Michael, that’s not that unlikely. According to the ICL model, on the day the stay-at-home order as imposed, March 23, new infections had risen to about 280,000 per day. Only six days earlier it had been 100,000 per day. The stay-at-home order immediately reduced it to about 80,000 per day, and it has now (or had) fallen to below 10,000 per day.
https://mrc-ide.github.io/covid19estimates/#/details/United_Kingdom
Those are figures for the whole UK, but most of the actual infections were concentrated in London and a few other major cities:
https://www.ons.gov.uk/visualisations/dvc811/multimap/index.html
Remember that people get better (or die) and become non-infectious. When R is less than 1, by definition it means that people are becoming non-infectious faster than other people are becoming infectious. So lets say each infected person is infectious for six days. At 10,000 infections per day there are only 60,000 infectious people in the UK, and if R is less than 1 that number is reducing all the time. So it isn’t that hard to find 2000 people together and no one has had it, especially away from hot spots.
This also shows how easy it could be to snuff out if only the government had half a brain. Infection numbers went up fast because R was about 4, so if we can get R down to 1 / 4 = 0.25, infection numbers will go down as fast as they went up, plus just the few days that a person remains infected.
The government needs to explain what we need to do, and why we’re doing it. They have to say “see this number, we have to reduce it to zero, and the harder we all work on this the sooner we go back to normal”.
You’re right that the first stay-at-home order was wasted. When you try to eradicate pests, you have to eradicate all of them or you were wasting your time; they just breed back. The solution isn’t just to put up with them.