I had to blog about it, even after putting up a Facebook post, in part because part of the hype and touting from the smart-stupid I *** Love Science. (I'm not a prude, but expletives used just for "coolness" I have the option, and the sometimes desire, of deleting.)
Still MUCH more science fiction than science reality. There was a story a week or two ago, about exactly how Google's self-driving cars, to date, have had the success they have had. Answer? Massive, massive, massive amounts of data, even for a very limited, restricted area of driving, almost all in the Bay Area. To navigate the whole US with a driverless car? Such a critter won't be made for 50 years. Anything Google sells before then will have a "EULA" more than a mile long, which will include terms strictly limiting where you can drive one of its cars.
Alex Madrigal wrote about that data gathering at the Atlantic earlier this month, as part of the hype versus the reality on this issue. Beyond actual miniscule road testing, there's this:
Today, you could not take a Google car, set it down in Akron or Orlando or Oakland and expect it to perform as well as it does in Silicon Valley.And, do you and I really want Google taking a million times more photographs with 100 times more precision than Google Earth now does? And, no, those numbers aren't hyperbole; they're off-the-cuff guesstimates. Lots of red-state gun nuts would be shooting at Google Cars photographers, methinks.
Here's why: Google has created a virtual track out of Mountain View.
The key to Google's success has been that these cars aren't forced to process an entire scene from scratch. Instead, their teams travel and map each road that the car will travel. And these are not any old maps. They are not even the rich, road-logic-filled maps of consumer-grade Google Maps.In short, Google's modified commercial cars have been using a massive cheat sheet. But, the reality of trying to have cars impersonally drive the entire US is massively different. How different?
They're probably best thought of as ultra-precise digitizations of the physical world, all the way down to tiny details like the position and height of every single curb. A normal digital map would show a road intersection; these maps would have a precision measured in inches.
Very few companies, maybe only Google, could imagine digitizing all the surface streets of the United States as a key part of the solution of self-driving cars. Could any car company imagine that they have that kind of data collection and synthesis as part of their core competency?Urgh.
Whereas, Chris Urmson, a former Carnegie Mellon professor who runs Google's self-driving car program, oozed confidence when asked about the question of mapping every single street where a Google car might want to operate.
So far, Google has mapped 2,000 miles of road. The US road network has something like 4 million miles of road.
And, the hype is worse elsewhere. The preview for last night's Charlie Rose had a clip of a guest talking about the GoogleBug learning from algorithms about how humans drive. Wrong! That will not be what's happening next to this car, because that's NOT what Google was doing.
Also, as far as the data, this is a Google Refrigerator on steroids. Do you want GOOGLE making a driverless car, one with telemetry that will tell you every destination to which you drive? I know I don't.
Beyond that, and in light of that, a GoogleBug, or, for the Atlantic story, Google's modified Lexuses, are about where moble robots were a decade or two ago with a massively higher learning curve. How much higher?
Besides, the average American in "flyover" territory is NOT buying this car. An all-electric that, even at its miniscule size, gets only 100 miles of range? And only goes 25 miles per hour?
The Guardian has more on the car side of the issue, namely the limitations Google hit with re-engineered street cars.
Hey, IFLS? Last I checked, sociology and psychology were sciences; you might want to bone up. It's not just IFLS, but, they're an easy target.
Besides, this leads to another bone to pick.
This car isn't really about science at all. It's about technology. That's entirely different. As is engineering. Which would have told you about those 3.998 million miles of unscanned roads.
That said, there is a science called "environmental science." Increasing use of mass transit is more environmentally sound than, even in the Bay Area, building a bunch of electric quasi-cars.