How the Internet of Things Will Change the World: Not by Much

Are you ready for the enormous, revolutionary, ground-shattering changes coming with the IoT?

If you said “yes” and by “yes” you meant you were prepared for breathtaking changes, you are a naive child wading in a murky pool of lampreys, your will putty in the hands of the same charlatans who brought you terms like “cloud computing” which still has yet to be defined in any concrete technical sense.

If you said “yes” and by “yes” you meant that you felt that the more things change the more they stay the same — then you are indeed prepared.

Cold War II, civil war in China, the breakup of the EU, abolishment of American drug laws, the DEA and an end to the Mexican civil war all at once — those are the kinds of things that will have a measurable impact on life. The so-called “internet of things” concept as heard in internet marketing is… well, not at all what the guy who coined the term “Internet of Things” meant.

We already have an internet of things. Has it cured cancer yet? Nope. But if we put RFID in every part of our bodies we will certainly be even more exposed to the will of outside actors. Not that the public has demonstrated that it cares about complete loss of its privacy, especially when “Google style conveniences in exchange for your life’s data” can be backed up by the rhetoric of fear necessitated by government “anti-terrorism” funding. (Yes, I mock this, and yes, I was a Green Beret in the US Army for 6 years — the direction that rhetoric is headed is toward government empowerment, and the government is exactly the least well equipped element of society to deal with terrorism.)

Want to see an internet of things? Tesla cars receive system updates across the network now, and can turn in performance data to help the maker improve on their designs and software. Open water jetski robots can follow automated routes and report hydrographic and bathyrithmic data back to a data processing facility to chart change over time. I was working on a (now defunct, but promising) design project to develop spotting scopes that were intelligent enough to peer data amongst one another within an operational space and change “spotter calls” into more generally interesting “shot requests” and aggregate shot providers in the area to engage targets based on type, effect and following damage reports. Whenever any peers had a network connection they could aggregate data externally.

Dude, we’re already there.

What we lack is standards. Oh, wait, nevermind… we actually have tens of thousands of those. What we lack is standards that people actually can use, that aren’t harder to learn and comply with than the handling of the basic user problems at hand are. These problems will mostly never be solved, not really. Truly understandable data must have a semantic foundation, and semantics are essentially arbitrary in most ways that matter in data processing. That means data must either be tagged in a non-trivial way or must be placed into a schema where relationships are what have meanings.

Note that “tagged in a non-trivial way” above means taking tagging systems to such extremes that they become their own ontologies. Think about that. It should make your face turn pale. That’s at least as difficult as developing an arbitrary formal language. (In case you didn’t notice, an “arbitrary formal” language is a oxymoron — though consortia and governments alike love nothing more than funding committee efforts to formalize the syntax of futile efforts in this area). Writing even trivial software using such a tagging system would require that programmers at every step of the system learn this arbitrary formal language of tagging before they do much of anything, and that’s a lot harder overall than just continuing on with our pile-of-ad-hoc-systems approach. Schema-based systems, while having some different tradeoffs (computationally natural descriptions of data as a “shape”, for example, is a really big win in practical application), ultimately suffer from the same complexity explosion at some level. In particular, applying a particular schema designed in the context of one problem domain will very often not fit in the context of another problem domain — and fully normalizing all data ever, ever would eventually require an infinite (and ever growing) number of relational definitions. Yech.

So… Internet of things? Yeah. We’re already living it. Don’t get too excited and don’t give into the hype. Just because you technically can read data from remote sensors or activate your house’s appliances with your phone (hint: you already can) doesn’t mean this is something you will want to do, or that the venture capitalists of the world will peel their lips off the adsearch cock for long enough to realize that there are more interesting things they could be funding than bounce-under ads and invisible iframe reclick-to-click javascript tech.

Rest easy. Humanity’s material circumstances will continue to get incrementally better (save the occasional dip due to predictably stupid things we do to ourselves) until The Singularity when we are all either suddenly eliminated due to obsolescence, or drive ourselves into a new mode of existence-as-slavery to whatever Google turns into (when all data is network accessible, privacy does not exist, all data is the private IP of a single aggregate, the rights of conscious uploaded entities don’t exist, the definition of “life” is still “way way way after birth”, and continued conscious existence equates to paying a service charge — that’s not really life). None of this is particularly dependent upon the hype surrounding the “Internet of Things”.