5 ms·

Why don't they just experiment to fully determine the elasticity of demand and then -- if they determine it's not profitable -- cut their losses while they're a

Why don't they just experiment to fully determine the elasticity of demand and then -- if they determine it's not profitable -- cut their losses while they're ahead? I think the strategy of trying to maintain a monopoly to starve others out is daring. There's no guarantee fully autonomous vehicles will be the norm anytime soon enough for the business to capitalize on it. There was an article on here -- days, weeks? ago -- about how easy is it to fool the on-board cameras of autonomous vehicles. Slap some black rectangular bars on stop sign, and now all of a sudden it's recognized as a speed limit sign. I think we're farther away from Level 5 than what is led on. Which is a bit of a shame, I'd love to be in a world where I don't have to drive myself, but I'm also being realistic. Ultimately, in my mind, uber is failing at, well, failing fast. Don't dance around the worry that your business might not be profitable at the price point you need customers to accept. Otherwise, I think they're just wasting time. You need to validate product-market fit as quickly as possible...

xapata 9y ago
> fool the on-board cameras Those kinds of situations can be solved by training on adversarial examples. It's easy to fool them now, but perhaps not after they've prepared for that possibility. Some years ago, one could fool a web search engine by repeating a word dozens of times on a page. It's not so easy now.
- Y7ZCQtNo39 9y ago
  If they're not training autonomous systems for these extremely simple use cases, I'd imagine there wouldn't be much confidence in what else will be missed by the "autonomous" system. Out of the infinite number of possibilities one can encounter while driving, can the systems be trained to apply common sense? You can't program an infinite number of if-then conditionals for every possibility. The system has to be able to "think" and apply reasonable interpretation, like any human would when taking in any visual or audible input and converting that into a vehicle maneuver. The data today suggests that autonomous vehicles get into less accidents per vehicle mile than human drivers, which is promising. But who knows if autonomous vehicles are being used in complex situations representative of human driving conditions: poor weather, non-highway driving, etc. If we're putting them on the 101 in sunny weather, then yes, I'd expect them to perform reasonably well and definitely better than the average human.
  - xapata 9y ago
    > extremely simple use cases When's the last time you saw someone put black tape over a stop sign? These adversarial examples are not simple use cases. They're situations that humans rarely if ever encounter. > apply common sense ... "think" Ever seen the "invisible rope prank"? Humans are easily fooled as well. > if-then conditionals That's not the way statistical machine learning works. It's layers upon layers of probabilistic functions. > who knows if ... representative of human driving conditions I've read reports that a few of the autonomous vehicle companies are indeed testing/training on regular roads, which would allow a representative sample of human driving conditions. Frankly, that's just basic science -- don't experiment with a biased sample.
malandrew 9y ago
> I think the strategy of trying to maintain a monopoly to starve others out is daring. The irony is that they are not. They are subsidizing far less than any other competitor. Show me one competitor with lower losses as a percent of gross bookings.

xapata 9y ago

> fool the on-board cameras Those kinds of situations can be solved by training on adversarial examples. It's easy to fool them now, but perhaps not after they've prepared for that possibility. Some years ago, one could fool a web search engine by repeating a word dozens of times on a page. It's not so easy now.

Y7ZCQtNo39 9y ago
If they're not training autonomous systems for these extremely simple use cases, I'd imagine there wouldn't be much confidence in what else will be missed by the "autonomous" system. Out of the infinite number of possibilities one can encounter while driving, can the systems be trained to apply common sense? You can't program an infinite number of if-then conditionals for every possibility. The system has to be able to "think" and apply reasonable interpretation, like any human would when taking in any visual or audible input and converting that into a vehicle maneuver. The data today suggests that autonomous vehicles get into less accidents per vehicle mile than human drivers, which is promising. But who knows if autonomous vehicles are being used in complex situations representative of human driving conditions: poor weather, non-highway driving, etc. If we're putting them on the 101 in sunny weather, then yes, I'd expect them to perform reasonably well and definitely better than the average human.
- xapata 9y ago
  > extremely simple use cases When's the last time you saw someone put black tape over a stop sign? These adversarial examples are not simple use cases. They're situations that humans rarely if ever encounter. > apply common sense ... "think" Ever seen the "invisible rope prank"? Humans are easily fooled as well. > if-then conditionals That's not the way statistical machine learning works. It's layers upon layers of probabilistic functions. > who knows if ... representative of human driving conditions I've read reports that a few of the autonomous vehicle companies are indeed testing/training on regular roads, which would allow a representative sample of human driving conditions. Frankly, that's just basic science -- don't experiment with a biased sample.

malandrew 9y ago

> I think the strategy of trying to maintain a monopoly to starve others out is daring. The irony is that they are not. They are subsidizing far less than any other competitor. Show me one competitor with lower losses as a percent of gross bookings.