Building an Open Source Nest

An anonymous reader writes "Google's recent acquisition of Nest, the maker of smart thermostats and smoke detectors, has sparked concerns of future plans for the devices, and how Google's omnipresent thirst for information will affect them. Thus, a team of engineers at Spark sat down and roughed out a prototype for an open source version of Nest. It looks surprisingly good for such a short development cycle, and they've posted their code on Github. The article has a number of short videos illustrating the technology they used, and how they used it. Quoting: 'All in, we spent about $70 on components to put this together (including $39 for the Spark Core); the wood and acrylic were free. We started working at 10am and finished at 3am, with 3.5 engineers involved (one went to bed early), and the only work we did in advance was order the electronic components. We're not saying that you can build a $3.2 billion company in a day. But we are saying that you can build a $3.2 billion company, and it's easier now than it's ever been before.'"

Comments

What?

[r.freeman]

What is the Nest. Do they mean like a natural nest build by bees or what - it is not clear form the summary, is it just me who doesn't find Nest an obvious thing like Apache or Linux that doesn't need introduction?

Re:What?

[_anomaly_]

https://nest.com/ ...and their blog post about being acquired by google: https://nest.com/blog/2014/01/13/welcome-home/

Re:What?

[Jeremiah Cornelius]

https://nest.com/ ...and their blog post about being acquired by google: https://nest.com/blog/2014/01/13/welcome-home/

Let's build a NEST clone, and spy on OURSELVES!

This message was brought to you Through a Scanner, Darkly...

Re:What?

[unixisc]

I agree. From the title, I thought it was about building an Open Source incubator or something of the sort

Antbot?

[spiritplumber]

OK, if this is the case, why in 2010 when we built an open source Android robot, the folks at Google literally told us that since they were trying to do the same thing, they would try to pretend we didn't exist? (They failed: It is hard to pretend something doesn't exist when it's humiliating you at Maker Faire, or making your hand bleed)

The hard part

[TheloniousToady]

The hard part isn't building a smart thermostat. The hard part is finding somebody simultaneously dumb enough and rich enough to pay $3.2 billion for a thermostat company.

Re:The hard part

[DickBreath]

The hard part isn't building a smart thermostat. The hard part is building relationships will all those energy providers.

Re:The hard part

[Lorens]

No, the hard part is writing a summary that doesn't leave the reader lost and perplexed at the third word.

Re:The hard part

[AthanasiusKircher]

The hard part isn't building a smart thermostat.

Meh. The hard part is realizing that you should NOT be trying to build a thermostat, period. Static temperature is relatively useless for comfort, which is why people end up moving the thing up and down all the time.

Our bodies don't sense temperature directly. They sense heat transfer, which involves evaporation rate of perspiration in addition to convection. This is the basis of "wind chill" (increased convection increases heat loss) and "heat index" (humidity reduces evaporation).

If there were actually a smart tech company out there designing such a thing, it would do something like keep a relatively constant dew point in the summer. The temperature is irrelevant. It can be 82 degrees and perfectly comfortable in my house, but on other days it can be 70 and unbearably stuffy. Cooling the house on hot non-humid days is stupid; having to adjust the thermostat down on cooler humid days just adds cost. (This is relevant in the winter as well. When it's really dry in the house, you often need a different temperature to maintain comfort than when humidity is at normal levels.)

It would be much more efficient to just stop the whole "thermostat" idea altogether... if we're really after "comfort" with least energy expenditure, why not program our houses to respond to what actually makes us comfortable (which is a more complicated formula taking humidity and temperature into account), rather than a scientific abstraction like temperature that has little human relevance?

Re:The hard part

[Registered Coward v2]

The hard part isn't building a smart thermostat. The hard part is building relationships will all those energy providers.

True. The real growth is not in home owners, most of whom will never replace a thermostat, let alone spend $250 for one or more replacements. The market is the installers and manufacturers to include it with a unit or as an add on sale. My AC guy gives away a $200 (retail) thermostat if you buy a multi-year service plan so it's not a stretch to see them offer a Nest unit.

Re:The hard part

[Qzukk]

1000x this. HVAC controls should be about comfort, not temperature.

I'd also love to see a "thermostat" with a "dehumidify" button: run for the next 15 minutes no matter what the temperature is. That'll fix both cool/damp and warm/muggy. And also feel great when I come in after the yardwork drenched in sweat and want to stand in front of the register with cool air coming from it.

Re:The hard part

[north.coaster]

All of this sounds good on paper, but what I would like to see is some detailed information on how to translate "comfort" into an algorithm that can be used to control a furnace and/or air conditioner. Then get the folks who developed the "open source 'nest' " to implement the algorithm.

Re:The hard part

[tgd]

The hard part isn't building a smart thermostat. The hard part is finding somebody simultaneously dumb enough and rich enough to pay $3.2 billion for a thermostat company.

Actually, it is... and even Nest can't manage to do it right. There's quite a large number of issues with the second generation Nest units failing -- and failing "on".

A thermostat should never, under any circumstances, be able to fail "on". That's a fundamental flaw.

This is great

[rtkluttz]

I actually only heard about Nest about a month ago and was VERY interested until I found out it was cloud based. I immediately typed a complaint to them about it. I'm very happy Google is heading this way with it. Even if Googles open solution is still cloud based, it should open API's and communication documentation so that people like me who are NOT interested in giving control of my house to a cloud app under someone elses control that can sometimes override by proxy. I'm a security concious guy and I simply do not want my homes firewall open to anyone but me. My phone or tablet should connect DIRECTLY to my in home equipment or server without anyone else having to be involved.

Re:This is great

[c0d3g33k]

What do you possibly have in your house that it's worth somebody's time to hack your thermostat?

The concern isn't necessarily about what's in the house (though putting the pets at risk due to extreme heat/cold could result from a hack). It's the information gathered by the thermostat(s) that has value. There's the usual metadata that can be added to your "profile" for purposes of advertisement. The Nest and similar devices can also reveal information about your habits, like when you are present/absent from the house, how long, time spent in each room (the Nest has a motion sensor). That could be useful for thieves, govt or law enforcement officials interested in knowing when the house is unoccupied or where you are likely to be should they wish to find you.

Re:This is great

[braeldiil]

Since I have a Nest (and love it), here's the scoop: It doesn't have to be connected to the internet at all. It would work just fine as a standalone thermostat. Most of the functions would be harder to use (only so much you can do when you have 1 button and a wheel), but as far as I can tell they're all there. The web access provides an easier-to-use interface to the Nest. I don't use it much - it did a good job of learning my schedule, so mostly I use the web interface to turn stuff on at the airport after trips. There's some reporting and statistics, but those are mostly fluff.

Re:This is great

[camperdave]

At risk: my other computer equipment and embedded devices.

Take a look at shodanhq.com, then perform the following steps.

1) Hack my Internet-accessible thermostat. You now have a conduit onto my home network. 2) Use the thermostat to attack the other devices in my house, or for that matter anything on the Internet. This has already been done (http://internetcensus2012.bitbucket.org/paper.html). 3) Do whatever you like. Steal information you consider interesting, sell access to any systems you compromise, or whatever. purpose. you. choose.

Moral: don't assume that the value of access to your devices is based solely on their *intended* use, or that your equipment will be compromised solely based on how interesting a target *you* are.

Want to sell a wifi-enabled thermostat? I still don't think it's a good idea at this time, but go ahead. Just ensure that the default settings have the radio turned off, and that I can leave it disabled forever. Yes, most people will still make bad choices based on convenience plus ignorance of the cost and risk. But it's a start.

--klodefactor

Hack my internet-accessible thermostat and you have a conduit to the thermal control subnet on my network. All you'll be able to access is my plausibly deniable torrent downloads.

patents ruining the day again

[StripedCow]

But we are saying that you can build a $3.2 billion company, and it's easier now than it's ever been before.

Were it not for patents...

What article?

[colin_faber]

Am I missing something here? Where's the link to the article referenced by Mr. AC.

So why didn't you do it first?

[Anonymous Coward]

It's pretty easy to build a version of most things once there's a working example in front of you - the real value is doing it first, not just copying.

Been there done that

[WaffleMonster]

Throughout the years I have seen instances of precisely this kind of arrogance in various forms.

Everything always "seems easy" at first glance on the surface. This is more often than not a reflection of gaps in ones understanding or failure to consider the problem space with sufficient detail.

The other major issue is failure to understand the sometimes monumental difference between building something that "works for me" vs "works for everyone".

Anyone can hack together an arduino that flips a relay when temperature sensor reads outside of a certain threshold and package it up to look like a cheap version of the nest. This proves precisely NOTHING in my estimation.

It's easy

[DerekLyons]

From TFS: It looks surprisingly good for such a short development cycle

It's trivially easy to *look* good - being functional is somewhat harder.

And building a 3.5 billion dollar company is just a *little* bit harder than writing a few scraps of code and soldering some bits together.

Re:It's easy

[Random2]

Soldering the bits together? Isn't that what Monster Cables do to get them there faster?

That seems to be worth a lot of money....

Re:the A/C companies are stagnant

[necro81]

build heaters and or air conditioners that can be controlled by USB or ethernet

Heaters and air conditions are largely operated by a single On/Off switch, controlling a 24VAC relay. It doesn't get much simpler than that. What would you want USB or ethernet for?

I don't understand the draw

[sideslash]

I love the idea of home automation, and have been involved in a couple significant DIY projects involving my own scripting (no custom electronics design, that's not my skill). I've focused on things like multi room audio and intelligent video surveillance. If somebody offered me a $250 thermostat (yes, I saw that they did it for $70 here), my response is "really? Isn't that kind of... boring?"
Don't get me wrong, I'm happy that they're innovating in the thermostat space. But I want much, much more than that.

Cold zones

[unixcorn]

I just bought a new thermostat. I really wanted a Nest because of it's cool factor however, I ended up buying a Honeywell. First, the Nest isn't as advanced; for example, the Honeywell has some features that allow me to run the fan periodically throughout the cycles. It also allows me to add an additional "slave" thermostat and average the temperature between my upper and lower levels. While the Nest allows you to view multiple thermostats in a single interface each stat required separate HVAC systems. The Honeywell also comes with a remote control that sense the temperature where you are sitting and will adjust the set point to make you comfortable. The bottom line is that sometimes new and cool isn't as good as tried and true when you actually do some research.

To avoid the need to wire...

[Anonymous Coward]

I can't see much use for USB, but Wifi would make a lot of sense. You could move your thermostat, or perhaps install an additional one, without having to run another low-voltage line down to the basement. It would also allow more sophisticated communication than ON and OFF.

Re:Typical Open Source Fan post

[kwalker]

Or, you know, use hand tools...

They're not saying they could build Nest for $70 in parts, they're saying they built a nest-clone device in less than 24 hours for about $70 in parts. Their time was theirs to spend, and they're not "marketing" this.

Re:the A/C companies are stagnant

[AvitarX]

For better control of the stages?

For information back on how often a boiler is actually burning when "on"?

For unlimited flexibility in the HVAC system (number of devices, zones, etc, not limited by number of leads on a device)?

Those are off the top of my head.

It's a fucking thermostat

[Anonymous Coward]

You set it. It controls the heater. Jesus fucking christ already, this is where we're at in 2014?

Re:That's what the woodpecker recently asked me.

[c0d3g33k]

What did the fox say?

Re:the A/C companies are stagnant

[mrchaotica]

First, newer ones can have dual-stage (or even variable-speed) compressors.

Second, it wouldn't hurt to have monitoring/instrumentation so that you can monitor the efficiency of your system and tell when things are wrong.

Re:the A/C companies are stagnant

[mspohr]

I just replaced my heating system with one from Carrier. It includes an "Infinity" thermostat which uses WiFi to connect to the Internet. I can control it over the Internet from my PC or phone as well as get notifications of status and service alerts. It has a nice display which shows the weather, also. Gives me historical energy use.
It's not as "smart" as the Nest (it's not watching me and my activity) which I think is a good thing.
Google wants the data and wants to be an intermediary with the energy companies.

WTF is an acquisitiona

[Hognoxious]

Acquisitiona? Isn't that Portuguese for credit note or something?

Re:the A/C companies are stagnant

[necro81]

For information back on how often a boiler is actually burning when "on"

I built an Arduino-based datalogger for my 5-zone heating system for exactly this purpose. It senses the states of the 24VAC and 120VAC relays and converts them into timestamped logical On and Off. The output is a CSV files on an SD card. I didn't go the extra step of putting it on the network, although that'd be pretty easy. The tricky part is the visualization of the data: I spent almost as many hours developing scripts to (offline) post-process and display the data as I did laying out the custom shield and writing the firmware. It'd be swell if I could accomplish that on a Raspberry Pi, serving up a furnace dashboard and interactive history. But, really, I'm not interested enough to go that extra step, nor do I have the time.

So, yes, it would be neat if this functionality were already built into the HVAC equipment, but such a tiny minority of customers would be interested in it that no manufacturer would add the cost of it.

Re:Replacing the software on the Nest

[chaim79]

I am ready to look into that, I have a nest (1st gen) and experience and tools in embedded development, looking at the mainboard (reverse side) for the nest there are plenty of touchpoints and even a set of contacts, how much you want to bet the JTAG interface for the Microprocessor is exposed letting someone (like me) install my own software?

Right now I'm looking around to see if anyone else has started this effort, no takers thus far but maybe that's just my search-fu being weak.