Hacker News Reader

Comment by nicktelford | Hacker News Reader

This stream of tweets reads like satire, but it demonstrates an extremely important point: the "Internet of Things" will never become mainstream so long as every appliance is a walled garden.

Without open protocols to allow people to connect together whatever they like, this technology will remain a niche for the minority that know how to write the software themselves - and have the patience to.

View on HN

I think the bigger point is that smart devices should gracefully degrade to dumb devices.

But then you wouldn't have to buy another one.

YES!

Light switch not online, still switches lights. Smoke detector not connected to wifi, still beep when detecting smoke.

The "smart" part is for convenience, not to create unnecessary dependencies. I'm a bit tired of seeing things that obviously only need access to local resources fail when the internet goes out.

Very good point. It always disappoints me when I encounter systems that seem to ignore the Fallacies of Distributed Computing[1].

1: https://en.wikipedia.org/wiki/Fallacies_of_distributed_compu...

I spent 5 years programming around those 8 fallacies for a web-conference company (ok maybe 3 fallacies reworded). Bottlenecks, network topology changes, different rules on different networks. Means you have to measure each connection and code multiple strategies. Probably should open-source that stuff, because its a growing problem and no sign of relief in sight.

the thing I keep failing to understand is: why do these devices need to block while updating? Can't they just use two (small) memory-banks and download the image to bank 2 while working from the first one and do one quick switcheroo in a couple of milliseconds?

(edit: question marks should mark the end of sentences which are questions)

From the security we've seen in these devices, I'm assuming that would be both a financial and technical no-go. The standard practice appears to be buying a minimum-cost board and loading it up with some stolen open-source software - sensible hot-swapping would be far too demanding for that.

(But yes, that really should be the case. It could also support elegant failures back to the previous firmware version.)

IoT devices seem to be the result of putting web people in charge of embedded development. Not to defend emb. devs... some truly horrendous stuff out there as well.

I think it's the opposite; they put the embedded developers, who (rightly) have no sense of the insane challenge of doing security, in charge of talking to the Web.

I disagree with that characterization. Engineers working on connected devices, Web or not, should be cognizant of security issues. That many aren't I see as a problem.

For that matter, I don't think most web developers have a proper sense of security challenges, either.

Does it have to be either/or? You need both types of developers, and they have to talk to each other, and you have finite hiring resources. These are very different cultures, and the failure mode is either broken system software, which will get noticed and fixed, or broken security, which likely won't.

You don't need that. Linux does that without using two separate memory banks.

(You only need to reboot for kernel updates, and that can be reduced with kexec)

Your idea is pretty similar to having a hypervisor running two VMs and switching from one to the another. There's no need to have two separate memory banks.

With an IC, it makes sense to work from two different RAM ICs. I'm not an electronics engineer, however.

Program flash is usually ROM and that is where you are executing the SW. You mostly keep variables in RAM, although you can do a couple of neat tricks with copying and fiddling with code in RAM.

Even more, while you usually can self-program the ROM memory, you can only erase it in blocks that are pretty big relative to the whole memory of the device. If you have a simple system that runs in a flat address space with no MPU and virtual memory, it gets pretty complicated for a running update.

Of course, that's from my experience with microcontrollers. SOrry if I went off-topic, but I just thought the discussion interesting. Embedded Linux computers I bet are more complex than that and probably execute code in RAM. These babies are connected to the internet so they probably have a far beefier SOC running Linux not some lightweight RTOS. It is a kettle, you need a lot of juice for that baby!

The kettle in the story does do that. The user simply chose to continue to try to get it working as intended, rather than pressing the boil button on the kettle base and going on with his day.

Any examples of devices that don't??

> it demonstrates an extremely important point: the "Internet of Things" will never become mainstream so long as every appliance is a walled garden.

I would say the point is "it is absolutely moronic to put a computer in a tea kettle."

I mean, seriously. What benefit can you possibly get from doing that? "Oh, I can turn the kettle on from anywhere."

Wonderful. For that benefit I would maybe pay $1. It's not like it's going to actually make the tea and bring it to me.

But how much would I pay for the assurance of absolutely perfect, air-gapped security from having hackers burn my house down by boiling my kettle dry, PLUS absolutely perfect assurance that I won't spend a second updating firmware or WiFi configurations or antivirus software on my kettle, PLUS absolutely perfect assurance that nobody is building a dataset about my life based on my kettle usage, PLUS absolutely perfect assurance that my kettle won't stop working because some service shut down or because some programmer mishandled memory...

What, you say? Those features are impossible if a kettle is connected to the internet?

People keep using this word "Smart". I don't think it means what they think it means.

Literally, the whole point is to be in Star Trek, to be able to say this: https://www.youtube.com/watch?v=R2IJdfxWtPM

... except you still have to manually put water in the kettle, and put the tea bag in the cup, and pour the water on the tea bag, and add and mix sugar and/or milk yourself.

Oh, and it doesn't actually have voice control. You have to pick up and unlock your phone and use an app, during which time you probably could have walked to the kitchen and turned on the kettle.

The Future™, ladies and gentlemen.

    > Without open protocols to allow people to connect
    > together whatever they like, this technology will remain
    > a niche for the minority that know how to write the
    > software themselves - and have the patience to.

That's me, and a good chunk of the HN crowd, but there's also people that will be happy to buy in wholly to a single ecosystem.

And unfortunately, that's actually more justified here than the "I'll get an iPhone/Mac because I have a Mac/iPhone so it'll work better" mindset that I'll never understand.

    > Without open protocols

That doesn't seem to be enough - Apple's HomeKit is... "extensible", shall we say; Amazon's Alexa has an open API; Google's Weave is an open standard, but the schema's pretty lax (vendors are "encouraged" to use standard identifiers like "toaster", but let's be honest, tonnes will use "acme_toaster"). ARM's mbed does enforce standard identifiers - using the LWM2M (`/:ObjectID/:ObjectInstance/:ResourceID`) format [0] - but that's four different systems and only the big players, so... https://xkcd.com/927/ (Standards).

[0]: https://docs.mbed.com/docs/mbed-device-connector-web-interfa...

The remote control for my television was an improvement because I didn't have to get up from my chair to change channels and when I got cable, there were a lot more channels to hunt. But most things don't need to be tuned because they don't create high bandwidth continuous variable information streams.

My smartphone and television are valuable because they're part of the network. My refrigerator is valuable because it meaningfully reduces my odds of death by food poisoning. The general appeal of configuring things like refrigerators and tea kettles is probably lower than programming the VCR ever was.

I'm skeptical that open protocols will change things much. Mostly because tea kettles are stand alone devices; the benefits of networking them are corner cases; and making them smart adds a lot of incidental complexity and that complexity manifests itself quickly because networks are unreliable.

My preferred 'program' for my refrigerator is "get cold and stay that way". Turns out I can write that with a ten cent analog knob the day I buy it!

I can imagine a world where it's worth programming my refrigerator, but the only possible justification would be that it integrates easily with a fitness band, to-do list, and preferably something else I care about. Ideally, I'd give them all some basic rules from a safe and convenient UI on my computer.

I've seen arguments for networking a kettle - usually "hot when you get home" - but I've never believed any of them. The thing takes 30 seconds to boil already, and I have no need to integrate it into my health or planning regimens.

When I imagine the world were I integrate my refrigerator with my shopping list it winds up not containing me. I mean I'd rather integrate my shopping list directly with my grocery store because integrating with my fridge smells like two integrations: there's also the fridge to the grocery store. Adding a fitness band is an n+1 operation in an O(n^2) integration process. And if I consider integrating at the feature level rather than the device level (which is where integration matters) the combinatorials are at the point where the responses can't be canned and the training sets don't exist to tune the integrations to my goals automatically.

> this technology will remain a niche for the minority that know how to write the software themselves - and have the patience to

Or just want that "computer, make me tea!" Star Trek experience. The frustration of getting it to work can easily be discounted by this guy's will just to make this thing work because the end result is awesome. I'd have gone through the same headache just for the delight of being able to say "Computer, make tea." and the computer making it so. Also, being able to do the same when guests come over just to show it off would have been worth every second of the aggravation it took to get it working - "Look what I did!" :D

Reply on news.ycombinator.com