Google's Nest service has been down once, twice, thrice, four times, no, scratch that, at least five times in five months, four of which were in the last few weeks. A similar thing happened toward the end of 2018. After each failure, a fix, an apology, more disgruntled users, and hours lost without any security recording for owners of the brand's cameras. Seeing the same headline with the same story every day proves that we can't solely rely on remote servers for the smart home, and local fallbacks need to be the first feature baked in, not an afterthought or a bonus.

I'm not singling out Nest here, but since we focus on Google's products on Android Police, we've been monitoring the brand a lot more closely than others and thus tend to be more aware of issues affecting its services. Outages have occurred in almost every server-reliant smart home device I've ever used, from Canary to Philips Hue, and are possibly even more frequent with security cameras (or maybe people notice and report them more there).

Nest outages reported on Downdetector in March 2020.

While most of these down times will be harmless for the majority of users, they'll still be detrimental to a small number of people. If the server is down and your smart camera doesn't catch a hit-and-run outside your house, or your alarm doesn't alert you of a burglary, the system has absolutely, irrefutably failed you. You'll never trust it again, will you?

And if the past few years with smart home gear have taught us anything, it's that server issues are inevitable. There's no if or maybe, the only question is when and how frequently.

10 important Nest outages since November 2019. (I removed minor ones with <1000 reports/hr.)

That's why local fail-safes shouldn't just be an afterthought that a few companies tack on because they want to appease a small percentage of vocal users, but an essential, legally-required feature (edit: I stated this at the end of the sentence, but for more emphasis, legal requirements should be for critical products and not all) for smart products on launch day, especially if that product is intended for security or if it provides a vital feature (gadgets for disabled or elderly people, for example). Just like the public at large has recently become more aware of privacy and security concerns, it should be clamoring for baked-in fallbacks or fail-safes.

Many of you reading this, specifically those in developed countries, have had little to worry about when it comes to connectivity and power. These server issues are the first time you've had to stop and think that maybe, just maybe, the "cloud" isn't all it was built up to be. But for me, living in Lebanon, with multiple power cuts a day and 16Mbps internet speed on a good day, the reliance on the cloud is the first thing I think about and research before I buy any product. I didn't get Hue lights until they added power-on behavior lest I woke up at 3am with my entire house lit like a Christmas tree. I bought a Nuki smart lock that can be opened from the outside and inside without an app. My smart Somfy blinds still have physical remotes. My Netatmo security camera records over a local SD card. Worrying about the products' inability to connect to the cloud at some point, and what would happen when they couldn't, is an integral part of my decision process.

Significant outages of 8 different brands in the last year, according to Downdetector.

Sadly, though, this concern has been dismissed by most gadget manufacturers. To them, cloud connectivity means paid subscriptions and longterm viability, but also less worry about local updates or building robust hardware. How could it not be appealing? A fail-safe is the smallest of their worries and the least eye-grabbing feature on their spec sheet, even though it's clearer than ever that it should be among the top priorities.

So what does having a fallback mean?

  • Any product should offer you the option to get alerted when it's down, whether it's because of a power blackout, an internet outage, or a server failure. You should not have to go to Downdetector or Google to see if issues are being reported or it's just you. You mustn't discover the problem when it's too late and your home has already been broken into or your dog has run away.
  • Products that can have local control buttons should still have them: your security system shouldn't just be a hub and an app, but offer buttons to turn things on/off the way it's intended. You shouldn't dump your TV remote and rely on voice controls for everything. Your microwave and oven and dishwasher should still have buttons to control everything, even if they come with a fancy app and Alexa/Google compatibility.
  • Products that rely on timers or schedules should save those schedules or timers locally and execute them regardless of whether or not they're connected to the server. A pet feeder shouldn't leave your cats and dogs without food, a smart pill dispenser or reminder shouldn't stop dispensing or reminding, and ideally, your smart lights and blinds should still follow their schedules without needing the server to issue that command every time.
  • Products should offer local control when unable to connect to their servers. If said product and your phone (with the installed app) are on the same WiFi network, you should retain access to the most essential features in the app. You don't need every bell and whistle in this mode. Livestream your camera, arm/disarm your security system, turn off your lights, lock/unlock your door, etc... These aren't big requirements and can easily happen without the cloud as intermediary. Google has promised local control through the Home since mid 2019. As far as I know, that's nowhere to be seen, or at least the companies that support it can be counted on the fingers of one hand. Besides, you don't need Google to allow that — even if it's a cool bonus, not everyone will have a Home speaker — and it should be a direct access from the app to the product.
  • Vital or critical products should offer basic processing at the edge when servers are inaccessible. Whether your internet is down or the company messed up something, you need to know you can still rely on your products. A smart camera has to be able to detect movement and record incidents locally (either built-in storage or SD slot), even if they're just static snapshots, and even if there's no person detection or zone control or any of those fancy cloud-based features. That won't come to the detriment of cloud subscriptions, but will compliment it. (I know that dozens of silicon chips are announced every year, promising better edge processing, local face recognition, and less reliance on the cloud for voice or image analysis. Technically, if manufacturers wanted to, they could move all of these processes locally, but I doubt we'd see that happen as it'll justify their subscriptions less and less. So I'm asking for a small fallback, which wouldn't jeopardize their business models.)

When the smart home was still a geek's expensive hobby, everyone who decided to venture into did their research, understood the risks, and made an informed decision about it. Now that smart home products are cheaper, more powerful, easily available, commoditized, and replacing every appliance in our house, it's become paramount that we look at their reliability. Most people buying them won't think twice before going for a brand spankin' new smart doorbell or light, and they won't know that they could fail because of a server outage until it's too late.