What's Wrong with the Internet of Things | Hacker News Reader

What's Wrong with the Internet of Things

tomtunguz.com

30 pointsghosh19 comments

Ugh. The example given here, the electric imp, is everything that's wrong with how people want to commoditize the Internet of Things. Had the "as a service" product been optional, I'd have already purchased a half dozen of them for various projects I'm working on. However, because it can only use the integrated wifi to connect to their servers, there's no way to build devices that will work in situations where internet access is spotty and/or latency management is crucial. Internet of Things is neat, and fascinating, but articles like this may very well smother it in its crib. Dependency on a central server is the one way to ensure that it stays just a geek toy.

IoT researcher here. You can buy devices with LTE or GSM connectivity, or Bluetooth LE if you prefer. Those still work well for battery devices since the new Bluetooth provides sleep cycle syncing. Also for LTE sleep patter synchronization is being added.

I know that part, and battery life is not the issue. My issue is more with the Electric Imp specifically having everything communicate up with the mothership. For background, the application I wanted to use it for is a costuming project with lighting and sound synchronization, where having to send my data up into the ether and then back down into the other costumes and props, in areas with potentially no GSM/LTE signal, is a no-go. There are quite a few bits of TCP/IP that would make my job easier, but dependency on an outside server isn't one of them.

I came to say the same; my application differs but the breakthrough is not SaaS, it's Moore's law even while Dennard scaling is running out.

I specifically mean the density at minimum cost per transistor. ARM cores are now providing what I used to need x86 cores for.

Embedded has been around longer than any other app.

IoT is a really unsettling prospect. I was expecting a text dealing with this, but alas, mostly click-bait. the text is superficial ideological bickering which takes IoT for granted. IoT is simply here, it seems. the text deals only with what kind of IoT you should like.

it's only a question, as one commenter here put it, of our psychological adaptation to it. like cows in a super-high-tech milking farm, we should adapt to being monitored and analyzed by devices 24/7. furthermore, according to the text here, the most important aspect of this might just be the need to centralize all the collected data! crazy.

not to mention, i still haven't seen anything other than gimmicks and empty talk about ubiquitous devices. which devices? what is the actual use here? you can change your thermostat temperature before getting home? good going for Nest and their product, it certainly seems useful, but there is no revolution here, other than perhaps a completely revolutionary loss in privacy.

and BTW it all sounds like some kind of bubble. like the need to find a way to sell more new stuff has recently gone into over-hype.

It sounds like you're failing to grasp the true implications here. Sure, there is loss of privacy, but there is massive efficiency to be gained for both businesses and end consumers. As with everything we will have to make tradeoffs, and as time goes on I hope privacy considerations will continue play a larger role in the debate.

i don't know why you think i failed to grasp that. there are benefits, i agree. i just don't see any such great or important benefits which would justify this almost complete destruction of the notion of privacy. can you give me some examples?

PS there's a lot more to life than efficiency ;)

You're using the negative connotation of efficiency that should really be discarded. It doesn't only mean "more, faster, because higher profits, work like crazy until you die", that notion is extremely narrow for what efficiency actually means.

Efficiency saves lives right now. It is also what will enable us to save our species from the destruction of the planet. In a world as populated as ours, efficiency is by and large what the very concept of sustainability itself consists of.

you are again projecting some notions of yours into me, than correcting your own projections as if i actually said that. if you would be interested in having a conversation, we might try that. till then, buh-bye.

I agree that the IoT should be about improving lives but that's true of any new product/service/thing and is not specific to IoT.

I do think the post misses the much larger point of how such services and products get created and deployed. Right now, developers either have to (1) use someone else's platform and be locked in or (2) build a system completely from scratch.

We've already seen what happens with option 1 on the web, where the company providing the platform can make an arbitrary decision that puts you out of business - or they get acquired and shut down or a whole host of other things. Option 2 is just too difficult right now unless you're doing nothing but 'appccessories' (toy devices that have some networked functionality) - mainly because we're dealing with distributed systems and the problems they pose.

What's needed are open-source tools to make option 2 viable for a new wave of developers. In other words, we need the equivalent of the open-source LAMP stack but for distributed systems. Just think for a moment of the billions of dollars of value that this open toolstack enabled. We need an equivalent stack to solve the modern problems and empower developers to make new products and services.

I see these modern problems as related to (1) how we create and deploy applications, (2) how those applications store and sync their data and (3) how they identify and securely connect to each other. Of course, there are solutions to all of these right now but they're cumbersome and inappropriate for the environment we're deploying to. I'm working on a toolstack that deals with these issues but in a much more scalable and resilient way. We're calling it the MISO stack and I'll be writing more about it soon. In the meantime, you can look at following post to get an idea of what we're doing and why.

http://nymote.org/blog/2013/introducing-nymote/

I think we love to think that. Developers care about the 'how'. Most of the users, sadly, don't.

And –to be perfectly honest– most developers will sell their service/product with the client data should a big fat offer come along.

It's the developers I'm talking about, not the users. I've yet to see any breakout company that's using a proprietary IoT platform (but correct me if I'm wrong). There's an inherent risk associated with trying to build a big company that way.

Yes, the developer could still sell everything to a third party but that's not really what I'm trying to tackle. It's the problem of building and connecting such applications/services in the first place. You shouldn't need a 'platform' to do that.

I agree, you shouldn't. However, I think dev's will flock to the first commercial API that allows them to concentrate on the specific problem they're trying to tackle. They'll worry about the 'walled garden' if it ever get's so far that it would become a problem. It could even go the Twitter-way. Open for all and pinching down on devs when they've grown (too?) succesfull.

AlertMe has done pretty well for themselves, although they do now recognise the need for an open platform.

To be honest, your tech, while interesting, is a step or two above the typical Internet of Things device, and the problems that will creep in there. The chips I'm working on for my projects that I'd like to have Internet access have at most a luxurious 32 kilobytes of RAM available to them, with some of them having 256 bytes; the closest they're going to get to a kernel is a set of utility functions. The problem isn't the sparsity of space -- explicitly defining variable sizes becomes second nature pretty quickly -- what is is a lack of well-defined, tiny protocols for sending data up and down to the devices. By the time that the device has even gotten to the uPnP stack, it's already been encapsulated and scrambled in some proprietary format because the developers have pretty much no other choice. The part of the stack you're focusing on already has lots of real smart people working on it, we need more people working on the front lines, building good open protocols designed around sharing so that one's sprinkler can talk to one's weather station so that it knows not to run while it's pouring rain, regardless of who makes it. Until we have those, all the TOR in the world isn't going to protect the huge corpus of data that is going to spring forth into the world.

In certain European farms, each cow carries a computer and produces 200 MB of data each year. Contamination and irregularities are found through analyzing the data they are producing.

IoT will certainly improve our quality of life, but I suspect it will be in the background for some time. The processes that bring you the things you use and interact with every day are already getting smarter and more connected, but I doubt most people will have an intelligent anticipatory device strapped to them for quite some time. Humans have to first psychologically adapt to a world where you don't need to stand in check-in lines at an airport terminal.

I think we're still waiting for the "IBM PC Compatible" moment on the IoT. Most of whats available today reminds me of the C-64 era of computing. Everything is limited, silo-ed, and locked. Lots of gardens, no ecosystem.

There's a big difference between complexity being hidden from the user and complexity being locked away in a portal, platform, or single service.

We had this discussion in our office recently - I'm a fan of using the Internet to add more powerful functionality but for a product to really be adapted at scale the Internet connection still needs to add value without requiring the user to interact for more than a few seconds periodically - if at all. I think the IOT category will allow products to better integrate into people's lives but also give power users more control and value.

My concern about the "internet of things" is not discussed here.

My concern is that "internet of things" is really a synonym for "things that used to work great independently but now have external dependencies that make them less reliable and less accommodating, for primarily anti-consumer reasons including surveillance and vendor lock-in"

Reply on news.ycombinator.com