Friday, 23 November 2012

The dangers of a fixed phase relationship.

By Iridescent
While I was walking home today, I notice that 2 cars were parked with their hazard warning lights flashing away and they had a fixed phase relationship. What that means is that the second flashed at the same amount of time after the first every time. This was between vehicles of two different marques.
Now while this might not seem a big point it is very new. In the past electromechanical system were used and this had a lot of varied depending on the actual specification of the devices, plus the the voltage and current from the battery and the temperature, on top of this was the different designs used by different manufacturers.
While it's unlikely that the lack of natural variation in indicator lights will cause any big problems, it is indicative of the grown synchronisation of all things. Somewhere, buried deep in the car, was a crystal whose vibrations were being counted and that crystal was very very predictable, it and its ilk are buried in good everywhere from Quartz Watches, to computers and washing machines. All counting time in very accurate nanoseconds.
Soon we will all these devices will be networked and to make things simple that will all be synchronised via the Network Time Protocol to one of a few Stratum 0 time servers which are calibrated by state of the art atomic clocks and synced with each other. Since the 60s they have been keeping Coordinated Universal Time, which is as near as damn it good old GMT, or as NASA refers to it "Zulu time".
So what harm could this possibly do?
Already the national grid has to predict when everyone will switch the kettle on during major TV events, what happens when all the kettles switch themselves on, not just over 5 minutes but over 5 microseconds. The cumulative surge voltage would be amazing put an enormous spike in the power demands of the county.
The world we live in is becoming more and more synchronised with finer and finer precision and that may cause problems, in the same way as people walking over London Millennium Footbridge did.
The Tacoma Narrows Bridge had a not unrelated problem once it started swinging at it resonate frequency, the oscillation just got bigger and bigger till the bridge destroyed itself.
A world of lots of interconnected devices has the potential to form a similar system causing power spike or message storms that become more and more synchronised and bigger and bigger till something blows up.
It is possible to stop this happening with the bridges but it is also possible for a new mode of excitation and reinforcement to come into play as happened with the London Millennium Bridge or one that the designers forgot to take into account, in that case synchronous lateral excitement. Bridges are have always been designed even when man was using rules of thumb they were created could be observed and if need a new rule added. The Internet has no such designer and no such design. Is it possible to test a whole global system for positive feedback between previous unrelated events and involving lots of different devices. I doubt it, we can avoid the problem by not syncing things that do not need syncing or by adding a bit of random noise to those devices which don't need timings to the millisecond. It will annoy the hell out of me and my fellow borderline OCD Geeks that our kettles doesn't switch on at the time we specify, down to the millisecond but it might just be worth it.

No comments:

Post a Comment