3 ms·

> all remain obsessed with case numbers and not actual hospitalizations and death It's a balance. E.g. Germanys current trouble is a direct consequence of offi

by detaro 5y ago

> all remain obsessed with case numbers and not actual hospitalizations and death It's a balance. E.g. Germanys current trouble is a direct consequence of official policy that had a core of "we only care about hospitalizations now", but ignored "... and based on case numbers and other statistics, what will future hospitalizations look like if we don't change anything about our approach" - not helped by election campaigns and politicians publicly admitting they haven't understood this whole "exponential" malarky yet. You really need to consider more than just one number.

formerly_proven 5y ago
The superseded government (by way of Jens Spahn) admitted that it held back on acting w.r.t. COVID in 2021 to avoid negative impact on their campaign. General inaction or adverse action in other areas (e.g. ICU staffing, leerdenker) did not help, either.
hef19898 5y ago
As someone who has quite some experience in running operations based on data, numbers and metrics the German approach to COVID on that front alone really puzzled me from the beginning. First the incidence of 50 that they pulled out of their ass for contact tracing, which was ignored basically the next day. Then the fact that resources weren't really increased to improve contract traceability. Then the decision to use a backward looking KPI, ICU occupation, to tackle a highly dynamic situation. Which also meant to ignore the incidence, the only real predictive metric we have. And it is good at predicting ICU occupation and severe cases when combined with vaccination levels. That's simply amateurish, and means all measures, not all of which are bad, are usually a couple of weeks late. Even better, the 7-day average incidence is based on the daily numbers. Those numbers are, what a surprise, limited to the testing capacities of laboratories, so the incidence is reported based on the day the test was analyzed, not when the sample was taken. This flattens the 7-day average curve. Well, until the new numbers are reported once labs git through their backlog. Thing is, the 7-day average is not changed once published. Again just amateurish.
- detaro 5y ago
  Yes, it's been really frustrating to see RKI predict this for weeks and just ... nothing happening. And then you still get headlines of "incidence fell slightly!" - from 392,4 to 392,2 (or sth like that), when testing labs are hitting capacity. We somehow lucked into a relatively good first wave and ever since then... so much unnecessary mess.
  - hef19898 5y ago
    The drop in incidence is the same BS when reporting in DAX or S&P moving by 0.2 % day-to-day. That's noice, no signal. EDIT: That's also noise for HF traders, because day-to-day numbers are pointless when you worry about milliseconds.
- iSnow 5y ago
  German government staff is as un-digital as you can get. They have no idea how a small modeling group would have helped. Maybe the incoming government will do better, but I am not holding my breath.
  - detaro 5y ago
    Access to modeling isn't even the issue, given RKI, Charité, ... Asking those resources the right questions and turning their results into policy and communicating it is the problem, which is even more fundamental "that's what a government is for" skills...
  - funcDropShadow 5y ago
    Being digital doesn't necessarily improve decision make. All the necessary data was there. RKI the German equivalent to the CDC in the US predicted the current wave rather accurately in summer, but politicians ignored it, because they wanted to win the election.
  - hef19898 5y ago
    It's not modelling, just having a basic understanding of your metrics and the ability to choose the right ones. German officials showed none of that. Well, I'm not holding my breath on that neither.

by detaro 5y ago

formerly_proven 5y ago

The superseded government (by way of Jens Spahn) admitted that it held back on acting w.r.t. COVID in 2021 to avoid negative impact on their campaign. General inaction or adverse action in other areas (e.g. ICU staffing, leerdenker) did not help, either.

hef19898 5y ago

As someone who has quite some experience in running operations based on data, numbers and metrics the German approach to COVID on that front alone really puzzled me from the beginning. First the incidence of 50 that they pulled out of their ass for contact tracing, which was ignored basically the next day. Then the fact that resources weren't really increased to improve contract traceability. Then the decision to use a backward looking KPI, ICU occupation, to tackle a highly dynamic situation. Which also meant to ignore the incidence, the only real predictive metric we have. And it is good at predicting ICU occupation and severe cases when combined with vaccination levels. That's simply amateurish, and means all measures, not all of which are bad, are usually a couple of weeks late. Even better, the 7-day average incidence is based on the daily numbers. Those numbers are, what a surprise, limited to the testing capacities of laboratories, so the incidence is reported based on the day the test was analyzed, not when the sample was taken. This flattens the 7-day average curve. Well, until the new numbers are reported once labs git through their backlog. Thing is, the 7-day average is not changed once published. Again just amateurish.

detaro 5y ago
Yes, it's been really frustrating to see RKI predict this for weeks and just ... nothing happening. And then you still get headlines of "incidence fell slightly!" - from 392,4 to 392,2 (or sth like that), when testing labs are hitting capacity. We somehow lucked into a relatively good first wave and ever since then... so much unnecessary mess.
- hef19898 5y ago
  The drop in incidence is the same BS when reporting in DAX or S&P moving by 0.2 % day-to-day. That's noice, no signal. EDIT: That's also noise for HF traders, because day-to-day numbers are pointless when you worry about milliseconds.
iSnow 5y ago
German government staff is as un-digital as you can get. They have no idea how a small modeling group would have helped. Maybe the incoming government will do better, but I am not holding my breath.
- detaro 5y ago
  Access to modeling isn't even the issue, given RKI, Charité, ... Asking those resources the right questions and turning their results into policy and communicating it is the problem, which is even more fundamental "that's what a government is for" skills...
- funcDropShadow 5y ago
  Being digital doesn't necessarily improve decision make. All the necessary data was there. RKI the German equivalent to the CDC in the US predicted the current wave rather accurately in summer, but politicians ignored it, because they wanted to win the election.
- hef19898 5y ago
  It's not modelling, just having a basic understanding of your metrics and the ability to choose the right ones. German officials showed none of that. Well, I'm not holding my breath on that neither.