While I do like the emphasis "take human behavior into account" in pandemic planning (that's my bread-and-butter! and duh!), I'm a little wary of being able to incorporate it into predictive modeling, though, for two reasons.
-
-
First, modeling reflexivity (which is what social science folks call the thing being described here: people will respond to what's going on, which will change what's going on) in any way that's truly predictive is not just hard, may well be intractable in such situations.
2 replies 2 retweets 13 likes -
Two, this "must be behavior by elimination" a wee-bit dangerous. Could it be? Maybe. But it could also partly be things we don't—yet—understand. Did cases drop so precipitously in multiple countries with so many different cultures at the same time all due to behavior? I'm.. wary.pic.twitter.com/mslteOF6Pc
3 replies 4 retweets 22 likes -
Alternative: society is complex & interaction networks are power law distributed. Many highly connected nodes got infected/immune disrupting network transmission. *BUT* 1) that still means many ppl could still be susceptible, and 2) social networks change w time e.g. with seasons
1 reply 1 retweet 2 likes -
The question in the original article was why did epi models miss the relatively rapid, widespread drop? People all masked up/distanced in fear seems less plausible to me as the main explanatory variable given the magnitude, though I think it does matter somewhat.
2 replies 0 retweets 5 likes -
There is a “freak out threshold” when COVID-19 become the major cause of death, ~cancer & heart disease levels, the news media pays attention, lots of headlines, and there’s a broad increase in public health discipline. I had hoped the threshold would be much lower
1 reply 0 retweets 0 likes -
Of course. But I think that mattered less in winter 2021 because of the ideological polarization. Fear certainly has an influence but by January 2021, the interpretation of news was highly-fractured. But the drop is fairly uniform. I think there’s a puzzle there given magnitude.
1 reply 0 retweets 3 likes -
(I don’t have an answer if that’s not obvious).
1 reply 0 retweets 1 like -
For sure, lots to learn. Hard to really capture the complexities of social networks in transmission models
2 replies 0 retweets 0 likes -
I think better real-time transmission data could help, including digital techniques, but seems that there are many challenges there. The challenges are social, political, and legal as much as technical.
1 reply 0 retweets 0 likes
But more than that... Maybe these things are outside the powers of Laplace's demon. Maybe prediction in the traditional sense is not something we can do here—even with more data and expanded techniques.
-
-
Yes, I agree. Personally, I just wish for better quantitative risk/transmission data, in near real-time. Not necessarily for prediction, but for closed-loop, real-time, fine-tuning of policies. Probably a pipe dream, since policy makers would probably not take advantage of such.
0 replies 0 retweets 0 likesThanks. Twitter will use this to make your timeline better. UndoUndo
-
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.