Based on this, I expect this coming Nov/Dec/Jan/Feb to be way worse again around SF. Except the baseline is much higher now, so I expect the elevated case rate to likewise be much higher
I’d like to register a prediction that in the Bay Area and in CA, rates will be drastically below where they are today in a couple of months.
The reason I think that is that everywhere else, we’ve seen delta skyrocket, and then collapse once the control system kicks in and herd immunity thresholds are reached as vulnerable people get infected. I think the same thing will happen here.
So contra the advice that this is the best time to do things, I think it’s worth waiting. If I’m right we should see evidence in the next few weeks.
Could you quantify ‘in a couple of months’? I don’t have a strong view on exactly what Delta will look like between now and November—I could pretty easily imagine it being about the same on average, slightly lower, slightly higher, or a decent amount higher, if no major surprises (e.g., a new strain displacing Delta) happen.
Off the cuff: In Alameda County, I’d maybe put 2⁄3 probability on at least 85% of weeks being between 5000 official new cases per day and 80,000 official new cases per day, until November?
And I’d separately predict with 70% probability the conjunction: (1) the average number of new Alameda County cases each day between Dec 15 and Jan 15 will be at least 1.5x the average number of cases in the last week of September; and (2) this Dec 15 - Jan 15 average will be at least 50,000 cases per day.
I wouldn’t be at all surprised if there’s a rise, followed by a collapse, followed by another (seasonally mediated) rise. I would be surprised if the collapse takes us far below 5000 official new cases per day—I’d more expect the collapse to take us approximately back to present levels, or somewhat higher than present levels? But I don’t feel super confident of any of this (hence my giving probabilities like 2⁄3).
In Alameda County, I’d maybe put 2⁄3 probability on at least 85% of weeks being between 5000 official new cases per day and 80,000 official new cases per day, until November?
Oops! I said “Alameda County” here, but I think I must have been thinking of CA, since I remember setting my lower bound (5000) at somewhat below case count at the time (in the chart I was looking at), whereas 5000 would have been massively higher than anything we’ve ever seen for Alameda. Hence my saying:
I would be surprised if the collapse takes us far below 5000 official new cases per day—I’d more expect the collapse to take us approximately back to present levels, or somewhat higher than present levels
(There were ~400 new confirmed cases in Alameda County on the day I posted this comment, vs. 11,000 new confirmed CA-wide cases.)
Likewise, the second prediction doesn’t make sense for Alameda County.
I could pretty easily imagine it being about the same on average, slightly lower, slightly higher, or a decent amount higher, if no major surprises (e.g., a new strain displacing Delta) happen.
A new strain did replace Delta.
In [California], I’d maybe put 2⁄3 probability on at least 85% of weeks being between 5000 official new cases per day and 80,000 official new cases per day, until November?
Looking at each Monday from Aug 23 through Nov 1 for what the NY Times says the weekly average was that day, I get:
Aug: 16k, 14k
Sep: 13k, 11k, 9k, 8k
Oct: 7k, 6k, 6k, 4k, 9k
By that accounting, 10⁄11 = ~91% of weeks were in the 5000 − 80,000 range.
And I’d separately predict with 70% probability the conjunction: (1) the average number of new Alameda County cases each day between Dec 15 and Jan 15 will be at least 1.5x the average number of cases in the last week of September
Average in the last week of September was ~7241. 1.5x that is ~10862. On Dec 15 new cases were only 5365, so my prediction is falsified. Dec 20 cases jumped from 3582 to 19084, so CA cases have been in the range I mentioned starting Dec 20 rather than Dec 15. (And I expected a climb to start in early November, like it did in 2020, rather than in late December—I mentioned ‘Dec 15’ because I expected this to give me more than enough buffer for seasonality effects to take over.)
and (2) this Dec 15 - Jan 15 average will be at least 50,000 cases per day.
Eyeballing the NYT graph, looks to me like the average was ~30k at the end of Dec, 50-60k first week of Jan, and 100k+ for most of Jan. (And again, with Omicron rather than Delta.)
Off the cuff: In Alameda County, I’d maybe put 2⁄3 probability on at least 85% of weeks being between 5000 official new cases per day and 80,000 official new cases per day, until November?
That’s a good way to think about it. My 2/3rds confident window looks something like 1.5k-25k per day for most weeks until November. So substantially lower than yours.
For example, if we had the same infection rate as the UK (cases are down from delta peak) it would be 500 cases/day in Alameda county. The UK might be trending back up now, but the slope is much lower. Their vaccination rate is a little higher than Alameda county but our vaccines are better, so that’s probably ~ a wash, or maybe slightly in the UK’s favor.
this Dec 15 - Jan 15 average will be at least 50,000 cases per day
In other words, you think the average for that month will be equivalent to the worst week last winter before anyone was vaccinated? Definitely seems high to me. Delta is much worse, yes, but vaccines really do help quite a bit, I think.
I buy that seasonality is a thing for covid, but I think it will be more like 15kish rather than 50k for that month.
That seems reasonable to me, and updates me some toward your view. I feel confused about how rapidly cases fell in the UK. If this has something to do with levels of natural immunity in the UK, I could imagine that meaning the Bay Area has less protection. (~9.7% of the UK has been confirmed-infected, vs. I think ~6.3% of Alameda County.)
I’m also curious how ‘lockdown culture’ differs between Alameda County and the UK—Berkeley has seemed pretty COVID-cautious to me (which makes me think the control system could be more trigger-happy, resulting in fewer cases), but I’m not sure how to compare that to the UK.
I’d like to register a prediction that in the Bay Area and in CA, rates will be drastically below where they are today in a couple of months.
The reason I think that is that everywhere else, we’ve seen delta skyrocket, and then collapse once the control system kicks in and herd immunity thresholds are reached as vulnerable people get infected. I think the same thing will happen here.
So contra the advice that this is the best time to do things, I think it’s worth waiting. If I’m right we should see evidence in the next few weeks.
Could you quantify ‘in a couple of months’? I don’t have a strong view on exactly what Delta will look like between now and November—I could pretty easily imagine it being about the same on average, slightly lower, slightly higher, or a decent amount higher, if no major surprises (e.g., a new strain displacing Delta) happen.
Off the cuff: In Alameda County, I’d maybe put 2⁄3 probability on at least 85% of weeks being between 5000 official new cases per day and 80,000 official new cases per day, until November?
And I’d separately predict with 70% probability the conjunction: (1) the average number of new Alameda County cases each day between Dec 15 and Jan 15 will be at least 1.5x the average number of cases in the last week of September; and (2) this Dec 15 - Jan 15 average will be at least 50,000 cases per day.
I wouldn’t be at all surprised if there’s a rise, followed by a collapse, followed by another (seasonally mediated) rise. I would be surprised if the collapse takes us far below 5000 official new cases per day—I’d more expect the collapse to take us approximately back to present levels, or somewhat higher than present levels? But I don’t feel super confident of any of this (hence my giving probabilities like 2⁄3).
Oops! I said “Alameda County” here, but I think I must have been thinking of CA, since I remember setting my lower bound (5000) at somewhat below case count at the time (in the chart I was looking at), whereas 5000 would have been massively higher than anything we’ve ever seen for Alameda. Hence my saying:
(There were ~400 new confirmed cases in Alameda County on the day I posted this comment, vs. 11,000 new confirmed CA-wide cases.)
Likewise, the second prediction doesn’t make sense for Alameda County.
Replacing ‘Alameda County’ with ‘California’:
A new strain did replace Delta.
Looking at each Monday from Aug 23 through Nov 1 for what the NY Times says the weekly average was that day, I get:
Aug: 16k, 14k
Sep: 13k, 11k, 9k, 8k
Oct: 7k, 6k, 6k, 4k, 9k
By that accounting, 10⁄11 = ~91% of weeks were in the 5000 − 80,000 range.
Average in the last week of September was ~7241. 1.5x that is ~10862. On Dec 15 new cases were only 5365, so my prediction is falsified. Dec 20 cases jumped from 3582 to 19084, so CA cases have been in the range I mentioned starting Dec 20 rather than Dec 15. (And I expected a climb to start in early November, like it did in 2020, rather than in late December—I mentioned ‘Dec 15’ because I expected this to give me more than enough buffer for seasonality effects to take over.)
Eyeballing the NYT graph, looks to me like the average was ~30k at the end of Dec, 50-60k first week of Jan, and 100k+ for most of Jan. (And again, with Omicron rather than Delta.)
That’s a good way to think about it. My 2/3rds confident window looks something like 1.5k-25k per day for most weeks until November. So substantially lower than yours.
For example, if we had the same infection rate as the UK (cases are down from delta peak) it would be 500 cases/day in Alameda county. The UK might be trending back up now, but the slope is much lower. Their vaccination rate is a little higher than Alameda county but our vaccines are better, so that’s probably ~ a wash, or maybe slightly in the UK’s favor.
In other words, you think the average for that month will be equivalent to the worst week last winter before anyone was vaccinated? Definitely seems high to me. Delta is much worse, yes, but vaccines really do help quite a bit, I think.
I buy that seasonality is a thing for covid, but I think it will be more like 15kish rather than 50k for that month.
That seems reasonable to me, and updates me some toward your view. I feel confused about how rapidly cases fell in the UK. If this has something to do with levels of natural immunity in the UK, I could imagine that meaning the Bay Area has less protection. (~9.7% of the UK has been confirmed-infected, vs. I think ~6.3% of Alameda County.)
I’m also curious how ‘lockdown culture’ differs between Alameda County and the UK—Berkeley has seemed pretty COVID-cautious to me (which makes me think the control system could be more trigger-happy, resulting in fewer cases), but I’m not sure how to compare that to the UK.