The proposed design makes no mention of using TPM/device attestation and explicitly calls out that it is bound to the same restrictions that Chrome's removal of third-party cookies is introducing.
If a site (Google or other) cannot get an attestation with the key and keys are unique per eTLD then there's no new tracking vector introduced. A site can't get anything more than they can with a standard, first-party session cookie.
> just force log me off of sites after a time limit
Since this opinion is seriously in the minority as the overwhelming majority of people do not like the friction of having to log back into services, that's never going to happen for any sizable fraction of websites [†].
Thus, for everyone's security it would be better if sites adopt DBSC to prevent session-theft/cookie-theft without introducing the nuisance of constantly logging in.
Not to mention, every log-in with a password is opportunity for phishing/key-logging. If people had to log in 10-100x more frequently, that would only increase another form of malware. Only FIDO/Passkeys offer the ability to make frequent log-ins phishing resistant.
[†] - banks being a big exception since, prior to DBSC, there was no good way to prevent session-theft/cookie-theft outside of making sessions/cookies very short-lived. And they need to be worried about session-theft/cookie-theft because... money.