A Vision of Future Vision
May 24, 2021 2:18 PM Subscribe
Project Starline is Google's latest and greatest prototype of the future of videoconferencing. How good is it? "Call it hyper-telepresence. Call it whatever you want. Either way, it’s pretty wild."
If it's like most other of Google's good ideas, it'll be graveyarded and leave many users unhappy. Surely by now Google's reputation for sunsetting apps used by millions of users -- but not enough millions, I guess? -- is going to prevent people from even bothering with a new product for lack of trusting Google to keep it running, even if it's awesome.
If it's like Google's bad ideas coughGoogleGlasscough it'll linger for years because it's some bonehead executive's pet project.
posted by tclark at 2:30 PM on May 24, 2021 [11 favorites]
If it's like Google's bad ideas coughGoogleGlasscough it'll linger for years because it's some bonehead executive's pet project.
posted by tclark at 2:30 PM on May 24, 2021 [11 favorites]
I look forward to viewing my rss feeds on it.
Oh wait, the fact that Google regularly abandons their products has already been covered? Ok then, I'll just say more seriously: Sure it looks sci-fi-y but after 15 months of Zoom calls that last goddamned thing I want to think about is Zoom++
posted by gwint at 2:34 PM on May 24, 2021 [18 favorites]
Oh wait, the fact that Google regularly abandons their products has already been covered? Ok then, I'll just say more seriously: Sure it looks sci-fi-y but after 15 months of Zoom calls that last goddamned thing I want to think about is Zoom++
posted by gwint at 2:34 PM on May 24, 2021 [18 favorites]
The thing I don't understand about this obviously very cool and expensive videoconference is why all my teleconferences have such bad latency. When I'm talking one-on-one on the phone I almost never step over the other party, but it happens regularly when connected to a teleconference. I'd be happy to get through a five minute conversation without an awkward exchange of "no, you go ahead"s that brings to mind a manual implementation of Ethernet collision avoidance.
posted by wnissen at 2:35 PM on May 24, 2021 [13 favorites]
posted by wnissen at 2:35 PM on May 24, 2021 [13 favorites]
Well, wnissen, I want your phone service. I have more latency problems with phone calls than Zoom these days.
posted by PhineasGage at 2:36 PM on May 24, 2021 [2 favorites]
posted by PhineasGage at 2:36 PM on May 24, 2021 [2 favorites]
Here's the full promo video, which I didn't see linked in the article.
I'm looking forward to the day when I get to climb into a little booth to do my interspace videophone calls.
posted by theodolite at 2:38 PM on May 24, 2021 [1 favorite]
I'm looking forward to the day when I get to climb into a little booth to do my interspace videophone calls.
posted by theodolite at 2:38 PM on May 24, 2021 [1 favorite]
When I'm talking one-on-one on the phone I almost never step over the other party, but it happens regularly when connected to a teleconference.
It's unfortunately the nature of internet teleconference -- rather than the old analog "party line" the latency is kind of the sum of the worst person's ping, your ping, and the amount of processing time the service takes to multiplex the feed. The services could do a lot to decrease the latency on the processing time, but the price can get pretty nonlinear pretty quickly, even to shave off 100 milliseconds which is usually enough to "feel weird" in latency terms.
posted by tclark at 2:40 PM on May 24, 2021 [3 favorites]
It's unfortunately the nature of internet teleconference -- rather than the old analog "party line" the latency is kind of the sum of the worst person's ping, your ping, and the amount of processing time the service takes to multiplex the feed. The services could do a lot to decrease the latency on the processing time, but the price can get pretty nonlinear pretty quickly, even to shave off 100 milliseconds which is usually enough to "feel weird" in latency terms.
posted by tclark at 2:40 PM on May 24, 2021 [3 favorites]
I was hoping it was going to be seamless auto-interpretation & auto-translation.
posted by aniola at 2:47 PM on May 24, 2021 [4 favorites]
posted by aniola at 2:47 PM on May 24, 2021 [4 favorites]
FWIW, Microsoft is working on this, too.
(disclaimer, I work for MS, and know people who are working on Mesh. I have not personally tried it, but 'how people do work' is baked into the MS business model, and we have a commitment to flexible/part-remote work for our own employees - up to 50% without manager approval - so we have actual skin in the game. And, the killing products thing? You can still run Win 95 apps on Windows 10, so...)
posted by parm at 2:48 PM on May 24, 2021 [14 favorites]
(disclaimer, I work for MS, and know people who are working on Mesh. I have not personally tried it, but 'how people do work' is baked into the MS business model, and we have a commitment to flexible/part-remote work for our own employees - up to 50% without manager approval - so we have actual skin in the game. And, the killing products thing? You can still run Win 95 apps on Windows 10, so...)
posted by parm at 2:48 PM on May 24, 2021 [14 favorites]
It seems a shame that an advertising company feels obliged to do techy stuff as well. It's clear their hearts aren't really in it anymore. Perhaps Google should just outsource all the tech stuff?
posted by thatwhichfalls at 2:50 PM on May 24, 2021 [6 favorites]
posted by thatwhichfalls at 2:50 PM on May 24, 2021 [6 favorites]
The thing about every smart & motivated person being gobbled up by a morally bankrupt misery train like google/Amazon etc is that they have these great ideas with the worlds worst applications. There's absolutely no need for this in video conferencing. You'd only think of that if you're the kind of person who regularly forgets other people are people. I bet if you gave this to some theater people they could do some fun things. Or medical researchers or for remote medicine. But no, we simply cannot have nice things. Give this to someone who manages to remember that other people are still people even when you can't see a face.
posted by bleep at 2:50 PM on May 24, 2021 [4 favorites]
posted by bleep at 2:50 PM on May 24, 2021 [4 favorites]
So…. How does that work in a N-way call with more than two people?
posted by njohnson23 at 2:55 PM on May 24, 2021
posted by njohnson23 at 2:55 PM on May 24, 2021
It seems a shame that an advertising company feels obliged to do techy stuff as well. It's clear their hearts aren't really in it anymore. Perhaps Google should just outsource all the tech stuff?
The techy stuff is how people convince themselves they're not only working to build a better ad.
posted by pwnguin at 3:10 PM on May 24, 2021 [9 favorites]
The techy stuff is how people convince themselves they're not only working to build a better ad.
posted by pwnguin at 3:10 PM on May 24, 2021 [9 favorites]
Unlike google's earlier abandoned projects this has something that could actually make it viable: it has extremely obvious sex-work applications.
posted by paper chromatographologist at 3:18 PM on May 24, 2021 [23 favorites]
posted by paper chromatographologist at 3:18 PM on May 24, 2021 [23 favorites]
I have a friend who worked peripherally on this project and it's very cool, but also a heck of long way away from being a product. He couldn't tell me many details, of course but he did tell me it uses:
* A fully tracked lenticular autostereoscopic display at ~250 ppi resolution at ~60Hz.
* Multiple cameras (normal color, IR, and RGBD), multiple microphones and directional speaker arrays
* Very carefully engineered lighting
* Bandwidth greater than most companies have for their internal networking, let alone publically available broadband
posted by riotnrrd at 3:21 PM on May 24, 2021 [17 favorites]
* A fully tracked lenticular autostereoscopic display at ~250 ppi resolution at ~60Hz.
* Multiple cameras (normal color, IR, and RGBD), multiple microphones and directional speaker arrays
* Very carefully engineered lighting
* Bandwidth greater than most companies have for their internal networking, let alone publically available broadband
posted by riotnrrd at 3:21 PM on May 24, 2021 [17 favorites]
Google might not sell it to porn, but porn will find a way now that the technology is on their radar, graveyard be damned.
posted by Slackermagee at 3:25 PM on May 24, 2021 [8 favorites]
posted by Slackermagee at 3:25 PM on May 24, 2021 [8 favorites]
It's doomed to failure, just like every other time people try to make this kind of thing work.
In-person works because you don't have to look at the person. You know they're there. You see them in your periphery. You can look at the same thing side-by-side. Et cetera.
This thing is a better illusion than Zoom, I'll grant that, but you're still sitting still in one place. You're still staring at your counterpart. Fidget and the illusion is broken. Turn around and you feel rude in a way you don't feel when in-person.
What people keep failing to grasp is that teleconferencing feels like a performance, not a collaboration. Performing is exhausting. Count me as someone who hates the phone, but even then, I can be on the phone and do things. I've chatted for an hour or more with my sister while I've done the laundry or even did work from home. Video-phone offers none of that flexibility or freedom.
Please don't put me on a stage just to have a work meeting, thank you.
posted by explosion at 3:33 PM on May 24, 2021 [44 favorites]
In-person works because you don't have to look at the person. You know they're there. You see them in your periphery. You can look at the same thing side-by-side. Et cetera.
This thing is a better illusion than Zoom, I'll grant that, but you're still sitting still in one place. You're still staring at your counterpart. Fidget and the illusion is broken. Turn around and you feel rude in a way you don't feel when in-person.
What people keep failing to grasp is that teleconferencing feels like a performance, not a collaboration. Performing is exhausting. Count me as someone who hates the phone, but even then, I can be on the phone and do things. I've chatted for an hour or more with my sister while I've done the laundry or even did work from home. Video-phone offers none of that flexibility or freedom.
Please don't put me on a stage just to have a work meeting, thank you.
posted by explosion at 3:33 PM on May 24, 2021 [44 favorites]
riotnrrd (I have always loved your nick BTW): I have a friend who worked peripherally on this project and it's very cool, but also a heck of long way away from being a product.
Yeah, this is a Google Research project. Emphasis on the Research. This will never go to market.
Research at an enormous software company (whether it's IBM, Microsoft, Google, etc) is where you put smart and scatterbrained people who like to make cool and impractical things like this. This creates good PR and gives your actual products a brand halo.
It also builds up a portfolio of software patents, in anticipation of a startup making a semi-practical version of this at some point in the future. You can then either buy out the startup at fire sale prices (because otherwise you'd sue them into oblivion) or you can add the patents to a defensive portfolio (so that when a competitor buys them out, you have another missile in the silo for your mutually assured patent destruction with said competitor).
As much as people like to hate on Google, generally they don't do the former and keep IP rights around for the latter. The only exception being when someone outright fucks them over like Levandowski.
posted by xthlc at 3:43 PM on May 24, 2021 [15 favorites]
Yeah, this is a Google Research project. Emphasis on the Research. This will never go to market.
Research at an enormous software company (whether it's IBM, Microsoft, Google, etc) is where you put smart and scatterbrained people who like to make cool and impractical things like this. This creates good PR and gives your actual products a brand halo.
It also builds up a portfolio of software patents, in anticipation of a startup making a semi-practical version of this at some point in the future. You can then either buy out the startup at fire sale prices (because otherwise you'd sue them into oblivion) or you can add the patents to a defensive portfolio (so that when a competitor buys them out, you have another missile in the silo for your mutually assured patent destruction with said competitor).
As much as people like to hate on Google, generally they don't do the former and keep IP rights around for the latter. The only exception being when someone outright fucks them over like Levandowski.
posted by xthlc at 3:43 PM on May 24, 2021 [15 favorites]
My current problem with teleconferencing is the opposite, I want to be reading the news or metafilter without being seen in meetings that are entirely superfluous to my life. This would make it worse!
posted by Carillon at 3:44 PM on May 24, 2021 [9 favorites]
posted by Carillon at 3:44 PM on May 24, 2021 [9 favorites]
Kinda torn, tbh. On the one hand, videophones were a flop. On the other hand, the world spent last year on Zoom. Somebody's gotta go first with the nextgen tech.
posted by mhoye at 3:45 PM on May 24, 2021
posted by mhoye at 3:45 PM on May 24, 2021
This hologram feature is a prank right? Or Google is turning into Houli as life imitates art.
Regarding Google Glass as an OG glass hole and one of the first outside of Google to develop apps please allow me a short rant.
Glass was a product done in by inept product management. Imagine a replacement for the helmet mounted Go Pro that also replaced a bluetooth headset and provided a simple heads up display for things like text messaging. Imagine being able to show someone what you were looking at on a video conference call without holding up the phone just look around the room.
Then the secondary uses for those needed assistive tech. There were apps for visually impaired people to help the recognize objects they were holding or looking at. Apps that did facial cue detection for people who struggle to read those cues. Even apps that tried to help face blind people by doing limited facial recognition based on people you’ve previously tagged.
Finally there were some great educational possibilities such as interactive field trips where students could be directed around a site and take notes using the camera and speech to text. The ability for apprentices in various professions to quickly record an activity and get an instructor / mentors feedback in real time.
This was the promise; but Google couldn’t deliver shy of those primary use cases or secondary. The hangouts video calling was dropped because they couldn’t solve the overheating problems because they never tuned their Linux Kernel from TI’s big cellphone factory settings. They never implemented the basic Bluetooth headphone capabilities like A2DP streaming. They couldn’t even keep the screens from falling apart after just a couple months of use — a basic manufacturing defect that should of been fixed. They didn’t listen to those of us who were living with the device everyday and attempting to build apps.
posted by interogative mood at 3:56 PM on May 24, 2021 [14 favorites]
Regarding Google Glass as an OG glass hole and one of the first outside of Google to develop apps please allow me a short rant.
Glass was a product done in by inept product management. Imagine a replacement for the helmet mounted Go Pro that also replaced a bluetooth headset and provided a simple heads up display for things like text messaging. Imagine being able to show someone what you were looking at on a video conference call without holding up the phone just look around the room.
Then the secondary uses for those needed assistive tech. There were apps for visually impaired people to help the recognize objects they were holding or looking at. Apps that did facial cue detection for people who struggle to read those cues. Even apps that tried to help face blind people by doing limited facial recognition based on people you’ve previously tagged.
Finally there were some great educational possibilities such as interactive field trips where students could be directed around a site and take notes using the camera and speech to text. The ability for apprentices in various professions to quickly record an activity and get an instructor / mentors feedback in real time.
This was the promise; but Google couldn’t deliver shy of those primary use cases or secondary. The hangouts video calling was dropped because they couldn’t solve the overheating problems because they never tuned their Linux Kernel from TI’s big cellphone factory settings. They never implemented the basic Bluetooth headphone capabilities like A2DP streaming. They couldn’t even keep the screens from falling apart after just a couple months of use — a basic manufacturing defect that should of been fixed. They didn’t listen to those of us who were living with the device everyday and attempting to build apps.
posted by interogative mood at 3:56 PM on May 24, 2021 [14 favorites]
You can then either buy out the startup at fire sale prices (because otherwise you'd sue them into oblivion) or you can add the patents to a defensive portfolio (so that when a competitor buys them out, you have another missile in the silo for your mutually assured patent destruction with said competitor).
As much as people like to hate on Google, generally they don't do the former and keep IP rights around for the latter.
As much as people like to hate on Google it's 100% deserved.
posted by bleep at 4:12 PM on May 24, 2021 [3 favorites]
As much as people like to hate on Google, generally they don't do the former and keep IP rights around for the latter.
As much as people like to hate on Google it's 100% deserved.
posted by bleep at 4:12 PM on May 24, 2021 [3 favorites]
Reminds me of William Gibson's quote: "The future is already here – it’s just not very evenly distributed." I don't see any way this could be commercially viable for the vast majority of people. Like, I can see this literally being a single setup in each major city, at the most. Cool tech, though.
posted by zardoz at 4:24 PM on May 24, 2021 [2 favorites]
posted by zardoz at 4:24 PM on May 24, 2021 [2 favorites]
Perhaps Google should just outsource all the tech stuff?
Should Google just become a VC fund?
posted by JoeZydeco at 4:35 PM on May 24, 2021 [2 favorites]
Should Google just become a VC fund?
posted by JoeZydeco at 4:35 PM on May 24, 2021 [2 favorites]
It ids going to get cancelled after the PM gets promoted.
posted by kadmilos at 4:39 PM on May 24, 2021 [3 favorites]
posted by kadmilos at 4:39 PM on May 24, 2021 [3 favorites]
." I don't see any way this could be commercially viable for the vast majority of people.
It's not intended to be. Cisco's equivalent a decade ago was into the six figures per room and this is more advanced so despite tech getting cheaper between now and then, I expect it'll be the same or higher. Part of the cost back then was you had to dress each room exactly the same to maintain the illusion. Maybe they can fake that in real-time with modern processing but I doubt it would be artifact free enough to not slip into the uncanny valley.
If it comes to fruition, it'll be for people to communicate between Google campuses and will be used to save travel costs for internal VIPs who want a more lifelike experience than Google Meet.
Cisco had dreams of commercial businesses where you might rent a room for 90 minutes for an interview or whatever but it never got widespread enough adoption to be viable and I can't see this being any different.
posted by Candleman at 5:12 PM on May 24, 2021 [3 favorites]
It's not intended to be. Cisco's equivalent a decade ago was into the six figures per room and this is more advanced so despite tech getting cheaper between now and then, I expect it'll be the same or higher. Part of the cost back then was you had to dress each room exactly the same to maintain the illusion. Maybe they can fake that in real-time with modern processing but I doubt it would be artifact free enough to not slip into the uncanny valley.
If it comes to fruition, it'll be for people to communicate between Google campuses and will be used to save travel costs for internal VIPs who want a more lifelike experience than Google Meet.
Cisco had dreams of commercial businesses where you might rent a room for 90 minutes for an interview or whatever but it never got widespread enough adoption to be viable and I can't see this being any different.
posted by Candleman at 5:12 PM on May 24, 2021 [3 favorites]
Things are slow at the “actual problems affecting real people” desk.
Not sorry. The best minds of a generation working on .... this. Sure, fantastic resume filler and the feedback loop gives one a big charge, but. Yeah.
Perhaps there’s some useful application (maybe ensafening sex work or therapy) that can’t already be met with existing tech. Likely. But so many other problems need solving.
Or maybe I have been in tech too long and should get off my lawn. “ I shoulda listened to mom and gone into advertising.”
posted by armoir from antproof case at 5:48 PM on May 24, 2021
Not sorry. The best minds of a generation working on .... this. Sure, fantastic resume filler and the feedback loop gives one a big charge, but. Yeah.
Perhaps there’s some useful application (maybe ensafening sex work or therapy) that can’t already be met with existing tech. Likely. But so many other problems need solving.
Or maybe I have been in tech too long and should get off my lawn. “ I shoulda listened to mom and gone into advertising.”
posted by armoir from antproof case at 5:48 PM on May 24, 2021
Instead of becoming a VC they could cut the same costs costs, and start paying out a massive dividend to their existing investors who will invest that money in other companies.
posted by interogative mood at 5:51 PM on May 24, 2021
posted by interogative mood at 5:51 PM on May 24, 2021
I can't recall ever seeing a tech announcement met with this degree of gratuitous cynicism.
Perhaps there is hope.
posted by flabdablet at 6:19 PM on May 24, 2021 [14 favorites]
Perhaps there is hope.
posted by flabdablet at 6:19 PM on May 24, 2021 [14 favorites]
Then the secondary uses for those needed assistive tech. There were apps for visually impaired people to help the recognize objects they were holding or looking at. Apps that did facial cue detection for people who struggle to read those cues. Even apps that tried to help face blind people by doing limited facial recognition based on people you’ve previously tagged.
This kind of thing is exactly why I was actually excited about Glass and rather annoyed that even here on MeFi all that ever got talked about was oymygodacamera. It was, and still is, a big missed opportunity to improve the lives of a lot of people
posted by wierdo at 6:22 PM on May 24, 2021 [4 favorites]
This kind of thing is exactly why I was actually excited about Glass and rather annoyed that even here on MeFi all that ever got talked about was oymygodacamera. It was, and still is, a big missed opportunity to improve the lives of a lot of people
posted by wierdo at 6:22 PM on May 24, 2021 [4 favorites]
Google has fucked me over with their hardware so many times they're on my never-again list. And once you're on my never-again list, you never come off.
I once bought a laptop from those fuckers and when I was in another country they refused to honour the warranty. A $2000 laptop dead in 3 months. I made it back to the states with a few days left before the warranty expired. They sent me a new one and 3 months later... same defect.
posted by dobbs at 6:26 PM on May 24, 2021
I once bought a laptop from those fuckers and when I was in another country they refused to honour the warranty. A $2000 laptop dead in 3 months. I made it back to the states with a few days left before the warranty expired. They sent me a new one and 3 months later... same defect.
posted by dobbs at 6:26 PM on May 24, 2021
Also, research projects like this are usually not great products in and of themselves, but the developments in technology required to make them work turn out to be transformative much more frequently.
posted by wierdo at 6:27 PM on May 24, 2021 [3 favorites]
posted by wierdo at 6:27 PM on May 24, 2021 [3 favorites]
To to summarise, it's a technology that works with a very expensive setup, prohibitive bandwidth, and a restrictive set of use cases, which won't ever be rolled out to the public. That's fine, everyone hates teleconferencing, no loss to humanity. I could very well see applications for high-end stereoscopic situational-awareness video though, let's say, in the sensor packages of drones and fighter-bombers.
posted by Fiasco da Gama at 6:45 PM on May 24, 2021 [1 favorite]
posted by Fiasco da Gama at 6:45 PM on May 24, 2021 [1 favorite]
It's a bit like NASA. Inventing ways to reduce twenty cameras down to ten and compress enormous point clouds down to something transmittable on regular backbones, you're still getting wins versus something you might productize someday.
And executives are convinced there's some magic to telepresence.
High fidelity 3D capture and real time mesh construction in a usable general case that handles pesky occlusions like glasses and earlobes is... Useful? I guess?
posted by abulafa at 6:54 PM on May 24, 2021
And executives are convinced there's some magic to telepresence.
High fidelity 3D capture and real time mesh construction in a usable general case that handles pesky occlusions like glasses and earlobes is... Useful? I guess?
posted by abulafa at 6:54 PM on May 24, 2021
Somebody's gotta go first with the nextgen tech.
Director: I have a great idea! What if we had everyone come into the office during the same hours? They could interact with each other directly (and even randomly) and all this synergy would generate the next generation of new product ideas and concepts. And if we subsidized (or even paid for, right?) a few meals while they were here, they'd spend even more time working together.
Middle Management: Brilliant!
posted by Insert Clever Name Here at 7:05 PM on May 24, 2021 [1 favorite]
Director: I have a great idea! What if we had everyone come into the office during the same hours? They could interact with each other directly (and even randomly) and all this synergy would generate the next generation of new product ideas and concepts. And if we subsidized (or even paid for, right?) a few meals while they were here, they'd spend even more time working together.
Middle Management: Brilliant!
posted by Insert Clever Name Here at 7:05 PM on May 24, 2021 [1 favorite]
To me this has "military" written all over it.
posted by rhizome at 7:15 PM on May 24, 2021 [1 favorite]
posted by rhizome at 7:15 PM on May 24, 2021 [1 favorite]
My company is doing something similar, on the camera side. You can do much of this on a pretty normal PC with a good GPU, and good home internet. Not amazing gigabit fiber internet, just a decent cable connection or similar.
This tech doesn't need an order of magnitude more bandwidth, or other huge scaling factor. If you have a couple more cameras, or an added depth sensor, that's 2-3x more data coming in. If your algorithm balloons that 2-3x to 10x or whatever, then your algorithm is bad, heh.
posted by ryanrs at 7:26 PM on May 24, 2021 [2 favorites]
This tech doesn't need an order of magnitude more bandwidth, or other huge scaling factor. If you have a couple more cameras, or an added depth sensor, that's 2-3x more data coming in. If your algorithm balloons that 2-3x to 10x or whatever, then your algorithm is bad, heh.
posted by ryanrs at 7:26 PM on May 24, 2021 [2 favorites]
Heheheh.
Speaking of which, does anyone know where the pick up some Google glass that isn't tethered to cooperate mandated usage or applications?
posted by firstdaffodils at 9:44 PM on May 24, 2021 [1 favorite]
Speaking of which, does anyone know where the pick up some Google glass that isn't tethered to cooperate mandated usage or applications?
posted by firstdaffodils at 9:44 PM on May 24, 2021 [1 favorite]
I do not like this valley. It is uncanny.
posted by prismatic7 at 10:14 PM on May 24, 2021 [4 favorites]
posted by prismatic7 at 10:14 PM on May 24, 2021 [4 favorites]
this has something that could actually make it viable: it has extremely obvious sex-work applications.
There's already NetMeeting for that.
... what, NetMeeting isn't around any more? Microsoft killed it? They never told anyone...
posted by Cardinal Fang at 2:13 AM on May 25, 2021
There's already NetMeeting for that.
... what, NetMeeting isn't around any more? Microsoft killed it? They never told anyone...
posted by Cardinal Fang at 2:13 AM on May 25, 2021
Has FOSTA/SESTA killed the dedicated sextech/teledildonics industry, or is that still a thing?
posted by acb at 2:20 AM on May 25, 2021
posted by acb at 2:20 AM on May 25, 2021
"I can show you this apple from Whole Foods, and I can see exactly what you’re looking at"
It seems like it's been well beyond high time to regulate this ad company, or at least put a severe and hard privacy kibosh on this kind of intrusive stuff across the board.
posted by They sucked his brains out! at 3:35 AM on May 25, 2021 [2 favorites]
It seems like it's been well beyond high time to regulate this ad company, or at least put a severe and hard privacy kibosh on this kind of intrusive stuff across the board.
posted by They sucked his brains out! at 3:35 AM on May 25, 2021 [2 favorites]
Will it allow me to turn my head and see what else my employee has on in their workspace? Asking for a company friend.
posted by Nanukthedog at 3:53 AM on May 25, 2021
posted by Nanukthedog at 3:53 AM on May 25, 2021
What people keep failing to grasp is that teleconferencing feels like a performance, not a collaboration. Performing is exhausting.
The whole history of telecommunications is peppered with failures to realise this. Normally because the companies doing the research are looking for ways of profiting by selling an every higher fidelity "performance". The problems go in two directions: first of all companies over-estimate the extent to which people will want to use video as a collaboration tool (especially so if the demonstration looks visually interesting). Secondly they fail to grasp the impact of technologies low bandwidth communications that let people split their attention or focus on a shared task. Thus both text messaging and tweeting were overlooked as technologies suited only for chat amongst engineers - while we have long seen as assumption that people will only be happy collaborating when they can focus on each other's expensively rendered faces (rather than what they are actually collaborating on).
posted by rongorongo at 4:31 AM on May 25, 2021 [2 favorites]
The whole history of telecommunications is peppered with failures to realise this. Normally because the companies doing the research are looking for ways of profiting by selling an every higher fidelity "performance". The problems go in two directions: first of all companies over-estimate the extent to which people will want to use video as a collaboration tool (especially so if the demonstration looks visually interesting). Secondly they fail to grasp the impact of technologies low bandwidth communications that let people split their attention or focus on a shared task. Thus both text messaging and tweeting were overlooked as technologies suited only for chat amongst engineers - while we have long seen as assumption that people will only be happy collaborating when they can focus on each other's expensively rendered faces (rather than what they are actually collaborating on).
posted by rongorongo at 4:31 AM on May 25, 2021 [2 favorites]
Please don't put me on a stage just to have a work meeting, thank you.
Too polite to ever gain any business traction, but presents a large enough wedge in team composition pies to show how some fail.
posted by filtergik at 6:37 AM on May 25, 2021
Too polite to ever gain any business traction, but presents a large enough wedge in team composition pies to show how some fail.
posted by filtergik at 6:37 AM on May 25, 2021
OK but then I also want the option to strangle the other person with my mind
posted by kleinsteradikaleminderheit at 7:37 AM on May 25, 2021 [2 favorites]
posted by kleinsteradikaleminderheit at 7:37 AM on May 25, 2021 [2 favorites]
posted by explosion: In-person works because you don't have to look at the person. You know they're there. You see them in your periphery ... teleconferencing feels like a performance, not a collaboration. Performing is exhausting.
I believe I completely understand where this is coming from, but isn't it interesting that this issue has nothing whatever to do with the actual act, quality, or technology of teleconferencing and everything to do with our feelings, habits, and expectations? The attitude toward teleconferencing. Culture.
I can't cite it for you but I recall a story about a successful medium-term collaboration (between two people) via teleconference. Some years ago now, in the pre-Zoom days. The two participants each had proper offices, with no kids or pets running around. First thing each morning, they would each start the teleconference on a second monitor. Then they would leave it running. All day. They more-or-less pretended they had adjoining desks, and that was the level of "performance" they aspired to.
They didn't have magical zero-latency connections, or special rooms to teleconference in, or 3D holographic display technologies. They had something that was simple enough and reliable enough to turn on and stop thinking about.
It's only a stage if we collectively agree it has to be.
posted by Western Infidels at 8:01 AM on May 25, 2021 [2 favorites]
I believe I completely understand where this is coming from, but isn't it interesting that this issue has nothing whatever to do with the actual act, quality, or technology of teleconferencing and everything to do with our feelings, habits, and expectations? The attitude toward teleconferencing. Culture.
I can't cite it for you but I recall a story about a successful medium-term collaboration (between two people) via teleconference. Some years ago now, in the pre-Zoom days. The two participants each had proper offices, with no kids or pets running around. First thing each morning, they would each start the teleconference on a second monitor. Then they would leave it running. All day. They more-or-less pretended they had adjoining desks, and that was the level of "performance" they aspired to.
They didn't have magical zero-latency connections, or special rooms to teleconference in, or 3D holographic display technologies. They had something that was simple enough and reliable enough to turn on and stop thinking about.
It's only a stage if we collectively agree it has to be.
posted by Western Infidels at 8:01 AM on May 25, 2021 [2 favorites]
xthlc: heh, thank you for the compliment.
I work in a corporate research lab myself (though not at Google) and you've totally nailed the reasons why private research is conducted: prestige, tech transfer, and patent portfolio growth. It's a very, very impressive project, but certainly several years away from any practical implementation.
posted by riotnrrd at 10:35 AM on May 25, 2021 [1 favorite]
I work in a corporate research lab myself (though not at Google) and you've totally nailed the reasons why private research is conducted: prestige, tech transfer, and patent portfolio growth. It's a very, very impressive project, but certainly several years away from any practical implementation.
posted by riotnrrd at 10:35 AM on May 25, 2021 [1 favorite]
It's just such a pointless, collossal waste of everything. It's like discovering how to make a black hole and then setting up a pipeline from all the worlds museums right into the hole. Why even do that? Just... don't? Don't do that?
posted by bleep at 11:44 AM on May 25, 2021 [1 favorite]
posted by bleep at 11:44 AM on May 25, 2021 [1 favorite]
Don't spend all of the worlds limited resources on shit that's going right into the toilet just because you have wealth poisoning? Please?
posted by bleep at 11:45 AM on May 25, 2021 [3 favorites]
posted by bleep at 11:45 AM on May 25, 2021 [3 favorites]
Nvidia did an interesting tech demo in the same sphere last year-- particularly the "AI Compression" stuff. Essentially building a 3d model of your face realtime-- then treating your 'data' like you would a game-character-- sending the details to animate the mesh, rather then essentially sending 60 photos a second. Their example shows a 98KB/frame stream getting reduced to 0.1KB/frame.
Mix that with that old kinetic tech demo-- where the guy was using the kinetic to track his head/eyes-- so if his TV acted like a really quite convincing window into a 3D world (as he shifted his head, the perspective on the far side of the window would accurately reflect that).
Someone like Apple, with their front-facing IR/depth mapping sensors (used for Face ID) plus their decent ML chips, seems like they could do something really close to this with everything they already have deployed to millions of people-- without the bandwidth issue.
posted by Static Vagabond at 11:52 AM on May 25, 2021
Mix that with that old kinetic tech demo-- where the guy was using the kinetic to track his head/eyes-- so if his TV acted like a really quite convincing window into a 3D world (as he shifted his head, the perspective on the far side of the window would accurately reflect that).
Someone like Apple, with their front-facing IR/depth mapping sensors (used for Face ID) plus their decent ML chips, seems like they could do something really close to this with everything they already have deployed to millions of people-- without the bandwidth issue.
posted by Static Vagabond at 11:52 AM on May 25, 2021
We do exactly that screen-as-a-window demo with our camera tech. You don't need the kinect, or really any special tech on the receive end, just a normal front-facing camera and face tracking. We use an iphone. The iphone is great because it's easy to move your head a large amount relative to the little screen, so it's a lot more obvious that you can peer around stuff and see nominally off-screen regions by craning your neck.
posted by ryanrs at 12:10 PM on May 25, 2021
posted by ryanrs at 12:10 PM on May 25, 2021
Apple's no better. In the no-true-tech-company world, they're a marketing firm disguised as a tech company, Uber is a taxi company disguised as a tech company, Netflix is a tv studio disguised as a tech company, WeWork is a real estate scam no matter how you slice it, and Amazon just runs arbitrage for Ali Express (disguised as a tech company).
posted by fragmede at 12:10 PM on May 25, 2021
posted by fragmede at 12:10 PM on May 25, 2021
If you're reading metafilter on an iphone you are using an Apple/Amazon tech stack from start to finish.
posted by ryanrs at 12:15 PM on May 25, 2021 [1 favorite]
posted by ryanrs at 12:15 PM on May 25, 2021 [1 favorite]
IMO this will definitely be used in the near (like... 3-5 years from now) future. Google is testing it with a big lenticular 3D TV and like 10 cameras and depth sensors, but that's like... the devkit version.
I would imagine that a finished version will use a trio of RGBD cameras, an actual lightfield display like the Looking Glass, and cost in the $15-20K range. It'll probably mainly be used for corporate telepresence at first but may trickle down to video booths you can step into at any [participating remote work company/global logistics corp here].
The way I think about it, it's not that we need a new form of communication, but that this is the next big step for fidelity when it comes to video comms. Video calls used to be grainy, laggy, and noisy. Now they're pretty clear and in decent HD. Next they'll be 60FPS and 1080p or 4K as bandwidth allows. But that kinda tops out clarity and temporal resolution, and you want something more - depth. They're looking a long ways ahead... along with Apple, Facebook, Microsoft, Nvidia, and everyone else.
posted by BlackLeotardFront at 1:37 PM on May 25, 2021 [1 favorite]
I would imagine that a finished version will use a trio of RGBD cameras, an actual lightfield display like the Looking Glass, and cost in the $15-20K range. It'll probably mainly be used for corporate telepresence at first but may trickle down to video booths you can step into at any [participating remote work company/global logistics corp here].
The way I think about it, it's not that we need a new form of communication, but that this is the next big step for fidelity when it comes to video comms. Video calls used to be grainy, laggy, and noisy. Now they're pretty clear and in decent HD. Next they'll be 60FPS and 1080p or 4K as bandwidth allows. But that kinda tops out clarity and temporal resolution, and you want something more - depth. They're looking a long ways ahead... along with Apple, Facebook, Microsoft, Nvidia, and everyone else.
posted by BlackLeotardFront at 1:37 PM on May 25, 2021 [1 favorite]
For our camera stuff, the big selling point, we think, is eye contact. We've found that even at 720p, the sensation of eye contact is strong with light field capture. This works with a normal computer display, it doesn't need to be 3D or VR. You can 100% tell when I'm looking at your eyes vs over your shoulder. Soooo many men are going to get caught inappropriately staring if this tech ever goes mainstream, ha ha.
posted by ryanrs at 3:33 PM on May 25, 2021
posted by ryanrs at 3:33 PM on May 25, 2021
It's just such a pointless, collossal waste of everything.
Much R&D starts like this. It took a long time for many of the things at PARC to become practical but their research was important. At some point in the not too distant future, this type of technology (or something influenced by it) will become practical for widespread use. Just like early videophones that were awful and expensive compared to FaceTime and Zoom.
posted by Candleman at 4:26 PM on May 25, 2021 [1 favorite]
Much R&D starts like this. It took a long time for many of the things at PARC to become practical but their research was important. At some point in the not too distant future, this type of technology (or something influenced by it) will become practical for widespread use. Just like early videophones that were awful and expensive compared to FaceTime and Zoom.
posted by Candleman at 4:26 PM on May 25, 2021 [1 favorite]
Nvidia did an interesting tech demo in the same sphere last year-- particularly the "AI Compression" stuff. Essentially building a 3d model of your face realtime-- then treating your 'data' like you would a game-character-- sending the details to animate the mesh, rather then essentially sending 60 photos a second. Their example shows a 98KB/frame stream getting reduced to 0.1KB/frame
Way back in the 90s, when I was involved in teleconferencing research - I worked with some people who were interested in "symbolic acting" - this was the idea that your avatar in a meeting could be animated to show your work state without you having to directly control it. This kind of signalling is something we do all the time in real life: we look down and shield our eyes to signal "busy, don't disturb", we leave our jacket and keys on the desk to say "away for a few minutes". Modern tools like Teams distill this down to a coloured icon next to our photos to indicate "away/busy/free". Like with rendering - this is also a form of compression too: work out what signals to send and transmit only those. Those of us who have spent endless hours in Zoom video calls, will know this feeling: you want to turn off your live video feed sometimes so as not to appear rude; sometimes you actually want to turn it off so that you can better concentrate on what is being said.
Google's "Project Starline" is dramatically fancier - but it is still the same mechanism as with Teams: we are showing an avatar of the other person rather than the person themselves. When the other person talks - the avatar talks, when they turn their head or blink - then so does the avatar. That is great when that person is the centre of stage - but what about when we need to depict them as a participant is 10 minutes into some boring explanation and they are getting sleepy and distracted by their kid outside the door. If we choose to depict the truth - then we are indeed obliging each person to perform so as to appear to be the attentive listener they don't want to be. If we lie - then we fail to send back cues to the person speaking that he is losing his audience.
posted by rongorongo at 2:50 AM on May 26, 2021
Way back in the 90s, when I was involved in teleconferencing research - I worked with some people who were interested in "symbolic acting" - this was the idea that your avatar in a meeting could be animated to show your work state without you having to directly control it. This kind of signalling is something we do all the time in real life: we look down and shield our eyes to signal "busy, don't disturb", we leave our jacket and keys on the desk to say "away for a few minutes". Modern tools like Teams distill this down to a coloured icon next to our photos to indicate "away/busy/free". Like with rendering - this is also a form of compression too: work out what signals to send and transmit only those. Those of us who have spent endless hours in Zoom video calls, will know this feeling: you want to turn off your live video feed sometimes so as not to appear rude; sometimes you actually want to turn it off so that you can better concentrate on what is being said.
Google's "Project Starline" is dramatically fancier - but it is still the same mechanism as with Teams: we are showing an avatar of the other person rather than the person themselves. When the other person talks - the avatar talks, when they turn their head or blink - then so does the avatar. That is great when that person is the centre of stage - but what about when we need to depict them as a participant is 10 minutes into some boring explanation and they are getting sleepy and distracted by their kid outside the door. If we choose to depict the truth - then we are indeed obliging each person to perform so as to appear to be the attentive listener they don't want to be. If we lie - then we fail to send back cues to the person speaking that he is losing his audience.
posted by rongorongo at 2:50 AM on May 26, 2021
It's just such a pointless, collossal waste of everything.
Much R&D starts like this. It took a long time for many of the things at PARC to become practical but their research was important. At some point in the not too distant future, this type of technology (or something influenced by it) will become practical for widespread use. Just like early videophones that were awful and expensive compared to FaceTime and Zoom.
I get this but Google isn't doing this for the public good, they're doing this to make law missles they can launch at other companies in court. At best. No one else will see a benefit from this. I would love it if anyone else was doing this for any other reason.
posted by bleep at 9:30 AM on May 26, 2021
Much R&D starts like this. It took a long time for many of the things at PARC to become practical but their research was important. At some point in the not too distant future, this type of technology (or something influenced by it) will become practical for widespread use. Just like early videophones that were awful and expensive compared to FaceTime and Zoom.
I get this but Google isn't doing this for the public good, they're doing this to make law missles they can launch at other companies in court. At best. No one else will see a benefit from this. I would love it if anyone else was doing this for any other reason.
posted by bleep at 9:30 AM on May 26, 2021
« Older She Stays Winning | Electric Blue IKEA Newer »
This thread has been archived and is closed to new comments
posted by justkevin at 2:28 PM on May 24, 2021 [68 favorites]