> Practically speaking everything in the comic is still correct. The fundamental problem is that apps aren't sandboxed from each other, so once a malicious prog

> Practically speaking everything in the comic is still correct. The fundamental problem is that apps aren't sandboxed from each other, so once a malicious program gets on the machine it's game over. On Windows, perhaps. On Apple and some Linux configurations, it hasn’t been correct for years because they implement app level sandboxing. Windows Sandbox, App Containers, etc. show that the Windows team isn’t ignoring that even though their backwards compatibility problem is much harder. > For instance even if you use VBS to protect a service that can encrypt/decrypt on behalf of apps, all of that's useless when the system can't be assured which app is asking to decrypt the secrets. There’s no technical reason why it can’t verify the app - indeed Chrome just started to do this to prevent infostealers from grabbing cookies – but also note that it’s still an advantage if malware can’t get the private key because that’s protected by VBS. Things like what password managers have done for years to require a handshake or rate-limit help to shift from the attacker getting everything effortlessly to only getting some things they specifically request or the first n things they try.