5 ms·

Does it matter that much if there is a setting or not? Google already has your documents anyway. They're not being used for training either way.

yawpitch 2y ago
Possession of a file doesn’t imply that Google was reading that file, and absent a very limited set of criminal actions for which a search warrant might be issued there’s always been a reasonable expectation that Google was not reading your files… now Google is reading that file, apparently by default.
- fauigerzigerk 2y ago
  Not sure what you mean by "reading". Google has always been indexing all personal content to power the search feature (not public Google search). It's a major reason why I use their services.
  - yawpitch 2y ago
    Let’s take “reading” to mean “learning the semantic meaning of” then… an index is really just a data structure populated with the words and link back to the document, but no developed notion of the semantic meaning of those words or a semantic meaning like “odds this is the user’s SSN” associated with the content of the document, and it could easily be held entirely within your user account, entirely accessible only to you. This is much closer to a Google employee opening your document and reading it and emailing you back a summary as a default action, even if I’ll admit that employee is not generally intelligent. If the technical details make it so the “semantic index” is held only within your user account and only ever accessible by you and cannot under any circumstances leak outside to Google at large, then okay I might be comfortable with that, but I’d want to know those details first rather than just having this “feature” enabled on a file type.
barnabee 2y ago
It matters to me. I don’t use Google to create personal documents precisely because I can’t trust or verify that their apps aren’t doing more than provide storage, versioning, conflict resolution, and an editing UI without my permission. That said, I think it’s way past time to give up on hoping to trust a cloud / “big tech” provider. There are now only three models for applications I’m happy to use, and I try to do everything in one of those three where possible: 1) Verifiably end-to-end encrypted. This is the only option way I can get comfortable with cloud software. 2) Self hosted applications. 3) Local first applications with “dumb filestore” data syncing via a provider of my choice (likely self hosted). In all cases I have a strong preference for truly open source (avoiding VC funded open core weirdness etc.) and try to donate more than the equivalent subscription or purchase price for proprietary alternatives to the projects I find that meet my needs.

yawpitch 2y ago

Possession of a file doesn’t imply that Google was reading that file, and absent a very limited set of criminal actions for which a search warrant might be issued there’s always been a reasonable expectation that Google was not reading your files… now Google is reading that file, apparently by default.

fauigerzigerk 2y ago
Not sure what you mean by "reading". Google has always been indexing all personal content to power the search feature (not public Google search). It's a major reason why I use their services.
- yawpitch 2y ago
  Let’s take “reading” to mean “learning the semantic meaning of” then… an index is really just a data structure populated with the words and link back to the document, but no developed notion of the semantic meaning of those words or a semantic meaning like “odds this is the user’s SSN” associated with the content of the document, and it could easily be held entirely within your user account, entirely accessible only to you. This is much closer to a Google employee opening your document and reading it and emailing you back a summary as a default action, even if I’ll admit that employee is not generally intelligent. If the technical details make it so the “semantic index” is held only within your user account and only ever accessible by you and cannot under any circumstances leak outside to Google at large, then okay I might be comfortable with that, but I’d want to know those details first rather than just having this “feature” enabled on a file type.

barnabee 2y ago

It matters to me. I don’t use Google to create personal documents precisely because I can’t trust or verify that their apps aren’t doing more than provide storage, versioning, conflict resolution, and an editing UI without my permission. That said, I think it’s way past time to give up on hoping to trust a cloud / “big tech” provider. There are now only three models for applications I’m happy to use, and I try to do everything in one of those three where possible: 1) Verifiably end-to-end encrypted. This is the only option way I can get comfortable with cloud software. 2) Self hosted applications. 3) Local first applications with “dumb filestore” data syncing via a provider of my choice (likely self hosted). In all cases I have a strong preference for truly open source (avoiding VC funded open core weirdness etc.) and try to donate more than the equivalent subscription or purchase price for proprietary alternatives to the projects I find that meet my needs.