To Control Its Destiny, Facebook Bets Big on Hardware | Hacker News Reader

To Control Its Destiny, Facebook Bets Big on Hardware

theinformation.com

111 pointsstate42 comments

As with previous articles, The Information has unlocked this one for HN readers. Thanks!

https://hn.algolia.com/?dateRange=all&page=0&prefix=true&que...

There's a related thread at https://news.ycombinator.com/item?id=21836862.

VR is neat, but I'm absolutely inspired by the idea of improved human-computer interfaces like those described and their more mundane uses. We have eye tracking, the ability to read electrical signals and positioning from hands or head, the ability to feedback to the body through vibration (or sound), and existing high quality peripherals like keyboards for high precision entry and correction. There has to be a way to combine all of this into a better vim.

A couple binary inputs for CTRL and SHIFT from reading brainwaves seems doable. Feedback on whether they're active from different high frequency vibrations on the hands, neck, or ears.

There are lots of easy wins here if I can just push a couple mental buttons with my brain. Trigger scrolling or text selection based on eye movement. Select a buffer or macro based the mental buttons I'm pushing. You basically get a binary digit for every mental button you're capable of simultaneously tracking.

I'm home sick, and maybe it's the fever talking, but I think I'm going to do some basic research and start hacking something together. Very small chance of success, but I want mental modifier keys with vibrational feedback!

Summary: BCIs are still bad, use foot pedals as suggested in the sibling comment.

I don't want to rain on the parade but BCI (brain computer interfaces here EEG that is interpreted as input for something) is far from being usable for everday things. The technology sort of works but it is slow and unreliable. Most reliable method was P300 [0] which requires several repetitions of a signal someone decided to select. Entering letters for example can be done by flashing rows and columns of a matrix of these letters and concentrating on the one to choose. It's reactive in the sense as it will only trigger when an expected stimulus is detected. A lot less reliable was a system trained to recognize imagined motions (raising the left or right leg or arm as different categories) and worst method I saw was a simple feedback loop between EEG for moving something on the screen in one of two directions. This rarely worked even though it was a simple binary choice. Any switch will be better than this for now.

[0] https://en.wikipedia.org/wiki/P300_(neuroscience)

Totally fair. There's a few things I'd like to point out:

  * I get to push the EEG button with just my brain, which is neat and fun.
  * I get to build multiple high resolution EEGs into something that doesn't look terrible like a pair of headphones or a hat, which is fun (yet flammable).
  * I play a few instruments (most recently bought a violin, which I'm still terrible with). It took years to become good at each instrument. I assume that it would take years to become good at using a BCI and properly adjust it to my brain.
  * I'm working on this with the initial assumption that I'm most likely going to fail, but I'm going to have a good time.

Edit: * It also gives me an excuse to work on this: https://news.ycombinator.com/item?id=21661567

I'm neither a doctor not a psychologist but here is a wild guess: I would assume that it takes about as long to train yourself to work with a BCI as it takes to e.g. learn to reliably wiggle one ear, raise an eyebrow, just any voluntary movement that you do not control yet. Reasoning: in both cases you would have a feedback loop and would need to learn to reliably create the appropriate output and that it does not matter a lot whether the output goes to actual muscles or via EEG to trigger a response.

Maybe you want to give this a try first https://www.emacswiki.org/emacs/FootSwitches Seems a lot more reliable and easier to implement than messing with EEGs.

I can't imagine this not being the future of how we interface with our computing devices, way before proper neural control interfaces.

I'm curious what research is being done on this front.

FB is kinda badly positioned when it comes to man-machine interfaces, not because of their technical capability, but the lack of trust that is needed for societal acceptance in that space. People who initially supported Oculus wouldn't have done so if it came out being from FB, they wouldn't have gotten over the initial hump.

I think that this is one of the reasons why Apple bets so heavily on privacy, they had a much better long term vision when it comes to wearables and man-machine interfaces in general. If Cook distances Apple from China they have such a big advantage there, if not FB will use Apples connection with China to their advantage.

This is the age of vertical integration in technology. Truly astounding to see, and not something we've witnessed much of historically.

Personally I think a lot of it is being driven by the comoditization of advanced chip fabrication. Now with TSMC you can actually have better manufacturing capabilities than Intel. A situation we've not been in for over 30 years.

> Personally I think a lot of it is being driven by the commoditization of advanced chip fabrication.

That's an interesting perspective. With mobile device chip manufacturing being modularized, Facebook is able to integrate forward with the end user by combining software (i.e. their portfolio of apps) with consumer electronics.

Aggregation theory at work: https://stratechery.com/2015/aggregation-theory/

> not something we've witnessed much of historically.

Goes in cycles. Xerox, hp, ibm

There's no getting around the fact that developing AR technology requires substantial investment in R&D. Only companies such as Facebook, ones that employ some of the best engineers and also have a strong incentive to make that investment, are able to pull it off.

Whatever product comes out of Facebook under its current business model will be optimized for serving ads. Not necessarily serving the user.

This in itself is not bad, as it will still stimulate the growth of an ecosystem where developers can create products with different business models.

And as such they will not do it because people don't use such devices currently. It needs to be the other way around: have the devices, people start using them more and more, and only then come Google and Facebook to destroy the system with ads.

> are able to pull it off

TBC

I don’t want to sound mean spirited, but I don’t really like or trust FB. I would check my feed briefly once or twice a month and post something whenever I finished a book and wanted to announce that.

My attitude changed when I bought an Oculus Go and a week later when I bought an Oculus Quest. These devices use the FB platform, and all the videos that I now upload to FB can now be watched in the VR theater on the Oculus devices.

So basically, the Oculus products are so good that I decided to give FB another chance and I use FB a bit more often now.

EDIT: and the Star Wars Vader Immortal trilogy on the Oculus Quest is my favorite entertainment experience ever, including the industrial strength VR experiences I used to work on. My respect to the teams that did the Vader Immortal material.

So you all have a subscription to theinformation.com? Wow. I can't just buy a subscription for every random "news" site that appears on hacker news.

I don't get why news sites don't have a "coin slot" where I can insert a few quarters if I like an article and get a pass to read that day's edition.

I've encountered exactly that feature multiple times and I've never used it because god knows I'm not creating an account to give a site 0.99 and read an article. Friction.

Maybe if someone did it as a layer on top of news sites (like Google/FB login)... Maybe ...

> god knows I'm not creating an account to give a site 0.99

Who says you have to create a new account on every site? A cookie set after the payment processor has completed is all you need.

That will create a ton of angry customers who had their cookies auto-deleted e.g. because of Incognito. Support costs will outweigh the few pennies earned.

Especially with a tech-ninded crowd, relying on the persistence of cookies is completely unworkable because of the large percentage of users with extensions that block or wipe cookies eagerly.

Sounds like a business opportunity for the Brave browser.

This article requires x BAT to access. Unlock? Y/N

That will lead to every news headline to be clickbaity to grab your coins

As opposed to ads and BAT, which are facilitated whether or not an article turns out to be click bait?

"Coins" would be a manual process. If something is clickbaity, then you don't have to insert a coin.

No subscription there, but the full article is available without a paywall.

Guessing there is some geo-thing or A/B test going on.

The entire site is paywalled, but they unlock some articles specifically for HN

https://news.ycombinator.com/item?id=21839936

It's interesting how a number of Facebook engineers are willing to comment on what they work on, while the company itself refuses to provide any official statements.

You can assume anyone named and quoted in an article like this is speaking officially on behalf of the company they are working for, with some level of preparation and filtering by their Comms team.

Nah. This kind of segmentation of what can be disclosed exists even within engineering. E.g. in https://softwareengineeringdaily.com/2019/12/17/kubernetes-a... Karl Isenberg is willing to talk pretty freely about their CI/CD pipeline but is very cautious and doesn't give specifics when talking about the pipeline for pushing firmware updates to cars.

I think a better analogy is probably that engineers are also not willing to implement any feature anyone asks for.

The company has a lot of public statements that are easy to find, and the people who aren’t “providing comments” are likely either working on the next ones or think providing this comment would not be helpful to the company (including taking into account how much time they can invest in it).

Of course none of this is specific to FB in any way.

Where are we standing with open XR? Especially equipment, including controllers?

I love XR (=VR/AR/MR), but I still didn't manage a good easy open source way to do that.

The main open XR project is called Monado and it implements Khronos OpenXR:

https://monado.dev/ https://www.khronos.org/openxr/

What is MR?

Microsoft's brand for their platform which covers both VR (their confusingly named Windows Mixed Reality range of VR headsets) and AR (eg. HoloLens).

MR and AR are essentially equivalent in the "industry". MR is Mixed Reality. For "mixing" digital images with the real world.

Mixed reality

Doesn't AR also mean Mixed since it's augmented?

I think that’s a _really_ fair question! xD

I’ve heard “mixed reality” being used to refer to the continuum between reality - augmented reality - virtual reality. In the above question I’d read it as asking about the state of problems particularly pertaining to the transitions along this continuum. (Made up example: of you solve locomotion in VR by putting people on moving belts, a problem in MR would be how to transition someone from walking around in unobstructed reality to walking in VR. I don’t know if that’s a real problem—please take it just to illustrate what I mean.)

Edit: Wikipedia’s article on MR includes a whole section on “differences in terminology”! \o/ https://en.wikipedia.org/wiki/Mixed_reality They also use the word “continuum”, but don’t seem to directly back up my further claims about using the word to refer to problems arising specifically at the interface of AR/VR.

I believe Mixed Reality was a term coined by Microsoft to differentiate their products, basically a marketing term.[1]

[1]https://www.vox.com/2015/7/27/11615046/whats-the-difference-...

Does anyone think AR has the potential to do in the next decade what phones did in this last one?

I think it has the ability to do more once it is sufficiently miniaturized. Small to the point that it’s not much different than wearing eye glasses.

Glasses can make a viral and sticky form factor that we already have centuries of experience with. If we can build an AR glasses product people want to wear then we have the potential to greatly increase the amount of digital information a human sees and processes. This information can be addictive, like smartphones, and of high utility, like smartphones.

Look at the success of wireless earbuds, they will be part of distributed wearable AR systems, they're screenless AR that exist already. AR doesn't always mean screens. Glasses can replace screens, but they won't be worn by everybody like Phones, so AR glasses will first compete with desktop, laptop and tablet screens in more professional environments. It can only go mainstream like phones in contact lens form, or some kind of iris projector that sits on your nose, likely still powered by a watch or a phone. Not sure if we see that in the next decade though.

If done correctly, then yes.

being able to overlay information, in context, onto objects that it relates to, will be massive.

But only eventually.

Reply on news.ycombinator.com