I’ve been wondering why this is not done. It seems so obvious, so there must be a good reason.
When doing covid testing, ask people questions like
- how many times did you get a bus this week
- or visit a hairdresser
- does your office/apartment have AC or natural ventilation
- do you share accommodation like in a barracks, retirement home, hostel
Then you or all this in MS excel and run a correlation. Of positives, how many people did X or Y or both. Of negatives, how many petiole did X or Y or both. The statistical functions tell you which patterns are important.
This instantly tells you the risk of each activity. It can guide people and authorities.
-
People could be given a target, normalised to the reproduction number - all your activities this week must add to less than 1.
-
Authorities can know the exact statistical risk of an activity. Exactly how big will be the effect of closing pubs, or hairdressers, or taking barman over 50 or of work? You can have as granular data as you like.
If you look up micromorts on Wikipedia it’s a similar idea.
All we know about covid risks, everything news channels and governments repeat in their echo chambers, is either extrapolated from other diseases, anecdotal, or speculation. We could easy grasp the true exact data, with as much granularity as needed, with no extra effort from what is done today.
This should be the headline news. Instead of “cases went up this week” we could know that “Cases among 40s-50s linked to AC ventilated pubs, where people stayed 1 hour, went up this week”
Why not?
Dunno why nobody has commented on that so far, but my best guess is a lack of legitimation for disclosing this kind of personal information. Then again, in case your location is tracked anyway, certain correlations like you suggested could be deduced, but finding an actual 1:1 correlation should prove difficult
In contact tracing, you are asked to give up a lot of personal information. That’s why it doesn’t work - people can’t be honest about their high risk activities, because it’s not anonymous.
This idea is an anonymous survey, for that reason. No personal information is gathered, but rich statistical information is gathered.
But how can you link each questionaire answer to either a positive or negative test, without identifying the individual? The technicalities would take some take to write out. Just know I’ve thought about it and there is (at least one) good way of doing it.