Why IoT Will Fail (and How To Save It)

Buzzword technology has two possible fates: they fail and disappear or they succeed and disappear. Remember at one time “multimedia” and “networking” were buzzwords. They succeeded and now they’ve vanished into ubiquity. Of course, there are plenty of failed buzzwords (like telecosm) that you probably don’t even remember. They just vanished into obscurity.

Unless you’ve been living under the CNC mill in your local hackerspace, you’ve probably heard or read about the “Internet of Things” (IoT). Companies big and small have realized that getting in early on The Next Big Thing is good for share prices and, right now, IoT is where everyone is trying to make a play.

There’s two things I’d observe, though: First, IoT is far from new. Connecting embedded systems to the Internet is old hat (I even wrote a book called Embedded Internet Design way back in 2003). Second, the way it is going, IoT–in its current incarnation–is doomed.

To see why I would say that, let’s do a thought experiment. In the early days of the Internet, there was no big money involved in it. It was just a quirky government and academic project. But pretend the Internet was being invented in today’s business climate.

Now imagine the ads in your local newspaper: Subscribe to KoolDNS! More free addresses than any other provider. Surveys show that 85% of all sites are available when you use KoolDNS. Another ad would want you to buy a browser that allows you to read over half of the most popular Web sites.

What? Sure, your typical Hackaday reader can manage IP addresses directly and could probably cobble together a reader for some non-HTML format. But the broader public relies on DNS services and a standard HTML format. It is hard enough that a very small percentage of sites today rely on some specific browser feature.

Yet this is where we are heading with IoT. Everyone has an open or semi-open broker. Groups of companies are banding together to create standards that compete with the other guy’s standards. In other areas (for example, WiFi) vendors have submitted to IEEE standards. The IEEE is working on an IoT standard and it might be out by 2017. Meanwhile, it is a land grab to get something to market now, hoping it will take off and displace everyone else.

If you don’t think this kind of thing affects you, you might want to go watch a movie on your HD DVD or Betamax player. As a consumer, being on the wrong side of a standards war can get pretty expensive.

How to save it? I’m avoiding IoT infrastructure that isn’t open source. There no promise the eventual winning standards will be kind to open source, but at least you (or someone) can likely bring it up to snuff. Plus, if you decide to stick with it, at least you can support it yourself if that makes sense in your application.

The truth is, I don’t think IoT will win or lose. I think it has been around for years and will continue to thrive in places where it makes sense. For places it doesn’t make sense, the fad will eventually die down, the big companies will go milk the Next Big Thing (smell-o-vision, perhaps) and the rest of us will continue developing embedded systems that meet our needs.

In the video below, Jeremy Rifkin says the Internet of Things is going to launch a new economic revolution on par with the industrial revolution and that there will be 100 trillion Internet-connected sensors by 2030. However, I can’t help but think that his argument of zero marginal cost is going to apply more to the Star Trek-style replicator than to IoT.

[Artwork courtesy of dadallone]

76 thoughts on “Why IoT Will Fail (and How To Save It)

  1. Good article, i pmuch agree, we will need an actual standard asap before we just end up with a enviornment where everybody ‘knows best’ and nobody wants to listen too the other players in the market.

    One small mistake though:
    “to create standards that complete with the other guy’s standards”
    should prolly be
    “to create standards that compete with the other guy’s standards”

    1. Yes, absolutely. IoT is currently nothing more but a hype about putting a WiFi interface onto a device. Connecting a cheap ESP board to something is not difficult, but how exactly does one set the temperature of a fridge? With one byte? How are temperatures assigned to the values? Or do we need two bytes, or more?

      There are plenty of standards for automation, and they all include such details, or else devices adhering to that standard could not communicate with one another, be it the ancient X10 or the more contemporary ZWave.

      “IoT” is lacking such definitions, everyone is happy simply putting a WiFi board onto something, writing an app for their phone and then being able to use that one app to control that one device.

        1. Yep but the point is, do you just send a byte, or an XML request, or some sort of special Extensible Refrigeration Language, or what? And if it’s just a byte, is it in C or F, or an arbitrary range where 0 is the lowest temperature a fridge can do? Or 255?

          The important thing being every other fridge uses the same standard, so you don’t need software that only works on one brand of fridge.

  2. This is also a problem for Augmented Reality advocates like myself.
    There is seemingly no real hope for a open standard to represent arbitrary geolocated data.
    Its like if every website required its own browser – that’s the state of AR now.

    Sure, theres stuff like Layar and Wikitude which let you submit stuff into their systems – but there also closed hubs using their own protocol and browser.
    There briefly was hope for a standard between a few of them, but it died (surprise) when one of them got bought out by a larger company.

    Theres simply no will these days for the use of open protocols or creating standards. Anyone should be able to make a client, anyone should be able to make a browser, and anyone should be able to host data. If you do that the ecosystem will flourish. But companies would rather you hand out in their corner -sigh-.

    Its a shame, as a good AR standard would also help IoT I feel. It wouldn’t solve everything, but it would mean devices connected to the web could have the option of a “virtual interface” hovering next to them. (for when a AR solution is practical at least).

    1. Ultimately though, consumers will choose one non-standard system that operates well over a standard that everybody implements differently.

      Take a look at Youtube’s new video player for example. HTML5 should be a common standard for everyone, but there’s slight differences in implementation that make it glitch in browsers other than Google Chrome. Things like the seek bar not keeping up with the video, or the voice/video jamming randomly.

      Of course it’s a deliberate effort on part of Google to abuse a standard that offers enough wiggle room to be non-standard in reality, just like how Microsoft “embraced” the open document format in a way that nobody but Microsoft can actually make a perfect implementation that would render MS Office documents correctly.

      Meanwhile, the Flash video player is the same for everyone and just works, despite being proprietary. That’s the problem with standards – what we want is one thing that works the same for everyone, but what we get is multitudes of different interpretations of what the standard means.

      It’s the same in engineering. If I want a snugly fitting M6 nut and bolt, if I order the nuts from one company and the bolts from another, it’s a crapshoot whether the tolerances end up being tight or loose. Despite having a standard, I still have to place an order with just one company to make sure the tolerances are as tight as I need them to be – otherwise my device will be wonky when assembled.

      1. Have you got citations? YouTube works as good in other browsers as it does in chrome for me.

        Its in Googles interest to make Youtube run as slickly as possible too – they get money from adverts , not browser usage. (and no, Chrome does not “collect your data” – Googles websites do that regardless of browser).
        Youtube they have to make it run on all platforms. They managed to kill of IE6 which helped them a lot (as IE6 support was litterally costing them development costs). But other then that, Googles not going to deliberately make it run worse on Firefox – hell, half their own engineers use it.

        Also, HTML5 is itself only the markup, we need a common video standard, hence the new push at the moment.
        There is NO common video codec or audio across all browsers.
        And that’s Apples fault, not Googles. Safari doesn’t even support Ogg for audio.

        Overall though, your point is valid, but not absolute; This was said on a html page published on a http server running on a TCP/IP system. I am running Valvdi browser – not “hackaday browser”, and my system is Windows7 not “Hackaday OS”.
        Standards have given consumers choice and sometimes they can win.

  3. 100% agree. IEEE is notorious in creating ugly standards, just look at WiFi. Zigbee is doomed by being vendor specific. I use openremote with great pleasure and if there is any lacking protocol, I can simply add it. Moreover, if there is CLI then it is possible to get it communicating with openremote almost instantly.

    1. one thing zigbee has going for it is installed base and market/mind share.

      true, the modules are HELLA expensive. otoh, they are trivial to integrate and can be used in simple (AT) or API modes.

      what I like is that we are seeing rf modules that use xbee form factors but are not xbees. that’s cool! the socket spacing is a PITA, but its nice to be able to design a board to take a module and then have choices on very different kinds of modules; bluetooth bees, nrf style bees, wifi bees, etc.

      so, maybe the socket ends up being the thing that lasts ;) kind of funny. but for now, I am using lots of xbees since its an easy solution to get POC’s up and running and I can revisit the radio technology some other time. important thing is to get links between boxes and start them talking in a safe and controlled way. the actual rf sublayer matters much less to me.

      1. I must agree it is not very popular. I am also just starting with it but I like that it is just an extension of the existing internet. IPv6 and you can use CoAP https://en.wikipedia.org/wiki/Constrained_Application_Protocol. Also IPSO Alliance are developing a smart object standards for interoperability http://www.ipso-alliance.org/smart-object-guidelines.
        The CC2650 from TI supports it and the current consumption from the datasheet, so you can see the difference comparing to WiFi and others :
        Active-mode RX: 5.9 mA
        Active-mode TX at 0 dBm: 6.1 mA
        Active-mode TX at +5 dBm: 9.1 mA
        Active-mode MCU: 61 µA/MHz
        Active-mode MCU: 48.5 CoreMark/mA
        Active-mode Sensor Controller: 8.2 µA/MHz
        Standby: 1 µA (RTC Running and RAM/CPU Retention)
        Shutdown: 100 nA (Wakeup on External Events)

  4. I also agree standards are needed. But in my point of view the Internet of Things is just plain old internet and the internet speaks HTTP. There is no reason why devices would not be able to speak HTTP.

    We at DPTechnics (www.dptechnics.com) make IoT devices such as the DPT-Board which are open source and they speak HTTP by default. This makes it super easy for anyone or anything to speak to them. They speak the REST/JSON dialect which support everything we need.

    Operability will only come with years and the effort will be done by ‘operators’ that speak to each other and work together. We have the BlueCherry (www.blueCherry.io) IoT platform which securely relays the HTTP data to anywhere in the world without the need for port-forwarding, connection-monitoring, … and are now actively working togeter with other IoT-platform operators to connect our cloud platforms.

    This completes the comparison with internet and basically any communication system. There won’t be one operator and every operator has it specific protocol’s (for. ex. GSM in europe vs CDMA in USA or IPv6 primarely in Asia and still a lot of IPv4 in Europe/US, …). You will choose an operator and the inter-operator communication will make devices work with each other.

      1. Hi Alan,

        We have indeed looked at MQTT and CoAP which are also widely used. The article you linked to looks really interesting and I will certainly read it. At the moment we have planned to make interaction via MQTT and CoAP possible via the cloud.

          1. one of the issues I have with ‘connected devices’ is that I’m not convinced, not even close, that each box should have an ip-stack and ip-addr.

            securing IP is hard. and IP stacks are VERY nontrivial. linux and bsd ip-stacks continue to get patches and updates even decades after they were ‘done’. will a hardware ipstack (‘network on a chip’) be updatable? very unlikely. shit, man, my PHONE can’t be updated!! I have a flagship google nexus (original) that gets no updates (no security, no nothing; 100% abandonware) and so I cannot trust it on a public network anymore. many phones are like that. dudes: if we can’t even have a good update ‘story’ for the most commonly carried computer on earth (‘the phone’) then what hope do we have of keeping the iot devices updated and secured? zero. Z.E.R.O.

            my idea is to run the thinnest radio protocol over the devices and have them concentrate back to a hub or hubs and those would be the ip-devices or proxies to some one central ip box. that would be the one thing you have to keep updated and secured, not the 100’s of little things that probably won’t even have data ports on them or might run signed code that does not allow updates at all.

            for my r/d work, I’m using lots of xbees and rollilng my own data protocol that rides on the transparent multicast mode of xbee. I’m not using data packeting of zigbee, I’m using the lowest radio packet layer of xbee and doing my own addressing, packet format, etc. all meant to send data from one box to another but never with any thought of getting it ‘out’ toward a cloud. if I need encrypted streams, I can associate 2 xbee’s and that’s that.

            xbee’s don’t ‘radio out’ very far, so hackers across the street or next door won’t even be able to get the signal. wifi leaks out MUCH too far (which is another reason I think wifi is entirely wrong for iot devices). xbee (or the nordic semi nrf chips; both are similar in how I’d use them) is a less well known target for hacks and there simply is not much TO hack on it; no local processes, no relaying of info, no user data, no logins, no routing, not much of anything.

            severely limit how much damage the end iot devices can do, keep them very dumb and local and design the network so that you have a single choke point to control traffic in and out. this is the way I’m moving forward in my own IoT r/d work.

          2. @bl
            “xbee’s don’t ‘radio out’ very far, so hackers across the street or next door won’t even be able to get the signal” You are very wrong about this: their range is limited because of antennas and radios on the, but nothing stops a hacker from using a better radio and antenna to get inside your network
            from far away.

          3. bl: using the minimum necessary – smallest processor, most efficient radio, smallist power consumption – is of course the grail… but as you already know, there’s no single popular standard that would be more suitable.Yet…

            Everyone has, knows and understands wifi, however. And just about any punter can implement a wifi IoT solution without knowing the intricacies of the IP stack. For $4 some company has solved all that for you. Ergo – wifi IoT is all the rage.

            It’s still early days… the biggest gap is envisioning where IoT will have the most impact. Immediately thinking of wifi-enabled fridges and toasters… is mostly missing the point.

          4. @bl : I am not a networking expert but here are my thoughts
            1- Many ways for securing IP devices exist from encryption to firewalls and others. I don’t think IP is less secure than the next thing out there.
            2- Of course it will be updatable with over the air updates (OTA). In fact we do not release any device that does not have one form of OTA. It is much easier to maintain OTA for embedded devices that have a software of few KB than maintaining it for Android and other devices. To make things better, most 6LoWPAN development is based on Contiki RTOS, which is open source and supported by many developers it will get the update without any effort from your side probably.
            3- WiFi is different than 6LoWPAN. Concerning radio power xbee and 6LoWPAN are the same. They can even run on the same hardware (CC2650 from TI).

      2. how is the interoperability working out in mqtt? are devices truly talking to each other? is there a management station that works with all mqtt devices?

        somehow, I kind of doubt it works well.

        I look at network mgmt (a field I’m very familiar with) and take SNMP, for example. its been around for decades and its a ‘standard’ but vendors do crazy shit in their MIBs (data stores) and simply publishing the mib is not enough. you can ‘walk the mib’ of some snmp device but there may be fine details that still need you to write custom code to create a high enough level dialog so that the snmp cmds can do something more than fetch a stat or column of stats.

        even with a protocol standard, I would not expect IoT devices to truly multi-vendor interoperate for a long long time. a ‘standard’ is not going to really help you as much as you may think.

        and even if there is a standard, business guys will always want to lock some things away using an ‘extension’ or private protocol, just to ‘differentiate’. this will break the open-ness of any open protocol as it actually gets implemented.

        and so, snmp has been around for a long time but you cannot manage a cisco box using the same mibs as a juniper box, for the most part. you can’t use ‘standard xml’ or json either; and netconf is still a joke and too immature. see, even in the big huge networking world, interop of mgmt info still is a laugh (or a cry) and if we can’t solve that problem on big expensive systems that have been out for decades, what makes people think we’ll have interop for devices that will talk to each other and to 3rd party mgmt apps?

        1. Well, MQTT works just fine talking to about a dozen different kinds of devices for us.

          From PIC32s to RPis to Mac laptops, to iPhone to Android phone, to AWS instances.
          From C to Python, but we don’t use any other languages. I know there are more.

          Your experiences may differ.

    1. http is not small enough for low-end devices to deal with. JSON is made for that.

      I do believe in the REST concept and so if you have a human hitting a website with html, write back to him in html. if you have a machine hitting a site with json, reply back to it using json. that duality is clean and works well.

    1. He posted twice because the first one was caught in moderation which can be confusing when you don’t see your comment in the thread. I deleted the duplicate post.

      I don’t think this is particularly spammy… sure the comment mentions the company and its products but also gives insight on how they’ve handled the problem Al spelled out in his rant.

        1. Sorry Daan, that’s party my fault. I saw you had posted twice but they were different enough I hated to delete either one and I approved both. We see this a lot, where the system blocks a post for someone to look at and then you (understandably) try again.

          Sorry for the confusion–didn’t mean to make you look like a spammer ;-)

  5. IoT will fail. It isn’t that there aren’t any standards, there are too many. Sure it is nice to have the variety, but as a vendor, which one do you pick. As a small vendor you live and die by that choice. As a large vendor you may survive a bad pick, but until there are only a few standards, this will never take off:

    http://www.eetimes.com/author.asp?section_id=36&doc_id=1327130&_mc=NL_EET_EDT_EET_daily_20150713&cid=NL_EET_EDT_EET_daily_20150713&elq=63f4574076bf4e8db7c2e3f448e62a68&elqCampaignId=23890&elqaid=26983&elqat=1&elqTrackId=70b8c37f0d1249ba80c5b18fa1e3d559

    The second paragraph:
    “At least seven different technologies are competing in this space – two versions of cellular, a new variant of Wi-Fi and at least four different 900 MHz protocols. All have their pros and cons, and so far none have the look of the thoroughbred who won the recent Triple Crown.”

    1. Well I think that’s exactly my point. Everyone wants to use a standard as long as its MY standard! And it is more than the vendors. Like I said in my thought experiment: imagine if there were 15 DNS standards for the Internet?

      1. Great article Al. you don’t even get into security (or the lack thereof) because I’m sure each vendor will want to implement their version which will happen to be closed source.

        I think there is somewhat of an issue of power as well. There are a lot of devices that you want to run for a long time on batteries (think door and window sensors). To do this often ends up with an 8bit uC due to costs of product and implementation which may limit the available processing power for full implementations of TCP/IP.

        That said, the space is evolving so rapidly that a product risks being out of date by the time it is finally launched. even for hobbyists, most stuff is designed around 8bit AVR implementations when there are now a dozen different type of ARM architectures to choose from (along with the ever-present PIC). Building a standard will mean choosing a minimum amount of processor needed to implement the spec. I’ll stay out of the RF frequency debate too because different spectrums are regulated differently in each region/country (how many 433MHz projects are borderline illegal based on United States regulations regarding the 433 MHz spectrum?)

        1. Well there’s been a lot written about security and the security or lack thereof doesn’t really factor in to having one standard, 100 standards, or no standards. The power is an interesting part too. BLE seems to be pretty efficient but that still means you need a gateway. I’ve been thinking about doing some BLE articles so stay tuned.

    2. today, if you want to connect your stereo boxes together and have a comprehensive remote control, its still pretty much vendor-specific. they have hardware busses and some have ir or rf for a box to talk to a box but nothing crosses vendor lines. that’s an ‘iot’ (really, m2m) example where its a success (sony sells lots of sony gear that way; pany sells pany and so on) and yet its completely non-interoperable.

      this is how its going to be for a long time. I actually don’t care much about ‘standards’ right now. we are in the ‘play and learn’ phase. its way too early to nail down standards. a standard that is too early and not well thought out is even worse than having none, as you will then have to support that bad standard for a long time. better for lots of little r/d efforts to work out the real issues, share the learning and THEN design systems that do what USERS want and in a safe, controlled way.

      companies are just moving too fast and this is a problem. we want this stuff (to some degree) but we want well-thought out designs and architectures, not half-baked bullshit which is mostly what we have right now. ip addresses in light bulbs. gimme a break! just because its easy and obvious does not make it a good, clean, affordable and safe solution.

    3. > IoT will fail.

      If you mean IoT in the form “everything will be online”, like everyone will have flying cars, then sure, it’s not gonna happen quite like that.

      Otherwise… IoT is already a success, and there’s tons of actually useful networking yet to be done.

  6. one thing not addressed in the ‘iot will fail’ declaration is the war between those of us who think ‘the cloud’ has a big part in iot, while others (like me) hate clouds, hate privacy violations and see very little reason to leak info upwards toward the public net.

    if there is a cloud in the middle, my first thought is ‘evil! they are spying and datamining’. we should try to get THAT concept across and associate “iot + cloud = evil”. start spreading the word that clouds are privacy violations (at least when it comes to iot; and lets avoid cloud COMPUTING for now as its not even relevant to iot and just confuses people).

    each time I change channels or change volume or pause, NO ONE ELSE has a right to know that. but iot remote controls, for example (an area I like to do r/d on) will always want to leak that ‘metrics’ stuff up northward. my own version has no plans to do that kind of evil shit.

    IoT used to be called m2m (machine to machine) and that sounds a bit more accurate. ‘the internet’ really does not have to be in the middle between you and your connected devices. decades ago when X10 was popular, there was no ‘internet’ tie-in; at best you could have a dtmf tone decoder to turn on/off relays but that’s it and it would not call out, you’d have to call in.

    there are truly rare occasions that ‘things’ need to go out of your private network and reach public wans. but that should be the small exception and NOT the rule.

    also, its worth breaking iot into 2 realms: eiot and iot. I spent a short stint at cisco (doing some iot work for them) and their take on eiot (enterprise iot) is quite different from consumer iot. in the enterprise world, their view is that all the sensors (etc) need to be connected and powered and put on their corp network. the idea cisco has is to run PoE to little adapter boxes that convert PoE to simple power and usb. sensors would be usb devices and the box makes those ‘ip routable’. at that point, you now have sensors that used network cabling for power and you didn’t have to run any new cables at all; just install small metal rj45->usb boxes and string in a bunch of sensors on the usb port; all up in the walls, floors and ceilings. zero bleed-thru to the internet and all data stays in the company.

    eiot is quite different from consumer iot. people should realize there may be even more ‘very different’ iot use-cases other than the 2 specified. moving buses and trains is yet another catagory and its not eiot or classic consumer iot. mobile iot? maybe that’s a good name for it ;)

    but iot won’t die. what I hope is that it gets SECURE and that we have GOOD TOOLS MADE that will help us manage and control the constant attempts to rob us of our privacy and leak too much dangerous info upward. we need non-vendor tools (like super firewalls) that can detect things trying to go out and stop them. this is the biggest problem that I see: we can’t even make firewalls for today’s host-based traffic and get it right and make it simple for regular norms to use. I’m not sure how we’ll deal with 100x more devices at home, all trying to SNEAK THEIR WAY THRU to the outside. we know vendors won’t want to play fair and they’ll try every sneaky trick in the book to get your data back out to them.

    1. I agree with this comment to. We at DPTechnics design IoT devices to be completely stand-alone. This means all ‘smart’ and ‘connected’ functions are on the device itself. Take for example a connected door lock. In the ‘classical’ approach one will design a dumb device that only listen to it’s cloud. On his turn it’s the cloud that contains the user credentials, this is something we wan’t to avoid at all cost.

      Firstly this makes the user completely dependent on his/her internet connection. If the credentials/user data are stored in a cloud solution one is dependent on the internet connection for everything. This is going to give problems sooner or later. If not security problems it will be stability or continuity problems.

      In our designs the IoT devices have full functionality without any internet connection, it was a decision we took 3 years ago and I don’t regret it. It is in my opinion the only way we can guarantee our users their data is stored at a safe place, because what’s safer than to store personal data at your own home after a firewall and NAT?

      Secondly it gives us the possibility to guarantee users their devices will keep working perfectly, even if we would ever stop to exist. The only thing we use the cloud for is to make a secure outside connection to the device.

        1. It’s a thing that is only connected to the internet if you wan’t to and doesn’t rely on it’s internet connection to work. The intelligence is not in the cloud but in the device.

    2. Using a central “cloud” to collect the data and act on it is much easier than any distributed method, and I think your concept of “iot + cloud = evil” is some of the most ridiculous tinhattery. No one cares what you’re doing with the device, and thinking everyone is trying to spy on you is a little egotistical.

      1. and meta-data cannot harm us, either? sure, lets let all kinds of info leak out.

        lets see, hvac info leaking out can tell others that you are away for long periods of time (vacation, business travel). your electricity patterns can also give info about occupancy.

        those are the most obvious leaks. I could list many more.

        the cloud is a marketer’s wet dream. maybe you think its not evil but I don’t think you are fully informed about it, to be perfectly honest, or that you simply have not thought this thru.

        seeing a pattern of movie pauses indicates you are home. watch the total trend of all the info collected and you can know where, in the house, the occupants are!

        not scary to you? fine. have all the iot cloud shit in your life you can install. the rest of us who can think ahead and see where this is going: we’ll be safe since we won’t buy into this lunacy.

      2. one more reply to brad: you cannot use the ‘hide in anonymity’ and have us, the techies, believe that shit! come on, man. this is not 50 years ago where actual people are sitting there going thru your mail. this is the age of automated BIG DATA. every tiny little thing gets hoovered up and stored away. bit by bit, it describes your life.

        are you really not understanding this or just fighting with me just to fight?

        look, if you are one of those ‘if you are doing nothing wrong, you have nothing to hide’ kind of guys, tell us now so I won’t waste any more time dialoging with you. some of us look at each and every bit of outgoing data and ask if it NEEDS to leak out and if not, lets stop it. that’s the proper way to approach this, not the ostrich approach you are advocating (sheesh!)

      3. It’s not ‘everyone’ that are spying, it are the big companies and the government. And that’s no longer called ‘tinfoil hat’ stuff because it has been proven beyond any doubt to occur on a truly massive scale.
        And of course you can also have a unintended fight with some idiot or group of idiots that have technical know-how and then they will not shy to mess up your shit.

        And then there is the cases where you are accused of something and the cops/prosecutors use all (unrelated) data they can get to paint you as a sinister character.

    3. > one thing not addressed in the ‘iot will fail’ declaration is the war between those of us who think ‘the cloud’ has a big part in iot, while others (like me) hate clouds, hate privacy violations and see very little reason to leak info upwards toward the public net.

      100% agree. There’s no compelling reason for the vast majority of “things” to be connected to the wider internet.

      I guess the only possible plus to the current ‘cloud’ IoT services is that they are turn-key solutions; you can buy an interface, plug in some compatible sensors and relay drivers, and have your thing on the Internet by dinner.

  7. I am curious if anyone watched the ridiculous video and then looked at the youtube comments. He has just enough knowledge to convince people that he knows what he is talking about, but most of it is complete nonsense. He doesn’t even seem to understand the definition of half the terms he is using.

  8. Could someone explain me, why I would want to have IoT things in my home, pocket or whatever? Why I would want to have for example a bathroom scale that puts my weight on the Internet? That would end up with diet pills and healthy food spam e-mails and ads on my IoT fridge. I hope this IoT thing die quickly or end up limited to important stuff, like environment monitoring, home and industrial automation (which is done already with no Internet) and such. Who would want to have his personal belongings and home appliances gathering info on his life to use it as profiling and targeting system for marketing weasels?

    1. think ‘home automation’ but without the privacy and data-mining.

      home automation is still a cool idea. people like knowing if they left their garage door open or if there is a boiler flood/break in the garage or basement. this is one example of a box talking to another box and adding useful value.

      its the marketers and ‘data scientists’ that has perverted a very useful idea and evilized it.

      1. We already have security systems that can tell you a door is open (and can close your garage door) on your phone from anywhere. It would be trivial to allow those systems to have general inputs for any kind of sensor. And if you are really worried about flooding the basement like I am (NEVER use non-braided washer hoses!), buy something industrial like a WaterCop. Better to have an intervening control device than just know about it and not be able to do anything from miles away.

        We should all know at this point that any more info going online is bad. For example, a friend had their CrapBook account hacked and now I am getting deluged with calls from incomplete or unknown phone numbers. Thanks again, internet!

      2. You are right,, it CAN theoretically be done right, but I doubt the commercial version will go for that, even if they pretend to.
        But of course tinkerers like HaD visitors can make their own gadgets and not have the world know what is sends or does, and that way achieve some security and privacy..

    2. You shouldn’t if you’re smart since you’re inviting hackers to hijack your system and make your house unlivable.

      Also it will ensure you have no privacy. Why do you think Google bought that thermostat company. If you use it, they know when people are home and when they aren’t. Same thing applies to other appliances. They’ll know your weight, when and what you eat, etc.

      Orwell’s Big Brother would be ecstatic over IOT. It’s the ultimate intrusion technology.

  9. I’m curious to see what the post-fail, post-fad IoT generation will look like. We’ll have to live through a dearth of Things that are Internettet for the IoT’s own sake, rather than for anything meaningful. After my coffee maker that requires a constant network connection and my tweeting toilet get thrown out, we’ll see what applications make real use of this incredible technology.

  10. The primary issue with IOT as it stands is that the money is going into “flat” cloud based solutions. For example, a Nest thermostat talks to Nest and you talk to Nest from your smartphone. This means that Nest actually controls your thermostat and not you. Not only that, but cloud based solutions are not scaleable when you are talking about hundreds of billions of IOT objects trying to talk directly to each other via the cloud. There is no reason why the architecture has to be cloud based. The Internet allows your phone to talk directly with your thermostat or maybe a home router/gateway that you control. What is needed is a “Massively Distributed”, layered, segmented, cloud optional architecture that puts control of objects in the hands of their owners and pushes intelligence out to the edges of the internet. In such an architecture, for example, objects in your house would talk to one or more “object concentrators” that co-ordinates those objects. These in turn could talk to a home controller that you also own and control. If you want to talk to objects in your house from your smartphone, you talk directly to your home controller and not via a third party. That is much more scaleable and puts control back in the hands of the owners of the objects. I am currently working on implementing such an architecture. Also for more on the social, political and commercial implications of a cloud centric IOT architecture see openiotfoundation.org.

  11. All this talk made me think of google’s ‘project vault’, where they use a computer-on-a-SD-card to enable security in a rather clever way (clever enough to even think it might be trustworthy while still originating from google). And then I realize this might be the answer to what confused me, why more and more phones and the like come without SD-card slots; namely to kill attempts at security (and rooting by the user too).
    But ironically google’s own tablets and phones avoid SD-card slots too, so they cripple their own project.

    So I’m still not sure in the end as to what it’s all about.

    1. I think they come without SD card slots just because they want you to upgrade as often as possible.
      Having a easily method to upgrade space (for dirt-cheap) obviously stops that.

      Its very,very easy to route any Android device so I don’t think Googles seriously trying to stop that.

      1. Maybe you’re right. But on the other hand, people always want the latest if they can afford it in any way, so it still seems odd to use a SD slot as a method to push it and maybe there’s yet another explanation, especially since even new phones released today don’t come with massive amounts of space compared to old models, it’s likely about the same in term of storage as previous models.

  12. Actually, IoT will fail because of crappy design.

    Take Nest’s Thread standard for example: it is based on 802.15.4-2006, which uses an organization algorithm that is pretty much, “talk whenever you want to, but check that no one else is talking first.” This very quickly leads to chaotic collapse of a network of even modest density. If you are looking at a situation like a high-tech house, where every switch, light bulb, appliance, thermostat, etc are all IoT devices, you could easily have 40 devices in one small area. With that density, you’d already be looking at situations of collapse, with high latency for turning on your lights, and poor service to all devices, and high duty cycles that drain batteries in days instead of years.

    The astonishing thing is: 802.15.4e-2012 exists, and has existed for a while, and solves this problem with a very elegant algorithm of time synchronized channel hopping, where all links are scheduled. Devices can run at <0.1% duty cycle, and still function very well.

    I really should have finished & published that paper that directly compared the two, showing that 15.4-2006 suffers chaotic collapse, whereas 15.4e just goes to constant throughput.
    Mark my words: Nest's Thread is going to flop, unless they come in and retool it.

    Then there's the hardware. WiFi is pretty horrible. It is optimized for high throughput. It is insanely complicated, and it barely works. It's still common to have wifi compatibility problems with new non-Mac laptops. The cost per bit for the radios is actually really low, but the radios are some deep core IP and are always surrounded in all the 802.11 compatibility stuff, and then hidden behind a driver. In standard operating mode, you can't get a wifi connection for less than 200 mA.

    Every year or so, there would be a newly trumpted project out of MIT (because they have excellent PR departments) wherein someone proclaimed that they would be able to bring mesh networking to all the poor Africans living in squalor, but doing point-to-point links on existing WiFi radios.
    It turns out, unless you rewrite the standard and re-engineer a radio from an existing IP core, that can't be done.

    Wifi is also crazy expensive on the embedded scale; ESP8266 is changing the landscape a bit, but the CC3000 modules are $20+ EACH.

    Let's not forget: Zigbee failed. It was a horrific pay-to-read standard. It required powered basestations. Very few industrial partners picked it up. There was no 15.4 support in phones or computers.

    In a lot of ways, bluetooth was a failure, too. Pay-to-see standard, and a very slow roll-out of hardware. The fad of bluetooth earpieces came and went, and the vast majority of them were horrifically bad, and phone support for them was an unending nightmare.
    Bluetooth BLE has already gone through extensions and renaming, and the ecosystem is already fracturing. People are still using phones with BLE support, and with the Android/iOS/WinPhone platform war, it only makes universal interop that must harder.

  13. At my workplace, we have 3 stories in our university IT building. Open concept(sucks), but there’s breakout rooms that fit 2 or 3 for doing real work. Now, there’s the standard Exchange room registration that you can email/request rooms. And you can see the registration status.

    But you can’t see if the room is currently occupied! Now, because it’s a trendy energy saving building, there’s a PIR for each room’s lights. But they aren’t networked.

    So, I built, using a arduino clone nano, a nRF24L01+, and a PIR sensor for each room. I use a clone-nano and a nRF24L01+ for my listener/control console. Super simple, and $5/room. I have the data going into a Processing sketch that draws the map of the building and turns the room green/red depending on occupancy.

    And I showed this off to the director, along with the price. He went absolutely crazy happy. The last engineers who tried something similar got mired in over-engineering and overpriced everything. I think they had a BOM per room per $800, and of course failed.

  14. Just because we are able to connect every possible device to the net, doesn’t mean we should do it. I do not see the benefit. So my toaster is able to be configured with my smartphone and it gives an alarm once it is done… wow.. using a poti and waiting for the noise it makes it such a struggle to get my toast done!.. the 20$+ for this feature are absolutely worth it… Plus, the app I need (if it doesn’t have a build-in webserver). Most smartphones are overloaded by bloatware which leads to confusion if you have too many apps.

    I don’t see too much innovation going on in this whole IoT thing.

    1. I agree, although a toaster is a poor example for what could be a real benefit of IoT, which is replacing annoying/limited/expensive/failure-prone UI electronics in appliances with a standard comms interface.

      The appliance UI is then run on a smartphone, or any other smart device. Appliances become cheaper, longer-lived (it’s always the crappy “smart” electronics that fail first on these things, and which are costliest to replace), and easier to use, not to mention, skin-able, customizable, remote-controlled, programmable, etc.

      Of course the way this will actually work in practice is that only devices with top-end electronics and UI will get the IoT interface, as an added marketing feature, so you’ll get two sets of failure prone annoying electronics with your appliance.

  15. Much later that I finally got around to reading this, but I can’t say that I agree. Sure, it would be nice if everyone played a long and made open standards, but that’s not what drives companies to invest–look at pharmaceuticals. Patents make them evil, sure, but they’re efficient.

    Blu-ray and HD-DVDs were competing closed standards, weren’t they? (I think so)

Leave a Reply to AmbientCancel reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.