However with FTF, any technology Google might potentially use in the future, they must file a patent for - this compression tech, or one of the DNN techs they've recently developed. Otherwise they could start using the technology and another company could copy the technology, file a patent, and be granted the patent because of FTF. They could then pursue Google for patent infringement.
In FTI they could develop and use tech without patenting it. in FTF they have to patent it because if they don't they'll lose the ability to use the tech to the first copycat who files.
FTF is just continuing the trend in the US of making it harder and harder for IP to be in the public domain - moving more towards the privatization of IP.
(To the extent that a bad actor can “steal” an invention, FTF vs FTI is irrelevant.)
A reference “is ‘publicly accessible’ upon a satisfactory showing that [the reference] has been disseminated or otherwise made available to the extent that persons interested and ordinarily skilled in the subject matter or art exercising reasonable diligence, can locate it.” Bruckelmyer v. Ground Heaters, Inc., 445 F.3d 1374, 1378 (Fed. Cir. 2006).
For instance, in Electronic Frontier Foundation, the Board determined that when a reference was only temporarily publicly available via a URL but no longer exists at that URL, the petitioner must show that the reference can be found through another source not including the URL. Electronic Frontier Foundation v. Personal Audio, LLC, IPR2014-00070, Paper 21 at 22 (PTAB Apr. 18, 2014) (Decision, Institution of Inter Partes Review).
[Note that the information being up somewhere where most people wouldn't be able to find easily enough doesn't suffice. "Reasonable diligence" isn't defined, but clearly isn't intended as a high bar.]
Printed publications can be an exception - but both those words are highly subject to interpretation.
How well known something must be is obviously going to be subject to interpretation. If you tell your co-worker at the bar, that’s probably not sufficient. If you publish on your blog that no one knows about, that’s probably also not sufficient. If you publish in a journal, that probably is more than sufficient. If you cannot reasonably claim that the other inventor should have been able to find your prior art, then the courts will rule against you.