Smart Homes Alter Human Behavior | Hacker News Reader

Smart Homes Alter Human Behavior

spectrum.ieee.org

I have a lot of smart home gear but almost no automations for this reason. I’ve never had an automation that got it right enough to keep using.

Pretty much what remains are basic things like a text message if we left the garage door open overnight or our garden is thirsty for too long.

I have a few automations, but they are also explicitly set to certain values that I chose. No ML or anything. Works pretty well for me.

But then I also avoid "smart" devices that require internet connectivity.

Yeah, I feel like there is a huge missed potential for high levels of home automation that are not "smart", but rather just powerful and well-interconnected. I'd be happy to have a high level of home automation that is flexible, extensible, powerful, and just does exactly what I tell it to without having to "learn" or "guess".

Check out home assistant. It is also 100% open source (and the core is beautifully written and unit tested async Python)

Going way deep down the rabbit hole is possible with HA, and custom stuff on ESP32 or ESP8266 chips with your own sensors, relays etc is possible with ESPHome.

It’s good fun and very inexpensive.

Indeed it is. Incredible project

> It’s good fun and very inexpensive.

Once you have the equipment and the knowledge, it takes around $2 of electronics to turn any random simple device into a smart device. Speaking for myself at least, it took around $1000 of tools and 3 months of trial-and-error self-teaching to reach that point…

Definitely good fun though :)

Same here.

I had a Nest thermostat for a while and ultimately ended up using it like a dumb thermostat with 7-day wake/work/return/sleep settings. It kinda put me off on automation. Now I have my own home-made BLE / video IoT gadgets that I manage myself.

It is waaaay more coding & fiddling, but in a fun way and not a "I paid $500 for this crap" way.

Yeah, you just spent countless hours of your precious time so you have "I have no idea of the actual cost for this crap" now. ;-) I have a closet full of half, 3/4, 9/10 completed projects that were going to be so damn cool and cheaper. Be laid plans blah blah. However, with each and everyone of those not completed projects, I learned a lot and I have not regrets on any of it. I'm guessing you don't either on your IoT gadgets.

I put 100€ tag on my hour of tinkering or hacking. That’s how I even don’t start projects without commercial intention. It would be really cool to upgrade old Geotrax trains with ESP32, but such 40 hours experiment does not pay off.

That's sad that you feel that way. There is so much potential unknown that can be discovered in these types of projects. If you only ever do things that will make you money, well...I don't know exactly. Boring is the best I can come up with. You don't ever do things just because it might be fun? Or is it that you actually do know everything and there's no point?

I do lots of tinkering and hacking in my day job. I rather build something using well structured Lego instruction in my spare time. I was thinking about designing my own smart home units recently, but ended up buying quality lamps with motion detectors.

That's understandable. All I do is bang on a keyboard doing code for someone else. Luckily, I do find it interesting so it gets my all. However, there's something totally different in comparing what I do for a day job vs tinkering with DIY electronic projects, soldering, wiring, designing, programming, etc. However, time is limited which is why I have so many partial projects stashed away. None of it was critical which also contributes to the lack of completion status on some of them. There were other projects that did get completed however, and those definitely applied knowledge gleaned and/or parts from some of the incomplete projects.

Everyone's spare time is their own, and there's never not enough of it, so I do understand the consideration of how best to spend it when available.

To me I wouldn't call this tinkering.

Ha. I was starting to build a WiFi mousetrap to notify me whenever one of my mousetraps triggered. My cats bring mice into our sunroom, then let them go. Eventually one will slip inside the house or just die in the sunroom then start smelling.

So, me being the technologist, thinks that creating a WiFi enabled mousetrap will let me know if it triggered. I buy a Cricket WiFi module [1], some batteries and reed switches. Start messing around with it, then think to myself "this is dumb", it only takes me 1 minute to check the three mousetraps, whereas I was going to ruin a perfectly good weekend in building a WiFi mousetrap. I put everything in a box, might revisit it later. After working in technology Monday through Friday, I definitely didn't want to spend my weekends doing it either.

[1] - https://www.thingsonedge.com

I totally get the fun and satisfaction of doing it yourself, but if you have any interest in an off-the-shelf WiFi trap I’ve had a pretty good experience with these: https://www.victorpest.com/smart-kill-electronic-mouse-trap-...

I've seen those, and have been tempted to get one, only because I wanted to take it apart. :)

One of the biggest impacts on my life recently was adding an automation for my smart lights (Philips Hue) in the living room.

The lights start dimming over a span of 2 hours until they are their minimum at 22:00.

The result is that it just feels like it is time to go to bed. I have always struggled with getting plenty of sleep and not staying up too late and this has had a tremendous effect the past two months.

I like this too, but I found it fails badly in a very common use case: I want the lights on brighter because of <thing>. So I do that, but then the automation starts fighting with me because it wants to dim the lights again. Very frustrating.

One of the folks I play board games with has this installed. Great for when they want reminders to go to bed: Not so great when we need the lighting to better see the table. So much fiddling with the light - I generally wish there were a simple override for stuff like that and honestly, it has turned me off such a system.

I just started using the newer Phillips Wiz. It’s trivial to override the automated rhythm in the app. Normally $12-15. Black Friday sale $7/bulb but still might be available at Home Depot for $10/bulb.

Its a tradeoff. The automatic fade / transition of Hue is really well done. Its where the system shines IMO. But if you want to be able to interrupt the process, you lose the previously sent transition. You have to do all the logic manually if you want to be able to bounce between the two.

My experience is that if you interrupt a single transition in Hue in any way (i.e. changing the brightness during the “Fade duration”) that it simply cancels the transition and will not attempt to override what you have specified. I imagine the problem arises from if you have multiple “automations” for different steps of brightness, in which case there isn’t a good way for the system to know if you intended to override future automations too.

I guess the HomeKit integration with Hue is the problem then.

Its just how the Hue hub works

I am using Hue via HomeKit, and it works differently. The automations are not cancelled, they continue.

Yes, transitions are a single command to the hub, so any subsequent update will override your previous. You need to write a lot of logic that captures the current state and target state then retriggers the fade when appropriate (e.g. after a debounce).

I did not have this experience. Interrupting the automation once cancels it completely in Hue.

You are experiencing what the article describe. What if you add this common scenario to your system? i.e. Making this manual override normal to the system, starting to dim with a new calc a bit faster or shifting the final dim time (based on user pref)

I solved this problem with dim lamps and no other lights. When the sun goes down my natural light goes away, and the lamps are just the right brightness. Much easier and no fiddling with automation.

Well, of course adopting any piece of technology is going to change human behavior. The question this study ostensibly posits is whether it’s for the worse in the case of smart homes.

And although they seem to have worked hard to answer in the affirmative, the conclusions reported by the article are suspect.

First, if a smart thermostat takes more time to calibrate than a dumb one, then surely it won’t be widely adopted. The question, then, is how accurate the researchers’ model is compared to available technology.

And the conclusion that humans would “change the order of their tasks to minimize their time spent changing temperature” is as baffling as it is amusing.

We're somewhat conditioned that facebook and other "free" things are an exchange for selling our privacy.

But why do IoT devices, which we pay for, then STILL sell our privacy/freedom?

If IoT is the internet 3.0, then things aren't going to be good. Ipv6 and nigh infinite addressability is just going to be nigh infinite granular surveillance.

Honestly, the lack of interop and standards barriers is a good thing at this point.

And I didn't even talk about the hackability/vulnerability issues.

The margins on hardware are terrible. Multiplied by how few people care about home automation and you get a project that’s probably not making money.

The two ways to close that gap are cloud service revenue (paid monthly or with privacy) or to charge thousands instead of hundreds for a thermostat.

Also, an ESP8266 costs 5-10$ and all the software needed to build your own gear can be one-click installed with the Arduino IDE. That’s what preserving freedom and privacy costs today.

>But why do IoT devices, which we pay for, then STILL sell our privacy/freedom?

Because they can. Do you routinely leave money laying on the table?

I still miss my mercury filled thermostat. None of this constant on/off from the digital contraption.

Wait, what is wrong with a digital thermostat?

There can be such thing as too much precision. I have in-floor radiant heat that benefits from a longer duty cycle. I buy thermostats that are designed not just for a particular temp, but will also run without what I would call thrashing. E.g Constantly on and off because someone walked by and caused a breeze of slightly cooler air.

Not to say a digital thermostat couldn't account for this (you have to find them specifically), and are probably better with good firmware, but murcury will have a wider margin before activation.

> but murcury will have a wider margin before activation.

Well, digital thermostats will have as much of margin as you program them to.

Until your comment, I assumed every digital thermostat had hysteresis since the 90's. I can't imagine why they wouldn't. The more expensive ones have PWM control of the heater or heat exchanger, but that's something extra beyond what the mechanical ones did (metal strip thermostats used to be designed with hysteresis too, what isn't something they automatically have).

It's a bang-bang controller. It's so simple that it's what I came up with for controlling power producers / sinks to fill / drain buffers (load matching) in modded minecraft many years ago. It's the very simplest way to do this and is far from optimal. The only thing it optimizes is simplicity of implementation. A PI loop tuned to either minimize error or transitions is going to perform better for whatever goal or balance of goals a user wants.

A bang-bang controller can have hysteresis or not, it's as simple as turning off in a higher temperature than the one it turns on and putting a low-pass filter (a capacitor if you do it in hardware) after the temperature sensor.

The simplest possible implementation of an electronic thermostat would be able to switch thousands of times per second, and this is completely useless for controlling anything mechanical. When you are making real controllers, you always need some extra complexity, but you don't need a lot of it to make a good one.

Filters add more latency than PI(D) loops. They're not used in practice because they perform worse. Would you trust a 20% tolerance capacitor when setting your poles?

Search for a digital thermostat with hysteresis options to solve this problem.

Or a high quality one designed for in-floor radiant heat, e.g. Tekmar 552. Very simple to set up, very few gimmicks, and pretty much just works.

(The Tekmar smart home interface for it is offensively priced. The thermostat works fine without it, though.)

if true, this is kinda sad. I use a free utility that does this with my pc fans and lets me choose exactly what the margin is. how hard would this have been to add to nest?

What problem does adding hysteresis solve on a digital thermostat, though?

Is it a problem that it stays closer to the asked temperature?

If you've ever had a basic electronics course with a raw button, hysteresis might instead be known as 'debounce' or making the button's single human action a single 'press' result on the other side.

I think a digital thermostat could work well enough if they weren't made with a focus on being stupidly cheap or possibly somehow still absolutely energy efficient.

* Wake up and respond FAST to user inputs.

* Show the numbers it's thinking of big (running average over 5, 10, 15 min?), but also the current reading small (but readable) updated whenever it polls.

* If there's a hold or timeout involved display that too, expose the user to the thought process, don't just be a nearly opaque box.

* Be responsive in other ways. If I switch modes at the very least kill the old mode's 'calling' line right then, even if the new one gets a cooldown to prevent equipment issues.

But none of this seems to need hysteresis for the actual activation/deactivation of heating depending on temperature readings?

One reason I could see is to preserve the life of electromagnetic relays, but modern thermostats should use solid state relays.

These points also seem to be solved for any somewhat modern (less than 15 years old?) run of the mill thermostat. Generally they always show the current reading and give the temperature setting only when you ask for it though, that's the only difference I see with what you want.

The comment was something like 'walking by' disrupting the air at the thermostat and that alone being enough to turn the furnace on at that exact moment.

A running average, but also making it clear to the user (both the value the unit is using for the decision and the current instant input) as well as a 'running average' over a given window address that specific comment. I added my own frustrations on top since they're also part of the larger 'dumb down the UI' and make the device frustrating.

The same thing as all digital to analog device comparisons. There is so much more forgiveness in analog. I really feel for those that have never worked with analog and only know the coldness of digital 0s and 1s. Digital is like living in a world of only black & white where analog ha[d|s] shades of gray.

Take an Arudino with both analog and digital pins. A digital pin is either high (1) or low (0). That's it. On or Off. An analog pin can have a range of values from 0v to 3.3v or 5v depending on configuration. That single pin can be wired to several different items that each have a separate resistance allowing them to be identified by their passed voltage.

At the same time, analog has its own problems with that lack of precision, so everything needs to accept a range. Put a resistance on a 5v line, and read from it multiple times and you get multiple readings, but within a +/- tolerance.

ecobee has all the options you are complaining about, min compressor/furnace time, temp deviation, thresholds, and lot more settings.

[1] https://support.ecobee.com/s/articles/Threshold-settings-for...

Even a simple digital thermostat will likely have an input for cycles/hour. Maybe in the form of some dip switches on the back.

I still have my old one on a shelf in case I ever want to go back.

Is the mercury switch the important part of that, or the bimetal strip that's tilting the mercury switch? I think you could probably find mercury-free bimetal thermostats still, though I haven't looked.

The mercury switch behaved kind of like the first hill of a roller coaster. It would linger at the edge of transition until it reached a breaking point then it would drop.

Some modern thermostats try to emulate this behavior. The Honeywell T6 line of thermostats have technician settings for configuring this behavior though no technician takes the time to configure them for a home.

They still exist:

https://www.amazon.com/Honeywell-CT87K1004-Available-RoundHe...

Those don't have a mercury switch.

I should have spent more than 15 seconds Googling. You're right. In fact, after more googling they appear to be banned in most states due to the mercury content.

I got rid of my last mercury thermostat last winter because it occasionally failed silently, heating far beyond the setpoint. Mechanical operation is a significant liability once you come to expect solid-state reliability.

Smart cars alter human behavior: Look at self driving and tesla.

Goes to say every technology changes human behavior. Candlestick, oil lighting, electricity, etc.

The actual title says "How Smart Homes Alter Human Behavior." The submission title makes it sounds like it's news that they do.

I’m still waiting for the answer to the headline’s question.

I fiddle with my dumb thermostat all the time.

The problem? One of the discomforting things about living inside houses is when the air doesn't move. Yes, you can use a portable fan or ceiling fan to move air around, but that's normally at a speed which is TOO fast. If you sit in a room for any period of time and the air temperature coming from the vent doesn't change, then you may not even feel it blowing, which is almost as uncomfortable as being too warm or too hot.

I feel a better solution would be to set a temperature threshold and allow the thermostat to have a bit of wiggle room to increase the comfort level of those who are in the house by moving the air around longer and altering the temperature slightly, which is what would happen were you outdoors.

I believe it's normal in commercial buildings to have CO2-sensors which are connected to the control of the HVAC system, activating/increasing the ventilation when the CO2 levels are too high. It seems like this solves the problem you're describing.

I was going to suggest CO2 might be an issue in a room. CO2 in the bloodstream upto 30% saturation has a stimulating effect which make no ventiliation in bedrooms a common reason sleep is not restful, over 30% saturation level it has the opposite effect and can then be viewed like a noble gas and the effect noble gases can have on the body, namely it can kill and the body doesnt react.

You can set Ecobee thermostats to have min fan usage and use remote temp sensors to help keep temps consistent.

don't all thermostats have some kind of builtin hysteresis algorithm? so it does it's best to keep temperature steady... ideally without feeling the wind?

At least for me, I want to feel the air moving, if only just a bit. It's weird and unnatural for there to be no air movement at all.

The wording of the headline seems to be blindingly obvious, but the actual finding seems to be "usually they get it right but if not you'll spend more time fiddling with it than a regular one." Well, OK. Doesn't seem horrible.

Amusingly, they used A.I. to simulate human behavior in response to faulty A.I. behavior.

Another funny part:

“The researchers also explored scenarios where two humans with different preferences were in the smart home environment at the same time. In such instances, the humans actually changed the order of their tasks to minimize their time spent changing temperature.“

The horror!

> if the smart home system was just a bit off in predicting a human’s preference, the human and AI developed misaligned behaviors. “This led to more time spent by the individual changing temperature and activities than without the smart home system—the opposite of the smart home objective

This definitely happens to me. My clothes dryer is "smart" which means it refuses to just run for a fixed time and has some sort of smart detection of when to stop. It doesn't work well. Instead I have to run it several times before the clothes are actually dry. There's also no easy way to set it to a fixed time long enough. You have to mess around with the dial, long press another button, hit another button over 12 times - then that works for one load and it goes back to default "smart" settings.

One way my "smart home" has altered my behaviour is that I have to worry about keeping numerous devices up to date software wise and fix broken configurations occasionally. It's partly my fault because I have used devices from different vendors and ecosytems and tied them together with Home Assistant. It works most of the time but also breaks in sometimes unexpected ways. For example I have to regularly (every four to six weeks) reset my IKEA light bulbs because they start misbehaving by loosing the connection to my server. It has been a fun learning experience, but I would not recommend this kind of setup to someone less technically inclined.

Given these experiences I have a hard time seeing how AI could improve my setup. It would probably just make it more chaotic and even less predictable.

I finally ripped all the Wemo and Eero smart devices out of my home. They were unbearably finicky and kept not working for some minor reason and you’d have to just keep resetting everything. Wasn’t worth the effort. Now I just have a good old fashioned mechanical timer on the lamp in my living room. It never fails due to some random update or Wi-Fi setting.

Why should be AI in control, while its learning? Normal thermostats, wearables, phone/tv usage, and perhaps motion sensors should provide enough data points to tweak learning in passive mode.

Reply on news.ycombinator.com