The problem of 99% products and what to do about it
OK, I know I can be a bit of a luddite sometimes, which is a strange thing to admit to after a lifetime of innovation in tech, but sometimes you just know that a product is not going to live up to the promise.
Take driverless or autonomous vehicles (AVs). I’m going to stick my neck out here and say that they are not going to achieve the aim of driving safely through everyday conditions on the types of journeys we all drive every day. Why? Well it’s not the tech, though I suspect the tech has still a long way to go to start dealing with edge cases like single track country roads, narrow city streets crowded with parked cars, magic roundabouts (US: forget it you don’t want to try to imagine a turn circle made up of turn circles), and four way stops (GB: forget it you don’t want to imagine an intersection where no one has priority). No, I guess the tech will get there in the end. It’s the people!
When I was 14 years old, living near a fast busy road, my friends and I decided that it would be a fun game to collect cat‘s eyes (US: Botts’ dots?). We would wait for a gap in the traffic, then run to the middle of the road and prize out the small retroreflective glass lozenges, then dash back to the pavement (US: sidewalk). Incredibly stupid, but that is the point, adolescent boys will do incredibly stupid and dangerous things for fun.
Now, imagine those adolescent boys out at night, maybe a little high, maybe just on exhilaration. What fun do AVs, with their wealthy passengers, offer? Lots! How about loitering by a set of Traffic Lights (US: Stop Lights) and when an AV stops, run out and put traffic cones (US: pylon?) just in front and behind it? Laugh yourself stupid as you watch the vehicle stalled and the passengers not daring to get out and face the ‘dark haired dangerous school kids’. Next move on to a game of AV chicken. How long you can leave it before stepping out in front of one and seeing it either perform an emergency stop or swerve wildly away, either way seeing the wealthy passengers thrown around will be the pay off. Having mastered that they’ll move on to leaving it too late and rolling over the front of the bonnet (US: hood) (yes, we used to do that too).
I just hope the tech companies trying to build AVs have some experts in adolescent behaviour.
I’ve not even started on the cultural differences in driving behaviour, signalling (wave, nod, finger, etc), though I’ve highlighted the language differences to prove the point.
So, if the kids are going to foil the revolution, what is the likely outcome? I think there will be three types of reaction: legal, physical and product definition.
The are tech giants building these vehicles have already had laws changed to get their test vehicles on the road. Why not continue to use the lawyers to get around these problems? Expect new laws prohibiting the unnecessary obstruction of a vehicle, and a change in the assumption of responsibility from the driver (there isn’t one) to the pedestrian. In the UK pedestrians always have the right of way, even if they just step into the road (there is no law of jaywalking) so reversing this would be an obvious target. Big techs will be suing the kids for their antics.
A more pragmatic response is to accept that they are imperfect and do what we did in the past when we wanted vehicles to travel faster than was safe for other road uses s – build protected roads like motorways (US: freeways), a disappointing climb down from ‘autopilot’.
I think having a message about your product which expresses a vision which you know is a massive stretch is a good thing. It conveys your direction of travel and opens people’s eyes to a previously unimaginable future. I did it when we built the first totally automated camera tracker, boujou. We knew that there would be shots where we would never be able to resolve the camera parameters and position, but once we got to around 80% we started adding tools to manually help the algorithm to find the solution.
I have read that the AV tech companies think they are 99.9% there. Presumably on some whacky asymptotic scale where the last 0.1% takes infinitely long. It is good to keep pushing on the tech and constantly refining it, but you must be realistic about the product offering. So, have that amazing stretch vision for where your product is heading, only by having that siren vision will you, and your engineers, discover what is actually possible, because you do not know yet. But, hold that vision lightly, be prepared to adapt it and most importantly be prepared to make great products that will delight.
I read in The I newspaper on the 15th of February that “Cities like Amsterdam, which have a lot of bicycles, are ‘very difficult’ places to start autonomous driving, according to a KPMG report”. Really? Most pedestrians who are not from Amsterdam find its streets difficult to safely negotiate without ending up colliding with a tram, taxi, car, or yes, bicycle. Imagine when they are ridden by obstreperous teenage boys!