Cwa-documentation: [CWA-Documentation-Github] Risk assessment document doesn't contain the newest attenuation values

[hoehleatsu]

Where to find the issue

Tables 3, 4 and 5 of cwa-risk-assessement.md

Describe the issue

As of 3207f6a28a5f0f23c69fc77f81859cf85cd31adc in the cwa-server repository, the attenuation risk has been changed from 1 to 2. However, this is not yet reflected in the three tables of the cwa-risk-assessement.md risk scoring example.

Suggested change

Update the values of (1) in the Attenuation rows to (2) in tables 3, 4 and 5 of cwa-risk-assessement.md and see if this produces changes elsewhere in the example.

---

Internal Tracking ID: EXPOSUREAPP-2945

[daimpi]

Update the values of (1) in the Attenuation rows to (2) in tables 3, 4 and 5 of cwa-risk-assessement.md and see if this produces changes elsewhere in the example.

In the same tables the values for "Transmission risk" levels I & II should be adjusted from (1) & (2) to (0) as this has also recently been changed in https://github.com/corona-warn-app/cwa-server/pull/763/commits/0599ecc09dd75faeffbd6234c8fa6dcc6db47dcf

This change would also require some text adjustments in the example calculations.

[thomasaugsten]

This are changes to react on the issues with iOS 13.7 .

When we are done with all changes we will update the documentation. The changes of the values are not changing the risk assessment on non iOS 13.7 device. We set the attenuation risk from 1 to 2 but in the same we change the divisor from 25 to 50

https://github.com/corona-warn-app/cwa-server/pull/775/files

[hoehleatsu]

Thanks for the information, apparently related to the fix of corona-warn-app/cwa-app-ios/issues/1165

[hoehleatsu]

When updating the documenation you might also want to check, if the exposureSummary in v1 of the GAEN framework really adds up all encounters per day per key before computing the risk score. At least the newer versions of the framework, which contain code releases, appear to calculate the risk score per exposure window and then only adds up these afterwards - see this blog entry. As explained in the post, I have not been able to find the exact code for how it's done in v1.

If the per encounter/window calculation assumption is correct, then the 10 minute bus ride in the morning and evening would not pass the minimum risk score of 11, because the duration has to be > 10 minutes for a value of 1 to be given for the duration part of the risk score. With the current configuration of the app, it's 0 and, hence, each encounter - no matter the other values for days, TRL or distance - would have a risk score of 0 and is thus removed from the subequent calculations.

[MikeMcC399]

The release of CWA version 1.9.1 (see https://www.coronawarn.app/en/blog/2020-12-16-corona-warn-app-version-1-9/) which changes the GAEN API usage from v1 to v2 ( see https://www.coronawarn.app/en/blog/2020-12-17-risk-calculation-exposure-notification-framework-2-0/), means that some points raised in this issue are now obsolete, however the document https://github.com/corona-warn-app/cwa-documentation/blob/master/cwa-risk-assessment.md still needs to be revised.

If the document revision cannot be carried out in a timely fashion, then the document should be marked in the interim with a comment like "undergoing revision".