Tread carefully when exploring the Internet of things – and for goodness sake keep your vehicle locked when it’s unattended!
With the Internet of Things largely unexplored, we are still in the process of sorting the mysteries out. These potential hazards fall largely into one of two categories first identified by US defense secretary George Rumsfeld: the Known Unknowns and the Unknown Unknowns.
To exemplify these categories, consider one of the emerging problems with unmanned car travel. According to sources at GE Software, one of the disadvantages of travelling in a driverless Mercedes is that all the other cars push in front. Or they do in the trials.
The automated car is at a disadvantage as it lacks the guile and cynicism of their human counterparts, who know they can nose in ahead at will. Yes, even those annoying menaces who drive up the slow lane and try to push in from the left! The poor driverless cars are too nice and will slow down as soon the car in front gets too close. Even if it got there by breaking the rules and pushing in.
We should have seen this problem coming. Given our knowledge of human nature, this IOT problem should have been one of the Known Unknowns. We knew human and robot drivers would clash – that much was never in doubt – but we weren’t sure how. My money was on the machines being the aggressors, but it turns out I was wrong.
How do we address this?
The programming of robot driven cars will have to become more complex, so they have the same sophisticated range of options open to human drivers. They may decide to flash their lights at the other driver, or lean on the horn, or close down the gap with the car in front to prevent anyone overtaking them. An algorithm could be developed to weigh up the risks, as no two situations will be the same. Somehow the robot driver must be able to work out when to call the other driver’s bluff and when to back off because they’re dealing with a nutter.
Humans do this by looking at a number of factors. A quick glance at the other driver’s eyes gives an indication of their mental state. This value should be cross referenced with an assessment of their car. Is it valuable? Is it a type typically driven by psychotics? If you calculate that the overtaker is dangerous and their car is of no great value to them, you’re probably better off letting them in, as they’ll almost certainly win any game of chicken.
Personally, I would like to see every fifth driverless car to be programmed as a complete psychopath. It wouldn’t slow down for anyone who pushed in and the vehicle must be indestructible. The outward appearance of these auto-psycho cars would vary too, so opportunist lane hoppers would never know which cars they could bully.
The Unknown Unknowns will be much harder to tackle. These imponderables are the responsibility of Mark McLaughlin, once a telecoms security advisor to President Obama and now CEO of Palo Alto Networks.
As more intelligence is loaded into the agents at all points of the Internet of Things, the more dangerous they will become, he warns. How dangerous, nobody knows, but the potential is there to hijack everything from drones to driverless cars.
Any IoT endpoint device may be the toe-hold that attackers need to move around laterally around a network until they get access to high-value data or people or whatever else we hold dear.
These days, time is our most precious asset. Much of my free time has been mugged by the marketing industry, as I spend all day opening unwanted emails and trying to unsubscribe from free offers with hellishly complicated small print.
The Internet of Things could help us all get more of our time back. But don’t be tempted to save time by pushing in front of that driverless car. It could be a nutter!
