Clay Morton

November 16, 2026 · 14 min read · networking, hardware

You Don't Need Internet For This

Why is it that we insist on connecting everything we own to the internet?

Laptops. Phones. Televisions. Lightbulbs. Refrigerators. Toasters. Toothbrushes. Where do we draw the line? Do all of these things really need to be connected to the internet? For some crazy reason, I don’t feel like my toothbrush needs to be on my Wi-Fi.

I set up enterprise-grade networks at Digital Dolce Studio and have had to connect more “smart” devices than I care to count, including in my own home. So before I shred them, just know, I am not anti-gadget. I do, however, think that we skipped over an important question along the way: what does putting this device online actually get me? Occasionally, the answer is “a lot” but more often than not it’s more like “I will open this app once to connect this and then delete it.” Even worse, sometimes the device requires a subscription to run the software on the hardware you already own! Oh, and it stops working when a company you’ve never heard of goes out of business. You can also unknowingly introduce security holes into your network simply by trying to set up a smart device. Let’s talk about setting these things apart.

Smart ≠\neq connected

Before we discuss this, I feel it’s important to rip apart two terms that aggressive marketing has smashed together.

A smart device does something on its own that a dumb one can’t. Think, a thermostat that learns your schedule, or a light that turns on when you walk into the room, or even a sprinkler that skips watering when it rained yesterday. A connected device is something that communicates with the internet. While these do overlap, they are not the same.

Programmable thermostats have existed for decades without Wi-Fi. Motion-sensor lights have been in garages and hallways forever, and they don’t need an account. Plenty of “smart” behavior can happen entirely inside the device or on your own home network, without ever reaching a server over the internet. The job that it does is largely the same as a device you would connect to the internet, and yet it doesn’t need internet to do it. So, when you see a company saying a “smart” device, a lot of the time what they actually mean is “connected to our cloud.” Is it really worth it though?

What do we gain from this?

In fairness to these connected devices, because they do some really convenient and helpful things, here are some things we can gain from connected devices:

  • Control from anywhere. Turning the heat up on the drive home, checking whether you locked the door, or letting a family member in when you’re not there.
  • Notifications. A leak sensor under the water heater that texts you before the basement floods is worth every penny. So is knowing the garage door has been open for three hours.
  • Automation. Lights that dim at sunset, a fan that kicks on when the humidity spikes, a porch light that comes on when the camera sees motion. Once it’s set up, you stop thinking about it, which is the whole point.
  • Information. Energy monitoring can show you which appliance is quietly eating your power bill. That’s useful data you wouldn’t get any other way (thank you stingy power companies).

If a device gives you one of those in a way that actually makes your life easier, great. Connect it. But every one of those benefits comes with a cost, and the cost almost never shows up directly on the box. It normally is deep in that end-user license agreement or terms of service contract that you agreed to without reading.

Every connected device is a computer

This is the part people don’t think about. A smart plug isn’t really a plug with a chip in it. It’s a tiny computer, running an operating system and software someone wrote, sitting on your network next to your laptop, your phone, and whatever you use to do your banking. And just like every other piece of software, that software has bugs.

The most famous example of this going wrong is the Mirai botnet. In 2016, it took over hundreds of thousands of internet-connected cameras, DVRs, and routers, mostly by trying a short list of default usernames and passwords that nobody had ever changed. Then it pointed all of them at a DNS provider called Dyn, and a big chunk of the internet, including major sites people use every day, went down for hours on the East Coast. The people who owned those cameras had no idea their devices had picked up a second job.

There’s another story that gets passed around security circles: a casino that was reportedly breached through an internet-connected thermometer in its lobby fish tank. Whether you picture a botnet or a fish tank, the lesson is the same. The weakest device on your network sets the security level for everything else on it, and the weakest device is usually the cheapest one.

Some governments have started noticing. The UK now requires connected products sold there to stop shipping with default passwords and to tell buyers how long the device will get security updates. That’s a good start, and it’s also a little alarming that it needed to be a law, but that’s beside the point.

Who’s updating your fridge in year nine?

Should our washing machine need software updates? That’s not a rhetorical question anymore. If a device runs software and talks to the internet, it needs security updates for as long as it’s plugged in. Otherwise, every bug anyone finds in it stays open forever.

Here’s the problem. Phones get security updates for a number of years, and the companies making them have gotten a lot better about supporting mainstream devices longer. But a refrigerator is supposed to last ten or fifteen years. A washer and dryer, about the same. A furnace or water heater, longer. Those companies are not software companies, and keeping old firmware patched for a product they stopped selling years ago is not where they make their money.

So you end up with a box in your kitchen that will keep working perfectly as a fridge for another decade, running software that stopped getting fixes a long time ago, and still connected to your Wi-Fi because nobody ever thought to disconnect it. Multiply that by every connected appliance in every house in a neighborhood and you can see why botnets keep finding new recruits.

When the company leaves, the device stops working

This one bothers me the most, and sadly isn’t very hypothetical anymore.

A lot of smart devices don’t actually talk to your phone directly. Your phone talks to the company’s servers, the servers talk to the device, and the device does what it’s told. That’s convenient for the company, because it lets them add features and collect data. It also means your device has a dependency you can’t see, and you have no control over it.

In 2016, Nest shut down the servers for the Revolv smart home hub, which it had bought a couple of years earlier, and the hubs people had paid for stopped working. In 2020, Wink told its customers that they had about a week to start paying a monthly subscription or lose access to the smart home hub they had already bought. In 2022, Insteon shut down so suddenly that people woke up to apps that couldn’t control their own lights and switches.

Think about what actually broke in those cases. The bulbs were fine. The wiring was fine. The switches on the wall were fine. What broke was a server in someone else’s building that had very quietly become a requirement for turning on a light in your own hallway. That’s a strange thing to sign up for, when the old light switch worked for fifty years without a login.

The features you already paid for

There’s an even worse, borderline silent, version of this that’s getting more common: features that live behind an app, a subscription, or a firmware update you didn’t ask for.

Printers are the classic example. Some printer makers have pushed firmware updates that stopped third-party ink cartridges from working, on printers people already owned. Some newer appliances keep certain cycles or settings behind an app, so the buttons on the machine don’t give you everything the machine can do. And a lot of camera companies give you the camera for a reasonable price, then charge monthly to actually save and review your footage.

None of that is illegal, and some of it might be totally fine with you. But it changes what “buying” a product means. You’re not really buying a device at that point, you’re buying a relationship with the company that made it, and they get to change the terms whenever they want.

Your TV is watching you back

Privacy deserves its own section, because whether you’re aware or not, connected devices are very, very good at collecting your data.

A lot of smart TVs use something called automatic content recognition, which identifies what’s on your screen, whether it’s coming from a streaming app, a cable box, or a game console, and reports it back so the data can be used for advertising. Back in 2017, Vizio paid a settlement to the FTC over collecting viewing data from millions of TVs without people meaningfully agreeing to it. Most TVs now ask during setup. Most people click “agree” because they want to watch TV.

Cameras and doorbells are an even bigger deal. In 2023, the FTC settled with Ring over employees and contractors having had access to customers’ private videos. Ring has made changes since, and I’m not trying to single out one company. The point is that when your camera footage lives on someone else’s servers, your privacy depends on their policies, their employees, and their security, not just yours.

Here’s my rule of thumb: If a device has a microphone or a camera and it sends what it records to the cloud, assume someone other than you can eventually see or hear it, and decide whether that’s okay with you. Sometimes it is. Sometimes it really isn’t.

Why I put smart devices on their own network

Should we call a device smart if it needs its own IoT VLAN on our home network to operate? Probably not, but that’s how I set them up anyway, and there are two good reasons.

The first is security. If a device is going to be a weak spot, I want it as far away from everything else as I can get it. Putting smart devices on their own network (or a VLAN, if your equipment supports it) means a compromised plug or camera can still reach the internet, but it can’t go poking around your laptop, your NAS, or your work computer. It’s the same idea as keeping the guest bathroom away from the safe. If your router doesn’t do VLANs, check whether it has a dedicated IoT network option. A lot of newer mesh systems do. If not, the guest network does most of the same job, because guest networks are usually isolated from your main one already.

The second reason is just practical. A lot of smart gadgets only work on 2.4 GHz Wi-Fi. Most newer routers combine 2.4 GHz and 5 GHz under one network name and use band steering to push each device to whichever band the router thinks is best. Your phone happily sits on 5 GHz, the gadget can only see 2.4 GHz, and setup fails over and over for no reason the app will ever explain to you. When I upgraded my parents’ network, that’s exactly what happened to most of their smart devices. A separate 2.4 GHz network just for smart devices fixed it, and it’s been solid since.

There is one tradeoff that may be a dealbreaker with that idea, and it’s with things like casting to a TV or AirPlaying to a speaker. These protocols expect your phone and the device to be on the same network. If you isolate your TV completely, you may have to allow that traffic through or keep the TV on your main network. It takes a little thought up front, but it’s worth it.

Local control is the way!

The good news is that the industry has started moving in the right direction on most of this.

Matter, the smart home standard backed by Apple, Google, Amazon, Samsung and a lot of others, is built so that devices are controlled over your local network. Many Matter devices use Thread, which is a low-power mesh network that runs between devices in your house, with a “border router” like a HomePod mini, an Apple TV, or a Google Nest hub connecting it to the rest of your network. If your internet goes down, a well-set-up Matter home keeps working, because the commands never needed to leave the house in the first place. (I have some opinions about how “universal” Matter really is, but that’s a whole other post.)

If you’re a little more hands-on, Home Assistant is open source software that runs on a small computer in your house and controls almost anything, with your data staying on your hardware. For cameras, systems that record to a box in your house, like a local NVR, keep your footage under your roof instead of on someone else’s server, and they keep recording when the internet drops.

The common thread (pun intended) is simple: the more of your smart home that runs on hardware you own, the less of it depends on a company’s server, business model, or privacy policy.

What this looks like when you build it yourself

I don’t just complain about this stuff, I’ve had to make these same decisions as a developer. When I built Cadence, my iOS app that plans your day around how recovered you are, the easy route would have been the route most apps take: send everyone’s health data and calendar up to a server, run it through an AI model there, and send the plan back down. I didn’t want to do that, and honestly, I didn’t think I should do that.

Cadence reads your readiness from HealthKit and your calendar from EventKit, both of which already live on your iPhone. Then it builds your day using Apple’s Foundation Models, which run on the device itself. Your sleep, your heart rate, and your meetings never need to leave your phone for Cadence to do its job. I don’t run a server for it at all. There’s no database of my users’ health data sitting somewhere waiting to be breached, because I never even built one.

The internet only comes into the picture in two places. If you use Cadence on both your iPhone and iPad, it syncs between them through your own iCloud account. And every once in a while, a request is too big for the on-device model, so it goes to Apple’s Private Cloud Compute, which Apple built so that requests aren’t stored and can’t be seen by Apple. That’s the internet being used for the things that actually need it, instead of being a requirement for everything.

That choice made the app harder to build for sure, but data is secure and it is yours. Getting data to sync between an iPhone and an iPad without writing it twice, keeping background updates from draining your battery, and fitting the AI work into what a phone can do on its own were the hardest problems in the whole project. A server would have made a lot of that go away. But it would have also made Cadence one more thing that depends on a company’s server, and one more place your personal data could leak from.

Here’s the test from earlier, applied to my own work: what happens if the company disappears? If Digital Dolce Studio vanished tomorrow, the copy of Cadence on your phone would keep doing what it does, because there’s no server of mine for it to lose. That’s the standard I’d like more “smart” products to be held to, and it’s one I try to hold myself to first.

What I’d connect, and what I wouldn’t

Everyone’s going to land somewhere different on this, but here’s roughly where I land.

Worth connecting: thermostats, leak and freeze sensors, garage door openers, and cameras that record locally. These either save you money, prevent damage, or answer a question/concern you actually have when you’re away from home.

Maybe, if they work locally: lights, plugs, blinds, and door locks. Nice to have, and really useful for automation, but only if they keep working when the internet or the company doesn’t. A smart lock in particular needs a physical key backup. I don’t want my front door depending on a server.

Probably not: the toaster, the kettle, the toothbrush, and most of the fridge. If the connected version doesn’t do its main job better, I’d rather have the version that just does the job.

Before you buy anything “smart”

I’m not telling anyone to rip the smart bulbs out of their ceiling. I’m saying it’s worth asking a few questions at the store, or rather before you click “buy”:

  1. Does it still work if the internet goes out? Look for local control. Matter and Home Assistant support are good signs.
  2. Does it need an account just to turn on? If a lamp needs me to make a profile and agree to a privacy policy, I’m already skeptical.
  3. How long will it get updates? If the company won’t say, assume not long.
  4. What happens if this company disappears? If the answer is “it becomes a paperweight,” that’s worth knowing before you pay $400 for it.
  5. Is there a subscription, now or later? Check what works for free, and what’s quietly behind a monthly fee.
  6. Does it have a camera or a microphone? If so, where does that recording go, and who can get to it?
  7. Does it actually need to be connected? This is the big one. A toaster’s job is to toast. A toothbrush’s job is to brush. If the connected version doesn’t do that job better, the extra features are mostly a way to sell you an app.

Smart should mean it makes your life easier

The smartest device in my house might be a light switch. It works every time, it doesn’t need updates, it doesn’t need an account, it doesn’t collect any data, and it will keep working long after whatever company made it is gone.

That’s the bar I want smart devices to clear. Connect the things that are genuinely better online, keep them on their own network, choose devices that keep working without someone else’s server, and let the toaster just make toast.

And no, my toothbrush is still not getting the Wi-Fi password.