The Internet Of Things

Nice intro by Benedict Evans in his most recent post:

My grandfather could probably have told you how many electric motors he owned. There was one in the car, one in the fridge, one in his drill and so on. 

My father, when I was a child, might have struggled to list all the motors he owned (how many, exactly, are in a car?) but could have told you how many devices were in the house that had a chip in. 

Today, I have no idea how many device I own with a chip, but I could tell you how many have a network connection. And I doubt my children will know that, in their turn.  

I woke up thinking about the Internet Of Things today. And I was thinking about something that Benedict talks about in his post – whether the brains of the “things” will be in the “things” or in the cloud or in the smartphone. Benedict thinks we will see all of those models co-existing. I am not so sure. There are tremendous cost benefits to using compute cycles and storage on smartphones or in the cloud. The old timesharing model lives again. And, as Benedict touches on, it will be easier to connect all the intelligence coming from these “things” in our lives if the data and the compute cycles are in the same place.

So my bet is that most “things” will be dumb and the smarts will be in the phone or in the cloud. At least that’s what I woke up thinking about today.


Comments (Archived):

  1. William Mougayar

    Then maybe we should call this The Internet of Connected (or Smart) Things.The old adage “Every Thing has its Place, and every Place had its Thing” applies.In this case, every Thing has its Cloud, and every Cloud has its many Things. It’s a lot easier to re-connect & integrate these Things in the cloud via software rather than stuffing them with excessive local processing.

  2. Mark Birch

    Perhaps, but there is also incredible sophistication at the physical layer that simply do not work well or translate easily to centralized control or smartphone devices. When you examine the enterprise the specialization required for advanced manufacturing and robotics, it lends itself to a model where “intelligence” is distributed, both resident in machine and in central controllers. Ultimately, it will depend on application, and when you look at the wide array of devices becoming “connected” (a picture of which I posted yesterday, the case can be made for both modes.

  3. Barry Nolan

    In a sense, IoT’s set-up resembles many app services such as Hailo – most of the smarts are the in cloud which connects the endpoints (passenger/driver smartphones). To play on Ben’s phrasing, “smartphones are satellites for the cloud”

  4. awaldstein

    Dunno–There are a bunch of closed systems, including even Time Warner I believe to control your home devices through them. Apple will do the same.Is the Internet of Things may evolve to be a bunch of closed systems bumping up on each other.

    1. fredwilson

      i sure hope not

      1. Robert Holtz

        I think you’re right, Fred. In my opinion, most “things” will be either sensors (inputs) or displays/notifiers (outputs). While there may be some kind of local physical device that acts as a networking hub, the real smarts will be in settings, profiles, and scripts that live in the cloud. I dreamt up an idea for this eons ago long before The Internet of Things was an actual thing but it was so early out that I didn’t put more energy into it. This was pre-iPhone, pre-App Store model, before the cloud concept became so embraced and accepted by the public. I still really believe in the space a LOT and I think you’ve got the right idea about what the architecture will be.

      2. LE

        Apple can do the same and that is the play. And people are fine with that.When you think about it the bulk of consumers want “turn key” solutions to problems. There is nothing special at all about dropcam or cctv cameras. I rigged together, back in 1992, a small video camera and a transmitter for the babies crib. (Even had one that was battery operated and didn’t need a power cord.) Years later of course multiple consumer grade products do the same thing (without the black electrical tape). Some today have really nice packaging and are baby friendly as well.So the point is, “Nothing” special about dropcam other than:a) nice designb) product literature written by english speaking peoplec) distribution channel (apple stores)d) Nice packaginge) ease and plug and play of configf) web 2.0 esq web site with a simple message. [1]But a-f is special.The price is high for what you get though. A foscam (which is much harder to install for sure) can get you the same resolution and pan and tilt but it’s a bit scary for average person.Same reason most people don’t use Linux. People will pay a premium to not have to think and have it “just work”. Of course once they are educated they might move to less friendly solutions.[1] Look at foscam website: not as easy to understand, right?

    2. William Mougayar

      If they bump in the cloud, it would be a good thing. Clouds can be friendly when it comes to integration.

    3. SubstrateUndertow

      Or a set of semi-porous systems with distributively judicious flow controls ?

      1. awaldstein

        Cable companies controlling our phones, internet and cable don’t have anything that is porous or flexible;)

        1. SubstrateUndertow

          True but it seems unlikely they will be able to keep everybody down on the farm for very much longer ?

    4. sachmo

      I think there’s some truth to this. The standards will be open (i.e. bluetooth), but the hardware design will be closed for sure. And hardware manufacturers don’t want to be commoditized, so if you can design hardware that’s non standard and hard to replicate – kudos to you.

  5. William Mougayar

    In other news today, Google/Nest is reported to be buying Dropcam.”Dropcam’s main product is a camera that saves its footage to the cloud, letting users check the recordings any time and anywhere. “Bingo.

    1. fredwilson

      i love my dropcam

      1. LE

        I actually bought a foscam model (does pan and tilt and nice resolution) but then discovered that you don’t connect to it by https but by http. So essentially you are wide open. I can fix that myself of course but most users aren’t able to do that. (And I can take care of my own cloud recording part if I want).As far as dropcaam cloud recording, I am totally uncomfortable with that feature. Remember they are getting not only video but audio. And you have no control over what happens to it on their servers (despite what they tell you they do you know all these guys “have it figured out”.)My latest use case for the camera is verifying that the cleaning people were the ones that watered down the liquor.

    2. Robert Holtz

      I also hear Nest is exploring building an alarm/security system next.

      1. William Mougayar

        Building or buying? That is the question 🙂

        1. Robert Holtz

          Exactly right… looks like a little of both.

  6. mike

    Watch the power companies for some clues. The smart power meters already include both wide area mesh networking and local (household) 2-way capability.

    1. Doug Gibbs

      Mike,I have seen their source code. Be afraid.The best way to break into a smart meter. http://concretemulticore.wo

  7. JimHirshfield

    Processing power is becoming increasingly cheaper all the time, so I think the “smarts” will be in both the cloud and the local device in proportion to data resident in cloud to device. And that doesn’t just mean data *already* stored, but the data, for instance, that my phone will sense, accumulate, and need to process locally for a better experience.

  8. pointsnfigures

    Privacy might become a big issue. Will people pay for it? As tech gets more sophisticated and cheaper, it will be easier to build private clouds. Could each device go into your own private cloud that you manage? Or, will storage capacity become so efficient, it will be easier to store everything you want on one device? I could see a lot of potential market segmentation.

    1. William Mougayar

      Private clouds already exist today. If you choose a Dropcam Cloud Recording plan, that’s for your eyes only, right?

  9. gredelj

    Internet of things is just another buzzword following the hype cycle. In a few years’ time, we’ll realize that it does not solve anything by itself, learn some lessons in the process, and move on to the next buzzword. The same thing has been happening over and over.There are much deeper and more fundamental problems behind all this that we must solve first. We need a solution for heterogenous systems (something that would make different things look alike, such as abstract APIs), distribuited transactions (several users/devices generating, changing and accessing data in parallel), replication transparency (so that the data is logically the same no matter how many physical copies are there), locking (to preserve logical consistency of data), location transparency (so that the programs need not be aware where they are running and where the data is), processing transparency (so that algorithms might be the same for any device or for the cloud), network transparency (so that it doesn’t matter what kind of connection and protocols are used), etc. We also need the mathematics for parallelization and optimization of execution, which we could prove to give correct and reasonably efficient results under all circumstances. When we have all that, we can build truly modular, powerful systems which would be able to talk and interact with each other. And we could see a really exponential growth, instead of just adding more monkeys to typewriter room.And then we’ll run into another set of entirely different, non-technical questions: who will protect us/our data, who will be responsible for life-and-death decisions that the algorithms might make, who will have authority and control over those systems, and so on.It’s a long way ahead untill we reach Type-1 civilization (planetary control). The Earth might decide to make us extinct sooner.

    1. ShanaC

      isn’t the math being worked on as we speak?And ethics, always an interesting area of philosophy

    2. SubstrateUndertow

      Maybe there is some tipping point here between over-synchronization/centralization and systemic stability.Distributed redundancy of form and function may play a key generic role in the stability of organically interdependent systems ?Failure to respect some fundamentally generic requirements for inertial dampening within organic systems could be the death of us ?We need a serious collective effort to formalize recurring themes of organic-dynamics as they apply to social-dymanics.

  10. FlavioGomes

    My company develops software that assists in the monetization of the high volume of transactions that Iot (M2M) generates.We are seeing a hybrid of cloud and edge intelligence with data flowing to the cloud.Future considerations that will weigh on edge or central will be privacy and data sensitivity.

    1. William Mougayar

      Is that a bit like Jasper… ?I didn’t realize you were in Cambridge. Let’s connect. I’m going to be in Waterloo on Thursday at the Tech Leadership conference. Are you going?

      1. FlavioGomes

        Yes i will be there. That would be great! I’ll ping u mobile.

  11. Brad Harrison

    A lot of good points made in the interaction between local and network – at the device level, the data level and the processing level. If you look at the power of “big data” – then one might argue that that we should expect a very sophisticated hybrid of interactivity between what transpires locally and how inputs from the network impact the local activity. For example, GE might have a line of connected devices that behave differently based on both local and networked inputs. They might be able to collect information from distributed devices and realize that they can optimize behavior depending on local variables – temperature, humidity, etc – but then depending on your personal configuration – number of appliances, type of heating/cooling, etc – each individual might have their specific device receive a different configuration from the cloud for personal optimization.For sure, it will be a brave new world….hopefully Skynets not in charge

  12. Richard

    In the short long run the “things” will self program “trying” APIs from other “things” to maximum some “pre-written” objective function. In the long long run the “things” will write and maximize their own objective functions and write their own APIs. In the long long long run, “things” will spread objective functions to other devices essentially attempting to overtake and control “weaker” devices.

    1. gredelj

      That’s basically a “war of things”? Natural selection? Birth of a new life form?

      1. Richard

        There will undoubtedly be a war of devices. These “things” (like man) will battle for resources including memory, CPU, energy. And will undoubtedly even form alliances. They will likely be too fast and too smart for man to intervene.

        1. gredelj

          It’s actually already happening. Only it’s not called cyclons, but malware (trojans, spyware, viruses, rootkits, backdoors…). You never know what a piece of software does. You cannot tell “good” software from “bad”. It’s just there and it’s doing something. And with the whole NSA affair, the line between legal and illegal is erased forever.

      2. ShanaC

        are the cylons coming?

    2. SubstrateUndertow

      As a human-biased control-freak I’m not really on board with that risk management approach !

  13. jason wright

    i’ll pass on the idea of carrying my fridge around in my pocket. it will be an islave to me, not i to it.

    1. Robert Holtz

      LOL… get ready to “boot-up” your toaster! 😉

    2. SubstrateUndertow

      My fridge can just cool its jets because it is never getting connected !Unless it can somehow produce free replicator foods.

  14. falicon

    “smarts” will be at the device level…and “wisdom” will be at the cloud level…but it’s all relative anyway (i.e. ‘dumb devices’ will probably *still* be way smarter than most of us regular folk).

  15. Julien

    I don’t think the distinction applies anymore. Eventually, these devices are like peripherals of the same computer. Your harddrive as its own chips, your screen uses its own video card, but for most people (and you probably) this is all the same computer.I believe this internet of things concept is the same: it does not matter where computing happens (and people will likely have no idea where), because all of this is connected.For example, my home scale gives me weight + BMI + Fat %. The calculation is obviously local (or is it?, because it needs my size, age and more, which I entered on a web app!). However, the history and trends is computed on a server (or is it? BEcause when I do not sync the scale for a couple days, it’s still able to show me the whole data)…So, in the end, dumb and smart will be too intertwined to even be distinguishable.

    1. SubstrateUndertow

      Isn’t being in control of the distribution between dumb and smart a key element of security/privacy ?

      1. Julien

        I am afraid not. Eventually, problems like privacy will have to be dealt with at a higher level than the per device level.The OSI network stack can certainly be applied to the internet of things and we are adding new layers above.

  16. Matt A. Myers

    It will come down to cost of energy or ability to produce energy, as to what model will work for which function.A question I have is whether or not we will eliminate redundancy fully or just if that excess data is just required as part of processes; A local copy and one or more copies in the cloud? Maybe a local copy and a local backup, and one or more copies in the cloud?

  17. mkodali

    IoT implies a need to be connected or usage of the cloud in some form, irrespective of what the device is. I can visualize many more apps using cloud for smarts, including Utilities (Nest), most healthcare (various devices) apps, transportation/logistics apps etc. Even a simple application of monitoring a gas tank (for fuel levels) that triggers a simple re-fuel action can use the cloud for timing/pricing/routing intelligence. I tried to look through all the apps on my phone to see if there is any that has intelligence on the device. Except for native phone apps, I didn’t find a single app that doesn’t use cloud for intelligence. While some processing is done on the phone, almost all of them depend on the cloud for full functionality.

    1. SubstrateUndertow

      Does “the cloud” need to be visualized as a single level entity ?Or can it be visualized as a multi-level filtered waterfall affair ?

  18. andyswan

    These new “life forms” will be much like us…. with incredible native processing and sensing powers for fast decisions capable of reacting to situations and learning… but with an exponentially growing base of “wisdom” in the cloud to learn from each other’s mistakes and successes.We are the creator. The student is becoming the master.A little post-holiday Genesis 1:27 fo dat ass.

    1. LE

      with incredible native processing and sensing powers for fast decisions capable of reacting to situations and learning…Sounds like the stuff they do on those tv’s at disney at some futuristic ride to break up the time in the waiting cue. If I had a broadcast quality voice I would totally soundcloud a vo of that.

      1. andyswan

        Sounds like a job for VoiceBunny

    2. ShanaC

      we still don’t know what makes us human

      1. andyswan

        I do, but I’m not telling

        1. SubstrateUndertow

          Some times it is easier to get perspective when you are on the outside looking in :-)Just joking could’t resist.

          1. andyswan

            LOL that’s good

      2. Dave W Baldwin

        The little part of brain recently discovered that makes question “why ” to everything. That leads to discovery

      3. Richard

        Being Charitable

        1. andyswan

          Dogs do that

          1. Richard

            ever see a (hungry) dog give (its food) to a stranger.

  19. LE

    There was one in the car,Fwiw if the car was gasoline there was at the very least the starter motor and the windshield wiper motor if the car was made around the 1920’s. And an alternator is really a motor in reverse basically, right?

  20. LE

    So my bet is that most “things” will be dumb and the smarts will be in the phone or in the cloud. At least that’s what I woke up thinking about today.My bet? If you give companies the ability to have an Eternal September of software “do overs” by allowing updates over the internet, we will go even more in the direction of manufacturers and developers pushing shit out the door and expecting end users to tolerate bugs and defects and then be responsible to clean up the mess.That ability to iterate and update, means less time is spent getting the product right in hopes of keeping ahead of the competition. And foists plenty of frustration on end users.

  21. MickSavant

    I work for a company that manages wireless connections for large enterprises. Much of the planned IoT roll outs have the brains in the cloud. Local devices are typically pinging back simple bits of data; this may be GPS, mileage, fuel levels for a car rental company; flow through rates, telemetry, and “health checks” for a pipeline company etc. Most M2M connections are low bandwidth and very low data transmission (kb not mb)–making the services wildly profitable for global carriers. The space is far too nascent to make any specific and accurate predictions for the future–but I will do it anyways: I think brain in the cloud with the ultimate goal being big data for the millions of simple transactions sent over wireless connections.

    1. MickSavant

      I guess one caveat — the type of connections I described above are “passive” and typically have one way communication. That seems to be the majority of what enterprises are rolling out now, but certainly active two way connections will take a greater share of this rapidly growing market.

  22. Dan Goldin

    This is why I get frustrated when I see car companies trying to integrate various technologies into their cars that would be better served by just allowing a better connection to a smartphone. Why does a car need Twitter or Facebook integration? Just connect via a phone.

    1. LE

      Why does a car need Twitter or Facebook integration?Halo effect that presumably helps them sell more cars. Almost the same reason companies want a sports or entertainment celebrity to help them market their watch or liquor.

      1. Dan Goldin

        It’s possible but I’m not convinced it has the intended effect. I’m also a pretty biased source so who knows. This is one of those things that’s hard to prove either way.

  23. James Ferguson @kWIQly

    Great,The device vs cloud question crops up in building control systems (BCS) as both imperatives and preferences (the priniciples doubtless extend to the IoT generally)A building must be able to operate given loss of cloud connectivity (think Gmail Offline) because eg hospitals need to be able to run during blackouts (In obstetrics scheduling decisions are often proxied to mother nature) and based on local sensors.However, a building informed of weather forecasts produced off-site may be are more efficientThere is portfolio wide number crunching that only needs summary stats of distributed behaviour for example benchmarking comparative performances across a group of peers.Finally there is peer2peer – Local buildings are subject to clusters of weather (think how a cold front or storm can have a clearly delineated boundary) – This gives scope for interesting network modelsToday I have been working on a restaurant Portfolio facing exactly these issues. Where to process ( during development it is centralised for ease , as it become stable distribution makes sense because IoT devices are often idle most of the time so you can shed server loads and increase robustness by distributing. Also distribution of intelligence tends to abstract it clearly making for cleaner modularity and better DSL interfaces.The bottom line is to reach for a great technology stack such that the solutions are flexible

  24. Salt Shaker

    Much ado about nothing. Commodity attributes (or will be in very short order). Google has “Nest, Samsung “Smart Home,” while Apple will announce it’s entree to the sector next week. Appliance manufacturers will make their products cloud compatible for both Android and iOS devices (the “smarts”). There ultimately will be too much risk (and confusion) w/ exclusivity. Next gen appliances, or next year’s models, will all be cloud compatible.

    1. SubstrateUndertow

      I’m not so sure their will be much traction for the likes of connected fridges. But when it comes to health and smart-home sensors/actuators vendor trust might just be the key differentiator and that could favour Apple ?

      1. Salt Shaker

        Yes, fridges perhaps not, but dishwashers, coffee makers, lighting, climate control, security systems, audio equip, dvr’s etc. are nice smart home add-ons. Vendors or appliance OEM’s may have heightened trust in Apple, but presume they’ll be equally trusting of Google too, no? Apple may launch w/ an exclusive with a manufacturer or two, but ultimately, prob sooner than later, smart home is a commodity and one of many product attributes tied to one’s phone or device.

  25. Twain Twain

    The processing smarts will be in the chips that mimic the biology of the brain:*…The data smarts will be in the Cloud that has a polyglot persistence model that mimics how we allocate memories and information and connect them. Our brains are complex and flexible so the Cloud architecture has to reflect this:*…There’ll also be a paradigm shift in our definitions of “smart” and “intelligent” away from the “Big Data” means smart data. The reason for this will be two-fold:(1.) Daniel Kahneman’s work which proves we are not as logical, rational and probabilistic — as the old models of our minds postulates — will make more developers question if probability correlating the “Big Data” sets really optimises for insights and actionable points.(2.) There will be more female engineers re-framing AI and eCommerce.

    1. Twain Twain

      Part (2.) I’m involved with via a “Mission to Mars” code challenge I’m leading.The problem for AI and how smart systems can be is found in IBM Watson creator’s quote: “Watson is akin to a human autistic savant.”Google are focused on building the Star Trek computer which is also autistic.Emotions are a domain expertise of the female brain. Academic research in Neuroscience has discovered that the hippocampus which controls our language, memory and emotion functions is 11% larger in the female brain and is oestrogen-sensitive.This has implications for AI because one of its key deficiencies is understanding the meaning in our words and configuring how to optimise memory in chips.This is why the next curve for tech innovation will involve female engineers and female technical founders who can frame the problem set for smart systems differently from male founders.

      1. Twain Twain

        “Mission to Mars” rather than moonshots and because John Gray wrote ‘Men are from Mars, Women are from Venus’.Our female supernovas are going to encode the currently autistic machines with emotional intelligence.Why does this matter?eCommerce, communications, discovery, networks are all underpinned and driven by our emotional intelligence.

    2. SubstrateUndertow

      Our collective smarts may well be in the Cloud but there are many autonomous layers of smart agents who’s volitional self-interests will favour maintaining a somewhat separately filtered/secured personal cloud.Each member within every species has its own local/private neurological cloud processing as its key survival strategy.Human communication technologies may allow us to create forms go collective smarts but only as an epiphenomenon riding atop a more fundamental community of local/private neurological clouds.This whole topic seems pretty cloudy to me !

  26. vruz

    If we think of these devices as agents that we need to use in order to interact with the world, providing an augmented experience everywhere, and we create the expectation that we must be served everywhere we go, the cloud just won’t cut it.Think a butler, more than a device. Once the expectation of being served is created, there is no going back.It’s possible we will need an in-car cloud, and the “smartphone” will be always a satellite to something else (think planetary systems, a star, a planet, a moon) but I don’t see the cloud staying the same or being the unique heliocentric thing in the long term.Mind you, transportation may end up being a relatively unimportant function of cars and other vehicles.Although considerable gravity must be accounted for the likes of Google and Facebook, whose entire business models depend on great concentration of data, I think the future will look a lot more complex than a handful mammoths running the table.

    1. SubstrateUndertow

      “Things” tend to be used locally and interact locally with other local things under their owns control.So maybe the “Rolling Stones” were ahead of their time with the song “Get off of my Cloud”.

      1. vruz

        Even if we don’t quite get over private property in a hundred years, we still live in a 3.5-dimensional world, and we still need a lot of caching (namely mass storage) and bringing servers closer to reduce latency. There’s no escaping physics.

        1. SubstrateUndertow

          Many local function may not need distal caching just transparent local redundancy ?Other functions may be better served by pushing selective data up the stack into more globally shared cloud instantiations.

          1. vruz

            Local redundancy is the same as no redundancy at all.

  27. Doug Gibbs

    Are the things dumb or smart does not matter. It is an implementation detail. If the device does something useful, it will be purchased. If that happens using trained squirrels inside the box, nobody really cares.More important, will this be then Internet of Things or the Net of Tracking? If my fridge knows I am out of ice cream, so it “suggests” a new brand on sale, that is great! Right? Not so sure. How much personal data goes out of your life to sell you more stuff. Remember, Meta data is data.What does the IoT allow that is new and different? Yes, it can make life easier, it can help with domestic life, but does it do anything new? Not much that I have seen so far.That means the deployment of new devices depends on old devices being replaced. This is not the cell phone or PC marketplace. It will take time, a long, long, money sucking time to see real market penetration of devices in the home.Industrial uptake is happening. That seems like the place to invest first.

  28. Jamyn

    ‘Internet of things’ is an exciting white space opportunity, but I believe the term will become as redundant and over-used as ‘digital’ or ‘mobile’ or ‘big data’. WRT to where the nexus of power will lie, I believe the smart-phone will be the gateway to interaction (whether it’s an Android, Apple or other models), but that the real magic will be at the data and services layer. And to that end, there will be no one ‘owner’ of those – not big players vying for ownership, like Cisco and IBM, nor emerging players, like Smart Things and IFTTT. The one player that seems to have end-to-end capability in the space is Google – from phone, to related home/car/wearable devices, to identity, to data, to computation, to storage, to services. But even they won’t own the space wholly, as the binary set-up spun in the press of ‘winner takes all’ is not reflective of reality and peoples’ inherent desire for choice. Regardless, the promise of an IoT future is one that has me humming with excitement. This is the stuff I dreamt about growing up.

  29. Cam MacRae

    I’m with Benedict.I’ll give you a contemporary example. Before I go for a run I put on my watch and turn it on. It connects to a web service over wifi gets an updated list of satellite locations and a copy of today’s workout. I strap my optical HRM around my upper arm and switch it on. It connects to my watch using the ANT+ protocol. Both my watch and the HRM connect to my phone over bluetooth.As I head out for my run the GPS and accelerometer in the watch start recording my run; pace; distance; cadence, etc. Of course it’s not recording the raw data, but smoothing, correcting, and removing noise in real time. A live feed of this information is sent via the smartphone to my workout diary using my phone’s wireless.The HRM detects my heart rate similarly smoothing, correcting, and removing noise from the optical sensors. When the watch receives my heart rate it calculates whether I’m the correct HR zone at a given moment of the workout and alerts me if I’m not. When the phone receives my heart rate it calculates the calorie burn rate for my food diary.I arrive home after finishing my run. The watch connects to a web service over wifi and dumps the entire workout into a website that allows my coach measure my progress.Where are the brains? Are they in the online data aggregation tools for my coach? The live tracking? Or in the watch? The phone? Or perhaps in the HRM firmware that does some very impressive calculations using data from the optical sensors?The answer, of course, is in all of them.He’s a smart guy, Benedict.

    1. Vasudev Ram

      >Where are the brains? Yes, in all of them. And in the humans who make them.

    2. LE

      Interesting. But why do you do all that instead of just going out for a run? (Note: I’ve been running for a long long time practically every day and I don’t even measure heart rate other than by taking my pulse, sometimes, after a run.)I can tell you from experience that the most important thing with exercise (imo) is keeping it up over a long period of time (assuming you are doing it for health reasons). As you get older you will find that if you are constantly stressing your body you will injure yourself which means you can’t easily do any exercise.

      1. Cam MacRae

        I’m not immune to the pleasure of running unwired.The wire is accountability. Left to my own devices I tend to run my easy runs too hard and my hard runs too easy. As you note, the former is a problem.

  30. Puneet Mehta

    Fred, I partially agree that most “things” will be dumb “sense and forward” with a possible “act on command” programmable interface. The only intelligence potentially needed is a ubiquitous security layer – intelligence in these dumb “things” about who could they talk to and accept commands from.

  31. Vasudev Ram

    Speaking of the Internet of Things, here a couple of posts I had written some time ago on the topic. One is about an acquisition and the other is more technical:PTC Acquires ThingWorx, Internet of Things Platform Provider:…websocketd and Python for system monitoring – the JavaScript WebSocket client:

  32. Matt Miesnieks

    It will depend on the value of the user interface to the experience. The richer & more natural (& thus more real time) the UI, the more compute on the device (sensor fusion more so than traditional CPU). Less time sensitive data processing will always be in the cloud. …. this will be true until network capacity grows & latency drops below ~30ms for a real time response

    1. Vasudev Ram

      Yes. This is roughly the same argument as thin client vs. thick client, or whether to do more processing on the client or on the server. As usual, there are pros and cons to both sides, and also, it depends on the application, as you say.

      1. Matt Miesnieks

        no this isn’t really thick v thin client. The UX for sensor driven applications is *different* in the same way mobile is *different* to a PC UX. It’s a new medium & will demand new paradigms of app design. As we’ve seen on mobile, if the UX can save a click or two, conversion rates (& the business case) shoot up. This has driven the apps vs mobile web trend. This trend will be magnified on more atomic devices, and the implications of handling the UX natively are both different & critical.

        1. Vasudev Ram

          OK, I guess you know more about this area than me. Thanks for the reply. Have you seen and can you mention any such new paradigms of app design for sensor driven applications?

          1. Matt Miesnieks

            think about any wearable products like jawbone or fitbit, or even Google Glass’s concept of Cards with only a few lines of text. The UX is highly constrained & you can’t just shrink down a smartphone app, you actually need to design different types of highly contextual (so the content is brief) and natural (so the input/output is usable) interactions. We saw exactly the same trends going from CD-ROM to Web (you can’t just put Encarta online & get Wikipedia) and from Web to Mobile (shrinking Flickr website doesn’t get you Instagram). The Apps & Interactions for this next gen will be Sensor Native and different to what we know on mobile.

          2. Vasudev Ram

            Got it now, thanks. Interesting stuff …

  33. NICCAI

    The dumb terminal was something we recognized early in computing, but the costs of cloud infrastructure were to high and cumbersome. The internet of things is exactly that – a network. The days of a standalone device with no piece in the cloud are over. The question that remains is which patterns of connectivity will prevail.

  34. rfreeborn

    Many of these comments are taking a very consumer app driven perspective to this – which I don’t think the #IoT will be.The real power and effect comes when it’s done on the industrial level and the decisions about where storage, compute and other functions are being done will be driven by application requirements. Consumption of the data on the other hand, well, that’s a whole different matter. I personally think that the greatest power and benefits of IoT solutions will actually come into play when people *aren’t* involved in the decision and it’s instead greatly automated.Cisco has an interesting take on it (see this that is. admittedly, very network compute specific and they focus on compute at the edge or “fog” (vs cloud) computing.My two cents…… rob

  35. Albert Hartman

    Many members, complex coordinated behavior. In an ant swarm, where does the brain reside?

  36. Karl Klept

    2 views: Google Chromecast has minimal direct interaction with the device being controlled, most relevant stuff for both controlled and controlling device is via cloud, Apple AirPlay projects data from controller to controlled device with minimal cloud interaction.

  37. stelajohnson1984

    With these share repurchases programs, the financial metrics of company will definitely improve. Earnings per share, P/E multiple, ROE, ROA and the share price which has been stagnant during recent times, all will be

  38. Pete Griffiths

    I don’t think there will ever be a definitive answer to this architectural question.As hardware and system software continue to evolve so shall the balance between central (cloud) and local.

  39. sigmaalgebra

    Internet of things (IoT)? Sure it’s possible. Certainly can do that. E.g., as I recall, some years ago IBM had a single chip with the TCP/IP stack in hardware; so, could have a single chip, today likely much smaller, with the TCP/IP stack, a processor, some micro code, some software, and some I/O, all small, fast, and cheap, and sell them like kernels of corn. Sure, could do that.But why would one want to? Is this a solution looking for a problem? Saying it is has long been a severe insult to the technology. But, of course, tough to believe that there won’t or can’t be a suitable problem, so far unsolved, and where the IoT is a big help for a solution.Sooooooo, can we look for such problems?While thinking of a problem, note that need to be sending some data via TCP/IP. So, for the data sent, have some proprietary protocols with encryption so that only one company’s ‘cloud’ can interact with the “things”. Right, not ‘open source’. Hmm …. But maybe one heck of a ‘network effect’. Could the recent situation that messages from an Apple product can’t go to an Android product be an attempt at such a ‘network effect’? Or, I mean, might anyone at Apple have thought of such a thing? Of course not because, I mean, of course all ‘good’ Apple users would never think of using anything else in which case there would not be a problem and no need for such a ‘private protocol’ issue; so, there would be no need for Apple to do such a thing to ‘good’ Apple users! Of course not!As I look around my house and life, I am slow to see what ‘things’ I might want to have Internet access. Okay, sure: Maybe the electric power companies want to have billing based on peak usage. So, have each ‘thing’ that uses a significant quantity of electric power be on the IoT back to a server that tells those devices when to turn on/off to minimize the peak usage subject to keeping the house going satisfactorily. Sure, that’s a problem in stochastic optimal control although maybe some simple heuristics might do well enough, especially if a customer doesn’t have good evidence, which tends to be the case! So the ‘control’ would be careful about power usage for the A/C unit, the water pump, any refrigerators, and the oven while keeping the oven temperature and refrigerator temperatures in a satisfactory range, the water pressure high enough, and the house temperature okay, Maybe, Sounds far fetched to me. For having my refrigerator telling a ‘control’ at AWS that I’m out of milk, beer, or Champagne is a bit much!Maybe some new services, not common now, could make good use of IoT.

  40. Hutchy

    Oh hey they re-badged SCADA sounds sexy now.

  41. Kevin Hill

    Possible, but I think Hadoop has been a good example for the fact that even in the ideal circumstances, networking can be a bottleneck to computation. If there is really that much thinking to do in the Internet of things, decentralized compute will be a necessity.

  42. Internet-of-people

    If I was you I could have written “… will be dumb and the smarts will be in the person -not in the phone or in the cloud…”

  43. eran shir

    I don’t think the edge can stay dumb for long. The prevailing model will be more interesting than that. Wrote why here:…In short, bandwidth, realtime, and resilience requirements will push intelligence down the pipe.

  44. Bernhard

    The key is to bring information that “dumb” device produce into the cloud to make these devices or infrastructures controllable. That’s the reason why we will see many “smart network adapters”, highly specialized in connecting one specific sort of device to the cloud. Closed home automation systems will have to open up since city blocks, industries and infrastructures will be upgraded to connect to networks, so smart home systems have to be open to be able to be fully leveraged as a sub-part in these system infrastructures.

  45. sachmo

    There’s software and then there’s firmware. The firmware is going to get it’s updates too. The user will interact with the device from their smartphone or the cloud, but the intelligence may exist on the device.How is this difference than intelligence existing solely on the cloud? B/c when the maker of the device (say an internet connected sprinkler system) releases new electronic hardware capable of NFC connectivity with your lawn mower, the latest application version ain’t gonna bring this new feature to you unless you get your hardware upgraded.The overall dichotomy paints it as a zero-sum game between hardware / software, which just isn’t true. It’s more like tighter integration between hardware and software.Anyway, I’m a hardware designer (that’s had to do a lot of software testing!), and I vote on smart hardware.

  46. falicon

    I have been intentionally leaving my phone at home from time to time lately…I wish more people would do that actually 😉

  47. William Mougayar

    Brilliant comment Charlie. It is Tuesday after all :)Clouds ain’t what they used to be.

  48. ShanaC

    Can you see bi-directional models happening.And how were the fish?

  49. leigh

    SETI that’s something I haven’t thought about in a long long time!

  50. Jim Canto

    @ccrystle:disqus … Charlie, I’m having a heck of a time recovering from a vaca I took recently. No useful cell or wifi connection for multiple days…a truly “fascinating” perspective. I can relate.

  51. SubstrateUndertow

    From the frontline “internet of things” all the way up through various personal-hub-controler devices and continuing on up to the mega-cloud-integrators like Google, any combination of sensors, actuators, data-objects, processing-objects and most importantly communication-filters/synchronizers are possible at each and every layer in that “non trivial causal spread” of organically-distributive synchronicity-based collective intelligence.Just like in biological eco-systems a fine-tuned emergent-best-fit symbiotic-coherence between those layered-functions will determine the fate and stability of the system as a whole.One has to wonder whether implementing this level of organic interdependence at the human social level inherits the same degree of trial and error resilience afforded its lower level biological precursor?This time around we are consciously in charge of building out our own self-refferenial organics-based social-substrate.What are the core recurring/reusable themes inherent in organic self-organizing systems. Do they apply to us? Do we need to colour inside those line or can/should we just make it up as we go along?I’m with James G. Miller(Living Systems Theory-1978) on this one!We need a serious collective effort to develop and popularize a clear, simplified, mass-culture set of organicnarrativesmetaphorslexiconsthat effectively reframe our presently emergent social reality as an instantiation of universal organic reusables.Postmodernist be damned, overarching road maps are important tools despite their frequent obsolescence and need for refresh !

  52. gredelj

    Because software is complex. Human mind is limited in the number of things you can think and reason about. There are just not enough people who can hold everything together, even if every person on the planet becomes a programmer. We need a way to describe complex systems using simple rules, and let computers do the heavy lifting of filling the gaps.

  53. LE

    I wish more people would do that actuallyYour competition isn’t leaving it’s phone home. And you know saying about competition.

  54. Richard

    Turn off your TV for a month would be a better idea

  55. falicon

    I just mean for social settings…I see people getting worse and worse about ignoring the world/people around them and just focused on their devices…but it’s today’s reality and I only see the trend heading towards *more* not *less*

  56. LE

    A dumb phone (like a law degree used to be) is always a good thing to have. I actually carry two phones when traveling so I can talk to certain people if needed by the non primary phone and/or if the primary phone is damaged in some way. Another thing you can do is put a google voice number in front of the primary phone in order to screen and block calls.

  57. LE

    We should be his two gay parents. You can tell him to coach more sports and I can tell him to get back to coding. Between the two of us he will have a great balance in life.

  58. PhilipSugar

    I would argue that having your phone always on and expecting to replace planning and organization with the ability to “always be reached” is both unscalable and unsustainable.It confuses activity for accomplishment.So not being a slave to your phone does not mean that you are going to lose to your competitors.It does if you have no organization and planning. If you are organized and plan, then you can schedule time and your phone can be an incredible productivity tool.I’ll give you an analog example. Ever talk to a big company person who says “I was so busy today, I had twelve hours of meetings, back to back to back” What decisions did you make?? What exactly did you accomplish? Uuuuuh….I was super busy!!!Yeah, talking, texting, and email on your phone 15hrs a day, might make you “feel” you are busy, but to me it shows no planning.

  59. LE

    Wait. You mean that you didn’t think that was clearly concisely stated in a way that the ordinary man could understand??? [1][1] Essentially the reason there are buzzwords and talking points to describe things.

  60. Dave W Baldwin

    No. As the higher smart is more prevalent which will begin with phones and moving then to devices and phones controlled by central station at home, the population of users will no longer be virgins who are controlled by phone.Kinda like observing those who realize they don’t have to answer because it will be there an hour from now.

  61. vruz

    That’s correct, and it only helps to accelerate the process in the direction I described.But this is history repeating itself… remember the mainframe, the Unix workstations and the PC?

  62. Vasudev Ram

    Ah, but after a while we’ll probably need a device to do that for us.

  63. Jim Canto

    Tough to say goodbye, eh? I’m so far from ANY beach… I try not to think about them too much anymore.

  64. dfooter

    Exactly. In order for the network of devices to work seamlessly for the different needs of the consumer in the moment, context will be crucial in determining where data is processed and for what purposes.