How the cloud is killing IoT
A while back, a bunch of very clever people designed something truly amazing — a decentralised network of computers that were capable of exchanging arbitrary information on a massive scale. We all know that story: the internet’s been around for so long we in fact have forgotten it ever being some other way. However, one thing that we don’t seem to realise is how much of the initial design goal of being “decentralised” has been lost, or the term was broadly misused.
A lot of that centralisation happened as a consequence of business needs — it’s much easier to make sense of vast amounts of data when it’s in one location, not scattered around. This has been the model that the tech giants of our age employ — gather as much information in one place as you can, then make sense of it, and monetise the results. Google, Facebook, Twitter, Amazon et al — all share a roughly similar business model that has been around for the past 15–20 years.
It then comes as no surprise that those same giants, when confronted with the emergence of the Internet of Things, had a bit of a lightbulb moment and thought — Hey, that’s just more data for us!
The not-so-connected things
The term Internet of Things is a bit of a misnomer in itself — it suggests that all we need to do is add some internet connectivity to otherwise mundane things, and they will suddenly become magically useful by virtue of that alone. In fact, it led to the exact opposite — things like the Bluetooth Salt Shaker (no, seriously) or the ill-fated Juicero have contributed to the emergence of another term — The Internet of Shit.
We’ve forgotten that first and foremost, IoT is about improving everyday products by enabling features that make our lives easier in some meaningful way. In the vast majority of cases, that means that things need to be connected to you, and preferably in the most direct way possible.
Where we’ve completely missed the point is how we’re enabling this connectivity.
Let’s talk about smart homes as a case-in-point. We’ve got plenty of solutions out there: from NEST to Hive to Apple HomeKit to Google Home and the plethora of wannabes and also-rans. What almost all of them share is an old-school business model of hoovering up your data, centralising it somewhere and making money out of that information. That’s the only way of operation those companies know. Not only does this lead to dramatic privacy issues (looking at you here, Alexa) but makes you and your home completely beholden to the provider you’ve signed up for and their ability to actually provide a reliable and secure service.
Oh you want to turn on a smart bulb? Sorry, our servers are currently experiencing technical difficulties. But you’re cool in the dark there, right?
Even though they all notionally connect the things to you, they also make sure that the only way for you to be connected is via them as a provider. Is this the future we’ve dreamed of?
Is there a better way?
The internet is supposed to be about inter-connected, but otherwise autonomous things. We need to stop ignoring that design and start embracing it instead. Connectivity is great, but usually all you need is for your things to be connected with other things (and you!) in a small, private personal area network: a home, an office, a factory, a car. Heck, even the human body — a pacemaker sure as hell doesn’t need an IP address and push notification support to be made “smart”.
Our perception is slowly changing for the better — Andreessen Horowitz recently published an amazing presentation about how the world is moving away from the cloud and into the edge — meaning more processing and more analysis is done locally, as opposed to offloading to a bunch of servers someplace. As with many things, this change has only been born out of necessity where the current model didn’t work any more — i.e. in the world of connected cars. You don’t have a fibre-optic connection running from your self-driving car and can’t rely on the cloud being readily available to process the gigabytes of data that the various cameras and LiDARs are generating. And in any case, why would you put gigahertz upon gigahertz of processing power into a car and then not use it?
The Cloud is great and gives us an unparalleled way of sharing things: one can scarcely imagine a world without Twitter, Facebook or Amazon. But it’s not a panacea or an endgame — and we sure shouldn’t make our connected products beholden to it. There are things that the cloud will always be a good place for — analysing huge chunks of data being one of them. But in the IoT world, that should be a secondary consideration — a side channel that’s independent from the main feature.
Here at IMONT Technologies, we’re working on a next-gen SDK for the connected world, that emphasises local connectivity and local control. We would like to see a world of many billions of things talking to each other in the way the internet was designed in the first place — in an ad-hoc, peer-to-peer kind of way.
