[a3n]

Roughly OT: I have yet to have seen anything that makes me look forward to the IoT, and when I talk to the voices of vendors in my head I'm basically telling them "Stay out of my refrigerator, my furnace, my toilet and my smart doorlock that I'll never have."

I just don't feel that anything relevant to the IoT is missing from my life. At all.

[wlesieutre]

I can see "Check if I left the stove on and turn it off remotely" being a neat thing to have. But not neat enough to let a 10 year old appliance with no software updates and the capability to burn my house down be accessible to the internet.

[exelius]

Yeah, but chances are you won't use the functionality often enough to configure it in the first place.

That's the real problem with the Internet of Things: most of the things we own are not all that useful when we're not in close proximity to them. Thus, not only are users and manufacturers unlikely to update them in the future; users are just as unlikely to connect the thing in the first place.

Home automation through things like light switches, etc. has a use case, but those products have been available and Internet-connected for over a decade and we still haven't seen wide adoption. I recently priced it out -- it would cost me over $5000 to swap out the outlets and switches in my house for Insteon devices. And that's just the hardware; not the electrician required to connect it all or the time I would spend configuring everything. Home builders aren't going to spend that kind of money building this into anything but the most high-end homes -- the IoT hardware alone blows through the fixtures and appliances budget that most home builders allocate.

People want systems that "just work". IoT does not "just work", and none of the current or announced implementations address the big problems around configuration (namely, every house is different so every implementation is custom). And in some cases like a stove or a refrigerator, any amount of configuration is going to be too much.

[mturmon]

About home automation of light switches: I put in ~10 X-10 devices, plus the ~20 controllers to activate them, circa 2002. I did it myself (I'm an EE), so the installation cost was 0.

It worked OK for a while, but the setup was not robust and eventually some controllers would not work some devices at some times. The annoyance factor in going from 0 errors to a 1% error rate is HUGE.

Five years passed, and I have been slowly replacing all these X-10 devices with hard-wired switches or with Insteon. Of course, the original 1959 wiring paths (12 gauge Cu FTW) still work fine.

Now, when I see connected/automated homes in design mags, all that tech seems more like a long-term maintenance headache than a desirable feature. If I had an unlimited budget, I would not build those features in, I would just install conduit and run old-school copper wires through it.

My lessons: (1) The design life for a home is decades, the refresh rate for home automation devices is years; (2) Upgrading/tinkering is fun the first time -- but only the first time; (3) Your spouse hates it more than you do; (4) The existing device does one thing without fail, replacing it with a device that does more things but sometimes fails is not a net gain.

[wlesieutre]

Agreed on all counts. The only ones I can see taking off in any sense are easy to retrofit things like light bulbs and the occasional wireless plug controller like a WeMo if you have a particular load that needs one, not as a standard on every outlet.

But prices need to come down, light output needs to go up (wireless bulbs seem to top out around 60W equivalent), and switches need to not be a $60 optional accessory (looking at you, Hue Tap). Controlling lighting with your phone is neat, only being able to control lighting with your phone sucks.

Long term, I'm sure I'll end up with more IoT devices. But it'll be because they got shoved down our throats and I didn't want to pay more to avoid new "features," not because I wanted a wireless microwave.

[exelius]

Insteon seems to have the best ecosystem of IoT stuff. But you're right; the lack of compatibility across all these different "open" standards is frustrating as a user. You basically have to pick one platform and stick with it.

But I'm personally not convinced by IoT. Every implementation I've seen adds complexity without really improving functionality. I have a few Insteon switches in my house, but only in a few places where I need them (e.g. on the lights in front of my house so they can be turned on with a timer when I'm out of town).

I actually have a "connected refrigerator" made by Samsung. I tried for 5 minutes to get it set up before I gave up. I really couldn't think of what I would need to use a connected refrigerator for. IMO this is the usage model for the vast majority of IoT devices: if I, a major geek about this stuff, am not willing to spend more than 5 minutes setting it up, who actually cares enough to bother with any of it?

[wpietri]

My guess is that for each person some small thing will make sense, which is sufficient to drive the costs down and the technology forward.

For example, good sleep is really important to me. So I used the Philips Hue bulbs and a NUC to build a smart home lighting system that behaves much more like daylight. When I describe it to people, quite a number really want it. Not bad enough to get my code off GitHub and install it, but they'd pay something extra for it. And I now see pre-packaged commercial products getting proposed, so I'm sure they'll have options.

The interesting thing happens when that cycle drives the costs a fair bit lower. Look at phones, for example. Smartphones were a weird, exotic thing. Then they were a high-end consumer thing. Now they're the default. A couple years back I went into a store looking for a cheap phone and asked for one without a web browser. The clerk looked puzzled and said, "Well, they all come with web access." I'm sure I could have ordered a dumbphone somehow, but it would have been harder and cost more.

So my question is: how much extra will you pay for your house not to be connected? Because that's the real test for me about where the IoT thing will end up.

[exelius]

I don't think so. Cost is one side of the equation, but usability is another. And at the end of the day, I don't think there's enough consumer demand to push IoT where it could be. I think eventually we'll get there -- at some point in the future, it will just be cheaper to buy an Android SoC than a handful of cheap analog micro controllers -- but for the next 10+ years we're going to keep hearing about IoT as "the next big thing" because there's no consumer demand for it.

The situation with your daylight system is a perfect example. People think "Oh, that's cool" but won't actually get off their ass and spend a few hours setting it up and configuring it to their liking. Unless it comes out of the box, it's a non-starter.

IoT devices as they are today require a systems integrator to come in and tie everything together and configure them. So your lighting system is integrated with a presence system that is also integrated with your thermostat. Each of these has to be configured on a case-by-case basis, because no two homes are alike. But if your products require an integrator, suddenly the integrator is your customer, and you begin sacrificing end-user focus for things that make the integrator's job easier.

And then you realize that the largest consumer of IoT platforms isn't consumers themselves, it's installers like ADT. Their focus is on selling simple products that require minimal support, not feature-rich ones. Why? Because users simply aren't interested in paying any amount of money for advanced features. Until that changes, IoT is going to be a niche hobbyist market. Even if it's built in to every device you buy, if it's not worth anything to you, you're not going to configure it in the first place.

[wpietri]

I think we might agree entirely. I think mobile phones were a relatively fast cycle because people were pretty happy to replace their phones every few years. That's not true with, say, washing machines or furnaces. So it will definitely take a while.

But if you look at devices with cycles in between, I already see it happening. 10 years ago when I bought a stereo receiver, I bought something entirely dumb. Last year I replaced it and ended up with something that was internet connected. Not because I really cared, but because the equivalent model came with that. And in retrospect I'm glad; their phone/tablet app is a way better remote control than punching a bunch of mysterious, no-feedback buttons. The same thing happened with my DVD player; when the old one died I just bought whatever Consumer Reports recommended and it too is internet connected.

I think that you're right that the development of pluggable Android is what will push this forward. And Google clearly agrees; their project Brillo is surely one effort among many.

[exelius]

Right; but in the case of a stereo receiver (or any TV-connected devices), the IoT model makes sense: this is a device you interact with on a regular basis. There is already an ecosystem of devices that support things like Bluetooth, AirPlay, etc. so when your receiver has them, it's actually kind of useful.

I've always thought that the reason for keeping Android as open source was to enable this use case. Eventually, Android-capable SOCs will come down to a very low price point, and for these use cases they don't need to push much more than a few kb per second. But it retains the advantage of having a development ecosystem that is well understood and widely available (you can't throw a rock in China or India without hitting an Android developer).

[walterbell]

> I used the Philips Hue bulbs and a NUC to build a smart home lighting system that behaves much more like daylight

Neat. Did you try to coordinate room lighting color changes with computer screen colors controlled by fl.ux?

[wlesieutre]

Flux can actually do that directly! https://justgetflux.com/news/pages/bigupdate/

[wpietri]

I didn't coordinate it, but they are roughly in sync because they're doing similar things. Code here:

https://github.com/wpietri/sunrise

It's minimally sufficient for my needs, but I'd be glad to work with others to expand it.

[semi-extrinsic]

I would say only the DIY part of the IoT is something to look forward to. When all it takes is $20 and a few lines of Lua to set up a display, some sensors etc. that's connected either to the actual internet or just your private wifi, that's actually a cool thing. Basically any time you've thought "Hey, I could use an RPi and automate this.. Nah, they're too big and expensive."