Do machines dream of IoT?

12/6/2011 5:13 AM EST

The heating thermostat is a great example.

I have lived with them all my life. About a year ago i replaced my central heating boiler and opted for a wireless thermostat unit so that wiring would not have to be strung through the house back to the boiler.

Bottom line: the overall system is more sophisticated and perfroms less well than most I have known. A long time constant means that the house cycles between slightly too warm and slightly too cold.

Steve Jobs clearly understood that customers don't want to be told "you can get it to work if you tweak it." Customers want a delightful experience out of the box. I am ready to be delighted and, occasionally, I am.

12/6/2011 3:39 PM EST

Yes, the too wide temperature cycling range is a nuisance. But it is not the fault of IoT per se.

That is a good example of what I'm talking about. Thermostats connected to HVAC systems have existed for a long time, and no one ever throught to give a a cathy name like M to M comms. Now that fancy thermostats might make use of IP, perhaps even to allow remote setting by a user, people make it sound like there's been a fundamental shift. But there hasn't. It's a simple evolutionary change, that may or may not fit people's needs.

12/6/2011 4:47 PM EST

But the process tends to be thus:

Vendors make systems more complex/sophisticated because they can. To try and extract extra value the vendors describe extra features and benefits the system WILL/SHOULD have.

Punters bite and the system works less well than expected for all sorts reasons; usually the unintended consequences of the added complexity.

Punter tries to return to old simpler version of system to find it is no longer for sale.

12/6/2011 10:00 AM EST

Oops... I said "Dave" but meant "HAL".

12/6/2011 3:51 PM EST

On the smart car technology, I've been saying for some time now that the time will come when we'll look back and marvel that people were ever "allowed" to drive manually. How was this even possible?

Consider this. There is a practical limit to how many roadways can be built, especially in sprawling urban areas. In the LA area, for example, they are even talking about double deck freeways. So one might wonder, why not use the same roads, but much, much more efficiently?

And this can be done. You automate the whole process. Cars can be made to move faster, and be packed closer together, if you take human drivers out of the loop entirely. This sort of upgrade can be installed only on the most congested freeways at first. Doesn't have to be all or nothing. Side streets, rural roads, etc., can go on for a long time with manual driving.

There are tons of examples of previously manual functions that are now always done automatically, and no one gives them a second thought. For cars, think of the choke, or the fuel-air mixture, or the spark advance, or even the gearbox. Automation makes all of these functions far, far more fuel efficient and less polluting. You wouldn't even think of installing an 8-gear manual transmission in cars, yet with an automatic, no problem.

3/12/2012 2:45 PM EDT

Not necessarily. I hate automatic transmissions, and my next cr will one again have an manual transmission. It is a much more enjoyable driving experience.

12/7/2011 9:23 AM EST

sharps_eng, I've seen all sorts of studies on so-called smart homes; there are one or two built by UVA here in Charlottesville. What I haven't seen is a convincing description of just how external scheduling of my washing machine was going to help anyone. Since I do at least 8 loads a week in my home, I presume that I would have to be up all night reloading the damned thing... and I want my coffee when a lot of other people do: first thing in the morning. That's probably not what is intended, saving a watt or two, but it is symptomatic of external scheduling control. Ok, so instead we concentrate on the long poles in the tent: heating and A/C... I'd like to see some real numbers, or simulation, that shows much more than minor reductions in peak demand from the elaborate infrastructure that will arise if the local power company gets to turn my HVAC system on and off. I suspect that all that will happen is that I'll pay twice as much for electricity when I need it as when I don't. Eat dinner at 11:00 PM, get up and shower at 3:00 AM... I just don't see how externally-scheduled events in my home are going to help anyone, let alone me...

12/7/2011 9:44 AM EST

I agree Bob, smart homes currently make little sense...one of the problems is very low cost of electricity...I spend $40 a month on it, even if I achieve 10% reduction by showering at 3am that will be buy only one latte...Kris

12/10/2011 2:24 PM EST

@Bob and @Kris - I was thinking along the lines of finer-grained industrial control, talking to the machine allows cooperative scheduling between on/off thermostat cycles. Many industrial heating installations use a lot of power over the day with a sluggish response and can wait for a slot on a minute-to-minute basis.

Probably improved PFC will provide more economies as more modern machinery are installed.

Domestic electricity is just a sort of chaos which averages out demand statistically except when people's activities get synchronised by TV shows; but internet usage is blurring that which helps.

BTW Our UK domestic power bill is $300 monthly (winter). It just went up 40% in 2yrs.

12/12/2011 10:40 AM EST

Well, I think there is room for fine-grain control... particularly in industrial situations. I don't know that the savings would be earth-shaking, but then, once the control infrastructure is in place, the cost of obtaining the synchronization would be minor. On the home front, you are correct: the power companies rely on statistical averaging. The big events are hard for them to control: you mention TV shows, but I am thinking of a big high pressure cell over an area, when every single air conditioner in the area starts to cycle on and off, whether or not anyone is at home. Many people don't have even basic scheduling thermostats...


May I ask if you power bill includes heating? People in the US are usually one of these: oil, natural gas, or electric. In the northeast, oil now costs $3.50/gallon. One of my relatives uses 800 gallons per heating season. I don't know how they do it.

12/6/2011 7:36 PM EST

Public transportation does not get used enough, for the simple reason that it doesn't go everywhere people want to go, when they want to go. Plus, you need to run buses that are just about empty for much of the day, if you expect the system to be at all usable. Few people would depend on a public transporation system that might leave them high and drive, e.g. if they have to stay late at work. Unless there are no options.

So it turns out not to be as efficient as you might think.

I use public transportation to get to work every single day (well, not quite, I do have to drive to get to it), but it's useless for me on weekends. AND I have no illusions about our system being usable by everyone. Far from it. It's very inefficient between suburban areas, for example, yet many are employed in suburban office complexes these days.

The future, I have no doubt, will be increasingly self-driving private transportation. At first, collision avoidance sensors. But it will go from there.

12/6/2011 7:44 PM EST

"... system that would leave them high and DRY," is what I meant.

12/7/2011 9:11 AM EST

Aw, Bert, I thought it was a clever double entendre... when the public trans system is unavailable, you're left high and (have to) drive... very good... ;-)

12/7/2011 3:52 PM EST

To a degree, I agree. But even in Europe, public transportation isn't always convenient, and if anything, since WWII, more and more people have been using private transportation. Even with the high price of gasoline. Public transportation will always be limited in terms of where it goes and how frequently it runs. And even in cities like Rome, a lot more people now commute from neighboring towns to go to work, than in the old days. So it's not all that different from the US.

But here's another example of automation in automobiles that we could never go back from. Back in the 1930s, I'm sure a lot of people would feel leary about foregoing the manual crank starter, at the very least as a backup mechanism. They probably sounded then just like those who oppose some of the autmatic new features today.

And yet, those hand cranks are not only dangerous, but totally useless when the compression ratio gets above about 5:1 or so. I doubt anyone would advocate lowering compression ratios today, just so we can have those hand cranks back.

Ditto with steam ships. The first ones had masts and sails as well as steam engines, just in case. And so on, and so on.

12/8/2011 11:35 AM EST

Bert's point about the disadvantages of public trans, and also, it's basic inconvenience, are well made. When I had to work in New York City, I appreciated the fact that the subways were there, but I hardly enjoyed using them. (Granted--other cities have far nicer subway systems, but you get the point.) Bert's points on advances in technology, that the changes are made to a partially-resisting public, are also cogent. But as far as the fact public trans will become more necessary as the population shifts into cities goes, what ever happened to work across the Internet? Just why do people flock into cities if they really don't have to? Far more people can do their work at home than do so right now. Much wear and tear on the infrastructure could be saved. But it doesn't happen as much as it could. Think about the fuel savings if businesses really de-centrified. They don't; it's too bad, but suspect I know why...

12/8/2011 11:56 AM EST

Bob: you really don't expect a farmer in India or China to work on Internet do you? The migration to cities is a well known fact but applies to population worldwide, high-tech workers in US is a very small sliver of that...Kris

12/8/2011 12:57 PM EST

Iniewski, no, of course not--at least no more than you expect someone living 24 miles from the city he works in to wait for a bus. ;-) No... many jobs can't be done over the 'net. But many more can than are in fact telecommuted. Here's my take: some people can drive an electric hybrid: I can't. So both heavy fuel users and light fuel users will continue to exist. Some people can work from home; others can't, but a lot of improvement is possible there. Many people can use public trans, for some of us it's hopeless. In each of these domains, there's no "one size fits all." But much improvement is possible, as long as the partisans of public trans don't get too busy banning cars from their cities...

12/7/2011 6:30 PM EST

True enough.

My original point, though, was that EVEN in networked control systems, the trend has been to migrate from legacy protocols to Internet Protocols. And this is not a recent trend. It has been going on for at least the past 15-18 years.

So this Internet of Things (as opposed to just people) is nothing new. You might even argue, the original ARPAnet was between academic and miltary computers, located in computing centers. If individuals accessed the Internet, they did it across non-IP links to their PCs or computer terminals.

I agree completely that the security aspects are real. My only point was, the IoT is nothing new. It is a progression of trends that have been there from the start. The new buzzword seems unnecessary, although I'm sure it will generate lots of scholarly papers, conventions, grants for schools, and all manner of other revenue for what are undoubteldy worthy causes.

12/8/2011 12:59 PM EST

Patrick, they'll shortly be telling us to go to bed, and not to watch TV too late. In a worst case scenario, we will become the mice in the wainscoting. Have a read (if you can find him) of William Tenn.

12/8/2011 5:06 PM EST

Machines talk to each other all the time, everywhere. The HVAC system I hear operating in my office, for example, has all manner of machines talking to each other, and no one gives that a second thought.

Some machines are already watching our every move. They deliberately prevent me from navigating to certain parts of the office intranet, for instance. Or during a fire alarm they prevent me from going through certain doors. Or while driving, they prevent me from locking my brakes during a panic stop.

In some airports, machines run local trains with no human operator. In power plants, machines monitor the plant and automatically go through different procedures, with no human intervention, to adjust to changing conditions. Even automatic shutdown, but that's only the extreme cases the press likes to write about.

Attibuting self awareness or free will to machines is premature, perhaps, but I'm not sure I understand how so many people, especially engineers(!), can have missed that M to M communications has been with us for a really, really long time.

12/13/2011 8:45 AM EST

Richard, your arguments are rational, and in the cases you cite the situation might have turned out better without a computer in the loop. However, consider this: you state a hazardous condition with special cases, and then postulate that in that case, having a human being in control would work better. I agree: if the said human was paying attention, competent, and could be relied upon to take the correct action. This isn't the general case--how many people do you see yabbering on their cell phone or waving their hands while driving? I think that, on the average, a computer in the loop is more valuable than relying on a person, late to work, holding coffee, doing her makeup, etc... suddenly becoming a first class driver. That's my disagreement with you: not that your analysis is incorrect, but that on the average, I think that a computer does better. In the case of the Air France crash, however, I make two points: (1) nose down to avoid a stall is always a good idea, especially near the ground. In that case, your options are to hit the ground under control, or out of control. I opt for nose down, under control. (2) The control system was probably not tested all that well for low-altitude, low speed passes over the runway for air shows. That's not a great excuse, however.

12/13/2011 11:35 PM EST

Bob, I do agree with you that given today's potential for distractions, computer aided assistance is not only beneficial, but mandatory. I do feel that this technology can be implemented in a way that coexists with the driver, rather than overriding him or her. Digital HUD speedometers have been utilized before; why not have a digital HUD target that tracks possible pedestrian hazards? The target color could graduate from yellow to red, with an alarm that sounds as the situation becomes critical. Radar detectors are effective at gaining the driver’s attention using similar progressive audio/visual cues, letting the driver apply the brakes before a speed trap.

12/14/2011 9:02 AM EST

Richard, it's somewhat banal to make this observation, but automation must be balanced against the both the average driver's capability and the poorer driver's capability. Your observation that the driver is often distracted is quite right... if cars were only driven by focussed, trained drivers, the situation would be different. Isn't much the same with the old argument of compiler vs. assembly? How many times did I hear "I can write tighter, better assembly than any compiler". And this may have been true: for 25 lines or so. On the average in the long-haul, though, compiler quality and consistency put most programmers to shame. (Not to mention documentation and maintainability--those are another thread.) Even in aircraft, pilots become distracted, or worse, disoriented, and some automation is there to keep the pilot from doing something that he really should think about first: a stick-pusher is an example. But agreed, agreed: automation is more than merely beneficial, it becomes more and more necessary as drivers are less and less engaged with their car, and more engaged with "whatever".

12/15/2011 11:02 PM EST

It looks like we’re both on the same page, but we are seeing points differently. A machine can certainly perform routine tasks more quickly, accurately, and with much more consistency than a human, but automation is also the system’s limitation. It cannot think; it can only follow its programming. That’s why the human element is just as great as a necessity as automation, and where you’re absolutely right that automation must be balanced. Where we have altering views is where the needle should sit, between the extremes of full human control (riding a bike) to full automation (taking a monorail). I suppose since I live in Taiwan where cars, pedestrians and armadas of scooters battle for right-of-way in sometimes very narrow alleys, I lean towards empowering drivers with more control. The potential for false positives exists at every intersection, where involuntary braking can be dangerous. Even if the computer braking system was highly debugged, just the sheer anticipation of “when is my car going to take over” would be an unintended secondary impairment on the driver, similar to having a jack-in-the-box on the dashboad.

12/16/2011 1:42 PM EST

Richard, this is actually a response to your posting of 12/15 @ 11:02... the system doens't allow unlimited reply nesting. But anyway, your point in that posting is well taken. The degree of automation must also take into account the service domain: are we in Taiwan's traffic, or on the mountain road that leads to my home? And of course, are we driving a commuter vehicle, or a heavy truck. All of this matters to the total control system. My Toyota Highlander has a strange control system. You are facing a downhill road, wet or poor traction. You put the car in REVERSE(!!!) and push a button. The car then controls both the accelerator and the brakes (!!!) to bring the car down the incline safely. (!!!) I tried this in my driveway, about 1/3 of a mile long, changing altitude by 200' (it's wild.) It is the absolutely strangest feeling to merely steer, while a computer operates the drive system. Still, it is the direction the world is going. (Not downhill ;-), but towards automation.)

12/17/2011 10:32 PM EST

Bob, yes, you are entirely correct. Love it, hate it or embrace it, increased automation is the new reality where rejecting it becomes less and less of an option for the majority of the population. Hopefully the level of enhancement and safety brought to our lives will outweigh the restrictions. I remain cautiously optimistic!

12/13/2011 3:51 PM EST

Dr. Brazil, have you been channeling George Orwell? ;-). Just kidding: your comment points out the enormous possibilities for intrusive government abuse when monitoring technology can insert its sensors into the home. Even more likely than being reported for storing meat in your 'fridge is the likelihood that your purchasing decisions will be known immediately to the marketeers, the Sierra Club, or any of a number of groups who know what is best for you. You can see a parallel attempt to corral all of us into monitored dormitories in the incessant drive to eliminate private possession of firearms among the population. There's a limit to what the government can do to an armed population, and that limit galls. There's a limit to what the government can do with little knowledge of what goes on in your home, but perhaps the accelerometers in many electronic devices can be persuaded that they are really microphones...


The government of the United States would never stoop to unauthorized surveillance of its citizens, would it?

12/14/2011 5:11 PM EST

//“The government ... would never stoop to unauthorized surveillance ... would it?”//

To belabor my point, neither would a corporation, union, neighbor, or computer intelligence so stoop …