Here is the excerpt from Edge :
But latency is what gets the most attention. And although it has already proven to be more than playable, [Madj Bakar, VP of Engineering] expects further improvements. "In the end, we believe that in a year or two we will have games that run faster and feel more responsive in the cloud as they do locally, no matter how powerful the local machine is," he argues. These improvements come via a term that sounds quite slippery. "Negative latency" is a concept by which Stadia can set a game with a buffer with the predicted latency between the server and the player and then use different methods to suppress it. It can run the game at a super-fast frame rate so that it can act on the player's inputs earlier, or it can predict a player's push of a button. These tricks can help the game feel more responsive, possibly more than a console game run locally at 30 fps with a wireless controller.
Now, moniker "negative latency" definitely sounds far-reaching. But there may be something in the idea of combining high frame rates with a buffer. And with Google's experience with AI and machine learning, it is undeniable that Stadia could predict button presses.
Bakers makes a very specific comparison between cloud-based games versus consoles that run at 30 frames per second and use a wireless controller. So it's obviously a stretch that this will be an apple-to-apple use case in the real world. Also, in general, players will raise their eyebrows at the concept of game streaming until we can test it in nature. Fortunately, it's only a month away, as Google Stadia will launch sometime in November.