, including 764 of the latest outbreak, 93256. Congratulations! You’ve saved 97792 – 93256 = 4536 people from getting sick by rounding 154 to 150 and linking 37 cases to known outbreaks!
Tell us we’re wrong.