Comment by kbenson | Hacker News Reader

Comment by kbenson | Hacker News Reader

I can't help but feel that most these companies working on these technologies are viewing them in isolation, when none of our peripherals are ever used in isolation. The mouse wasn't developed to replace the keyboard, but to supplement it. You don't generally use a mouse to select characters from a representation of a keyboard on screen (there are reasons when you want this, but it's not the general use case), so why do we always see the equivalent of that in new peripherals?

The two main cases I'm thinking of are eye tracking, and brainwave tracking (like EMOTIV). Individually, neither looks like a compelling way to control a computer when you have a mouse and keyboard, but together, I think that might really yield something interesting (and sooner!).

Instead of using brain waves to move a cursor around the screen, use eye tracking. Instead of using gaze lingering for click, use brainwave tracking. Individually they seem cumbersome and annoying to use. Together they could be a really compelling interface device, IMO.

The point of eye tracking in VR isn't really about being an input control (even if that's what this company was focusing on).

There are many big problems in VR that get solved if you know where the user is looking at. The most important one is, your eyes really only see high resolution in a very small region around what you're staring at, and if the VR headset can only really render that region in high quality, this solves many issues (processing power, framerate, even wireless transmission).

Of course, for any of this to work you need perfect eye tracking, which is very hard.

Either that, or most of the companies know that the technologies they are building are likely to be acquired to build part of the feature set of a larger offering, and are building them for precisely that reason. Too cynical?

I know a person that built a company for acquisition like you mention. Don't think they were ever successful. To be an acquisition target is one thing. But to be acquired by companies like Google most people would have to meet their hiring bar (aka passing their interviews) and have the right composition for the team (mostly engineers they would have a hard time hiring away). Of course there are exceptions, like they're solely purchasing for the technology, but that means the tech they built was pretty good and couldn't easily be cloned (of course this excludes if they're buying for market share of users if the product is super popular). My opinion is incredibly risky building for acquisition. Anyone have examples that it isn't?

Why is that cynical? That seems like a very reasonable thing to do if you want to start a company with very narrow domain knowledge.

This operates on the assumption that eye tracking and brain waves signals are positively or negatively correlated (for which the setup you propose can be thought of as an estimator with control variates). Is this indeed the case?

I'm not sure what information you're working off that makes you think they need to be positively correlated. I'm referring to using two separate systems in conjunction. Eye tracking to cursor placement (and anything else that it can handle well), and brainwave tracking to handle intent. At the simplest level, eye tracking for mouse x/y coordinates, and brainwave tracking for mouse button presses.

Eye movements and EEG are actually somewhat tricky to record and/or interpret simultaneously.

Eye movements cause travelling waves to sweep across visual brain areas (which are actually a huge chunk of the brain), potentially swamping whatever was going on before. Some of this is probably supports visual processing (by, e.g., persisting or resetting neural activity across eye movements), and some of it might be epi-phenomenal.

Furthermore, EEG electrodes pick up huge, non-neural signals from the movement of the ocular muscles and the movement of the eyeball itself. This can be reduced with clever pre-processing and more expensive amplifiers (e.g., that don't saturate or recover from saturation more quickly).

Reply on news.ycombinator.com