* Getting through unhindered
* Getting turned away at the border
* Forcing better civil norms (courts / warrants / lawyers etc.)
* Not everywhere has the infrastructure necessary to upload large datasets to the cloud
* Most cloud providers are in not-great jurisdictions for some threat models.
* Lying to border authorities, even by omission, ends badly.
The Shatter Secrets approach is fail-closed, it's designed to trigger extra-jurisdictional oversight in the (likely) case you end up in a room
Which is where the success criteria comes from, we reduce harm by using technology to force the powerful to deny entry or comply to stronger jurisdictional standards.
That is a different dynamic.
A significant part of the research around this was/is focused on the *usability* of approaches. Safely interacting with, and choosing parameters for, cryptographic protocols is not an accessible skill for many.