Doghouse – IoT and Resource Consumption

Recently I was giving a talk at a virtual conference on how to start your own business using free culture. I mentioned the basic, underlying premise that no one ever buys software or hardware. People buy a solution to a problem, whether that problem is running a business or playing a game. The fact that they find it easier to solve that problem with computer hardware and software is a side issue.

A young man started giving a presentation about how we had gone down the "technological rathole" and how technologies like IoT should be eliminated due to the amount of data transmission that would be used by IoT as well as the amount of electricity that was used by computing in general.

He gave his credentials as being a developer in a networking organization.

I must admit to being shocked, for I view the use of computing resources as something completely different.

Yes, there are people who leave their computing systems on all the time even when they are not actively using them, but these are typically low power devices that use much less electrical power in the long run than the alternative methods of solving the same problem.

There are people who give little or no thought to where and how they should be "solving their problem," choosing to ignore a good solution for "one that works" however poorly.

As people are beginning to realize, not everyone needs a very powerful desktop or laptop computer to do the computing they need day to day. Many people are doing web work or types of "office work" that only really require a Chromebook or Netbook type of computer.

It was slightly different when I started in computing in 1969. A single transistor could cost $1.50 in US dollars, with a power transistor costing $2.50.

Even mainframe computers costing millions of dollars (and that was when one million dollars was "a lot of money") typically did not have even a 32-bit address space, much less a 64-bit address space – it was simply too expensive. Today you can get a million transistors in a single chip for less than $1, with most CPUs having billions of transistors on them.

Computers "in the day" used to use kilowatts of electricity and use water cooling to execute solutions that would be deemed "simple" today.

I remember a comparison long ago saying that if computers and automobiles had evolved at the same pace, cars would be able to drive around the world on a cup of gasoline. Of course this was a simplistic comparison, but computers have come a long way.

Yet go into a business and you may find many desktop computers that are turned on, with disks filled with data, using electricity and generating heat that air conditioners have to remove.

What most people want today is fast and consistent access to the data that they need at this moment. People do not want to wait for their computer (or phone) to boot, and they want their data to be sent to wherever they want to be.

The person who was giving this talk was really concerned about how IoT would be sending gigabytes of data over the network and how much electricity would used and how many new routers it would need and so forth. He was particularly anxious about the amount of energy it would use for autonomous cars.

Today the controls of a car are seldom controlled physically. What used to be mechanical linkages that transformed to hydraulic controls are now electronic sensors and actuators. It is cheaper to put in electric windows than it is mechanical hand-crank windows. Most cars have multiple computers in them already.

I pointed out that if cars were truly autonomous we would need a lot fewer cars. I would foresee more cars "rented" only for the time you actually need them, with no cars parked in your driveway or at your place of business. The number of cars manufactured would drop. The number of buildings and parking lots needed for garages would drop dramatically. The amount of resources saved would dwarf the needs for producing the computers (which would probably already exist in the car anyway).

Cars would rarely be coming to a stop at a stop sign or stop light, but simply slowing down as they got to an intersection, and would weave through the intersection. Energy would be saved.

Much of the processing of data would be done "on the edge" of the local network, never having the raw data sent back to a central processing place.

While we should pay attention to data and electricity costs, time has shown that use of computers and data processing typically saves energy and resources with things getting better, not worse.