Google's parent company is disabling old smart home devices

[wuench]

I'm sure Blockbuster video felt the same way. Nobody is really going to stream video when they could have uncompressed local, right?

In 5 years I'll bet latency and stability are far better, and cloud is actually a valid option. It's not today, but 5 years ago streaming video was also niche.

 
Latency is a function of distance and the speed of light.  But yeah, you never know that guy at NASA is working at faster than light travel, so we might get better latency if that pans out in the next 5 years.    
 
I know the world is predicting that everything is going to go more that way and a lot of it will for a while, but I am going to go out on my own limb and say a lot of it will come back the other way in the end.

 

[Dean Roddey]

That's not quite the same thing. The level of interaction with streaming media systems is small and very infrequent and the worst case scenario is that you don't get watch Wheel of Fortune today or something. The requirements for automation, if it's serious automation, that requires low latency feedback, on which reasonably critical processes depend, and such, is quantitatively different enough that it's arguably not really comparable.

 

[pete_c]

It is still further away than 5 years. 
 
The issues of being able to do transactional processing to thousands of folks all at the same time is still playing ketchup to what can be done today with a single/multiple processing on a personal dedicated automation box at home.  I played a bit with that stuff and the software programmers always blamed the transport and the transport people always blamed the software folks for any latencies; it was vicious.  Today you can't really complain to a computer  / server in the cloud; cuz it won't matter really.
 
First newbie cloud automator not having ever done it the long math way (at home based automation) automation will appear to be or just be fast because typically the newbie automator doesn't have anything to compare it to it.
 
I have to admit that playing with the Securifi Almond + or Amazon Echo that was doing all it's stuff in the Amazon cloud, I personally didn't see any lags in the response times comparing the cloud connection via telephone or via computer or via the LCD screen.  Maybe I just can't tell though cuz I am getting older and slower than faster so it just seems fast to me.  So it's really an easy button solution and you don't have to know how it works.

 

[rismoney]

There is no overcoming Little's Law.  The issue of latency will compound in relation to the number of customers.  This is as true on the internet as it is at the deli counter.  As an aside, I design low latency financial market trading systems.
 
Remember you home router's connection is just your 1st hop, with the 2nd to the ISP.  Home and many other lower cost routers optimize around bandwith, not latency.   Then you have all the other routers, switches load balancers that sit between the ISP and say Amazon. Then you have the overhead of a cloud resource, probably running on a vm (versus baremetal), making a http request against a web server that services many other simultaneous requests.  Facter in dns lookups, health checks and a myriad of other things that happen, and instantaneous is typically front end trickery and foolery with lazy background activities happening unbeknownst to the requestor.

 

[pete_c]

There is no overcoming Little's Law.  The issue of latency will compound in relation to the number of customers.  This is as true on the internet as it is at the deli counter.
 
But it is overcoming little's law and the issues relating to latency are starting to dissipate. 

 

[xlurkr]

I don't know what Little's Law is.  I will look it up on Wikipedia after posting.
 
But I now have an Echo Dot, and I've noticed that having a conversation with your automation controller changes your expectations about latency in subtle ways.  I once was pleased that when I walked into a room from an adjacent room, the vacated room lighting dimmed and the newly occupied room lit up in what appeared to be an elegantly orchestrated manner.  Or that the "watch movie" scene dimmed all the lights at the same rate.  With my Dot, if the action I request happens during Alexa's verbal response, I'm impressed.  If you've never seen the first two happen, and your only experience is the last one, you'll still be impressed, and likely not understand the impact of latency.
 
I know and you know that you can trigger complex scene transitions from cloud interactions, and have both experiences, but I hope you understand what I'm saying.  There's the old magic and the new magic, but it's all magic, which is cooler than no magic, the first time you see it.
 
-Tom

 

[xlurkr]

btw, I wonder if this ins't part of the plan.  That is, when we conversev with someone, we expect them to think it over for a while, even an infinitesimal while, before reacting.  Wouldn't it be a sign of a lack of judgement if a person didn't do this?  Doesn't the same apply to software meant to be thought of as a person?
 
-Tom

 

[Madcodger]

So, here's my personal experience:

I am in our second automated home over the past 10 or so years. Both have been pleasurable to own, but most of the problems have been related to local issues, not issues occurring in a remote, central server. Granted, we have been blessed with remarkably good Fios connections in both those homes, so speed / latency and reliability of connection have not been an issue. I tried implementing voice control / interaction many years ago in various ways (as an example) but never got it right.

Then along came the Echo. And....... It. Just. Works.
Earlier, it was the Nest...... It. Just. Works.
I moved the accounting and project management software of our businesses to the cloud and..... It. Just. Works.

Oh - and WAF has never been higher, or even close to where it is now.

I will never convince those of you who hate remote processing and storage (the cloud) that it is a good - even acceptable - option. But as a busy homeowner and business person, I am largely done with trying to manage things locally. Yes, I know there are risks and I accept those as reasonable and generally manageable. And yes, every blue moon something is not accessible, so there is a need for local "overrides" on some things. But my cloud solutions are MUCH easier to manage than my local solutions ever were, and when time is money and your most precious resource, this becomes a big issue. For me, the debate is over. Central process and storage is reliable and very convenient for me, and I am not sufficiently concerned with privacy to alter course, as my suppliers are well known companies with high levels of competency that exceed my own ability to employ and manage these functions.

Rant on, friends. I have defected to the "It Just Works" - easily and at relatively low cost - camp.

 

[pete_c]

Yup the use of Cisco / Microsoft / Google / Amazon cloud services has become very cost effective for offices that use 50-100 or more computers. 
 
I know of a local company here that manages some 20-30 businesses via the new cloud services and only needs one/two resources at the home office to do this.  It is a win win these days having to only pay one/two resource for managing 20-30 locations.
 
Google is giving away its cloud services to the public sector (really cheap) and it is working fine for them.  What is better than not having to pay for cloud services (and still be broke?)
 
You do not need anyone at the business site to manage the computers / software etc. 
 
It's an easy peasy win win utilizing the cloud services now to manage a business that relies on desktops.
 
That said QOS of a residential ISP connection is still not soup.  In fact many today are still utilizing 20 year old cable wires.
 
Most ISP's have not proactively upgraded any of their last mile stuff in older residential areas and mostly have relied on old infrastructure to provide internet transport to the houses.  TV's still work there.  They do not want to spend the money.
 
I switched over to Verizon FIOS in one home when it was first available in a small 50 home subdivision on a key.  I was using DSL before FIOS.  The cable there was originally installed way back and never upgraded.  IE: HD television didn't work because it was just old.  My newest Motorola cable modem has Gb in and out these days.
 
I tinker here with hardware and software automation.  Automation is a hobby here and I like to be able to manage my own stuff whether software or hardware right here at home with the choice of no Internet dependences to turn on an automated light, use automated irrigation, security et al or remotely connect when I chose to remote control anything here with my cell phone. 
 
Many folks today just like the option of having a hardware hub that is connected to the internet 24/7 that they can get to with the cell phone for remote control 24/7.   They care more about the remote animated widget on their cellular phone than how long it takes to turn on a light.
 
It is easy button automation and cheap one light switch at a time (baby steps). 
 
Automating an entire home's lighting (nothing else) is still not a reasonably priced endeavor. 
 
Some day soon though you will probably be able to purchase a $5-10 automated light switch.

 

[wkearney99]

That's not quite the same thing. The level of interaction with streaming media systems is small and very infrequent and the worst case scenario is that you don't get watch Wheel of Fortune today or something. The requirements for automation, if it's serious automation, that requires low latency feedback, on which reasonably critical processes depend, and such, is quantitatively different enough that it's arguably not really comparable.

 
The average response time I've encountered with the Echo (between speaking a command and getting audio feedback) is typically no more than a 1-3 seconds.  For actions where the Echo is active (like playing an alarm or music) the device immediately mutes audio based on detecting the wake word.  This effectively ceases the action perhaps faster than if a full round-trip of commands was taking place.
 
One area where automation can suffer is lighting delays.  I don't use my Echo for demand-lighting (where I need illumination right now).  But I do use it for triggering scenes.  Scenes which are configured using the RA2 lighting software and triggered as a group.  I've read of others using lower-end lighting that make the mistake of trying to control the lights individually.  This leads to the 'popcorn effect' where the bulbs are coming up one-by-one instead of as a coordinated group.  This isn't a new concept with lighting, but it doesn't become an apparent issue until you get into automation.  Having a lighting system that properly supports groups/scenes and coordinated actions is key to avoiding a low WAF.
 
Whenever I hear technical-types deriding some new, low-cost and disruptive because of 'latency' I can't help but think of the dinosaurs bitching about all those annoying meteors and scrawny mammals...  Yes, plenty of things suck because of potential delays.  But you have to ask whether the long-drawn-out delays and expense of 'doing it right' are worth it?  
 
What's the saying?  "The perfect is the enemy of the good enough."
 
The trouble is the perfecting that goes on with many initiatives often fails to identify that which is most important to the end-consumer.  This is often the most tediously difficult part of the process... the user interface.  Sure, it's hard enough lashing all the technical bits together, but the last part of the 80/20 equation is often the most difficult.  Making it controllable, editable and usable by those not immersed in the nuances of the background engineering involved.  The more a product makes that seem like 'magic' the more the customers will flock to it.  Perfect or not.

 

[BobS0327]

Google Research http://research.google.com/ might be a good resource for all your theoretical computer science informational needs including but not limited to latency etc.

 

[Dean Roddey]

 But as a busy homeowner and business person, I am largely done with trying to manage things locally. Yes, I know there are risks and I accept those as reasonable and generally manageable. And yes, every blue moon something is not accessible, so there is a need for local "overrides" on some things. But my cloud solutions are MUCH easier to manage than my local solutions ever were, and when time is money and your most precious resource, this becomes a big issue. For me, the debate is over. Central process and storage is reliable and very convenient for me, and I am not sufficiently concerned with privacy to alter course, as my suppliers are well known companies with high levels of competency that exceed my own ability to employ and manage these functions. Rant on, friends. I have defected to the "It Just Works" - easily and at relatively low cost - camp.

 
But there's no such thing as cloud based automation. There's cloud based monitoring and reaction to things that are happening locally. In any serious automation solution there will still be LOTS of stuff local, and plenty of things that can still go wrong. It's not like running Office remotely, where all you have locally is essentially a terminal. 
 
Therefore the benefits aren't really the same, either. If that company cannot provide you with a solid state, reliable little controller for local control, then how likely is it that it can provide you with lots of sensors, thermos, cameras, media players, drapes controllers, and so forth that never fail? And of course it won't provide those things at all in most cases, because no company can afford to make all those things. So it's ultimately no different in terms of the effort required to create an automation solution, since the bulk of the effort is in getting all of that stuff reliably integrated.
 
Using the Echo with RA2 isn't remotely anything like cloud based automation, so how well it works or doesn't isn't really much of an argument either way. Cloud based automation would be when your RA2 controller was on the other side of the fence. How would you feel about that?

 

[Dean Roddey]

Whenever I hear technical-types deriding some new, low-cost and disruptive because of 'latency' I can't help but think of the dinosaurs bitching about all those annoying meteors and scrawny mammals...  Yes, plenty of things suck because of potential delays.  But you have to ask whether the long-drawn-out delays and expense of 'doing it right' are worth it?  
 
What's the saying?  "The perfect is the enemy of the good enough."
 
The trouble is the perfecting that goes on with many initiatives often fails to identify that which is most important to the end-consumer.  This is often the most tediously difficult part of the process... the user interface.  Sure, it's hard enough lashing all the technical bits together, but the last part of the 80/20 equation is often the most difficult.  Making it controllable, editable and usable by those not immersed in the nuances of the background engineering involved.  The more a product makes that seem like 'magic' the more the customers will flock to it.  Perfect or not.

 
Why do you insist on arguing that anyone who disagrees with you is somehow incapable of understanding and making use of recent technology? We've supported the Echo for a long time now. 
 
As to your other part of the last 20%, yeh, that is the hard part. And the fact that it's not been done yet demonstrates that it's really not doable. If any of these new doodads were actually doing that last 20%, you'd have a point, but they aren't. The IoTs world is not at all remotely there or even going in that direction. So that's where I don't get your point. All of these recent thingies are basically very simple ways to do very simple things, most of which, for most people, isn't even automation, it's just control on a smart phone.
 
Do you really think that all of the people who have been working automation for decades (and the enormous bucks spent) are too stupid to do something that some of these new companies can just pop up and take care of? They just aren't going to do it, because the problem isn't solvable by the things that these companies sell, nor is it solvable by the things that any automation companies sell. The complexity lies in the environment you are trying to control, and the fact that a truly smart home must of necessity be highly tuned to the needs of the specific customer, and once you go there the complexity factor goes up drastically.
 
Anyone, at any time, could have created solutions to make it easy to do very simple things, and of course some of those have been around for some time (like Roomie and things like that.) But that's a very low price market, with low barriers to entry and therefore lots of competition, and it still isn't going to cover that last 20% for anything like what would be considered a truly smart home.

 

[wkearney99]

Millions of dollars/pounds/francs/marks/etc and countless CENTURIES were spent in the creation of print and typography.  Desktop Publishing and the advent of xerographic transfer printing (aka laser printers) pretty much obsoleted them overnight.  So clearly the amount of time and money spent means little in the face of transformative disruptive technologies.  
 
Define 'smart home'.  Then address selling the consumer on that concept (aka marketing).  
 
Lots of people, clearly, want something 'in-between' universal remotes jammed into touchscreens and a Crestron/Savant/Homeworks/etc level of integration.  Anecdotal evidence suggests the high-end implementations are quite often negative factors for sales growth.  Ask anyone with a high-end implementation if they'd honestly recommend it to someone else...  Trying to 'do everything' for rich people is a well traveled path.  Trying to bring that down to the masses rarely works for the ones already succeeding in fleecing the rich.  
 
You speak of the IoT initiative as if where something 'too simple' for people to bother with.  Why?  Doesn't their 'cutting their teeth' at the low end build a smarter customer in the long run?  Clearly you're not prepared to sell to them now.  So obviously they've not going to be your customer, nor any of the suppliers above your price tier.  The smart money would seem to lie just ahead of that point.
 
Back to the original post, Google (among other companies) has had a long track record of buying something up and abandoning it.  It's good to see larger conversations being had about long-term viability of systems that depend on a continuing relationship with a vendor.  Eventually it'll circle around toward standardization and portability across providers.  While it's rough for the early adopters it'll hopefully lead to more robust solutions down the line.

 

[wkearney99]

Why do you insist on arguing that anyone who disagrees with you is somehow incapable of understanding and making use of recent technology? 

 
I don't.  That YOU make that assumption is not something I can control.
 

Do you really think that all of the people who have been working automation for decades (and the enormous bucks spent) are too stupid to do something that some of these new companies can just pop up and take care of? They just aren't going to do it, because the problem isn't solvable by the things that these companies sell, nor is it solvable by the things that any automation companies sell. The complexity lies in the environment you are trying to control, and the fact that a truly smart home must of necessity be highly tuned to the needs of the specific customer, and once you go there the complexity factor goes up drastically.

 
Here's where we disagree.  The complexity is not in the environment.  It's in the people using the environment.
 
Where large companies like Google, Facebook and Amazon have an edge is in taking user behavior into very serious account.  I'd venture more money gets spent by them on analyzing and predicting user behavior than most automation companies entire gross revenues.  This is where the cloud stands to hold an incredible edge over standalone implementations.  As patterns emerge they stand a much more fluid process of integration at the cloud-side vs a small box nailed to the wall at home.  Yes, a melding of the two would make incredible sense from a reliability standpoint.  But without active data analysis everything the sensors learn at home is going to be wasted.