Self discharge on a 60kWh battery shouldn't be more than say 10 Watts, I believe.

Tesla batteries shouldn't be self-discharging faster than say 10% per month. That's like 0.3% per day. Plus, I think this guy had a 60kWh car, so your 1.3% is too low.

A car battery contains about 1 kWh of power. So this kind of draw would drain a car battery in on day. You could probably leave a car parked for a month-or-so without worrying about the battery, so figure the Tesla is using power about 30x faster than a normal car. That further implies a normal car is running at about 1.5 Watts (which sounds about right for a computer running in low-power mode and occasionally checking for things like a nearby key fob for keyless entry).

Of course, you'd expect to lose charge in a 60kWh lithium ion battery at a rate of about 5-10 Watts. Adding the 1.5 Watts that a car's computer can expect to use, and the Tesla should be using about 6.5-11.5 Watts when parked. I can't tell you where the other 33.5-38.5 Watts is going.

'Netflix might say, "I'll pay in order to make sure that my subscriber might receive the best possible transmission of this movie."'"

Isn't that exactly what net neutrality people are worried about? Because it's hardly a big jump from that to "pay us or your subscriber will get the worst possible transmission of a movie".

My position has always been "I am the ISP's customer. I am not the thing they sell to Netflix." If it's more expensive for the ISP to deliver me video than emails, that should be a negotiation between my ISP and me. It shouldn't be a negotiation betwen my ISP and Netflix, that I end up paying for anyway. Or even worse, that negotiation goes bad, and Netflix just sucks for me with no way for me to improve it... and my ISP tells me "but Hulu works fine... you should just switch to Hulu... trust us."

They don't draw a line. These people would want to see those other species free as well. At least they're consistent, I guess.

Rather than "explanations", why don't we call the untestable ones "wild guesses"?

But it's entirely possible that the whole fire vs age curve might just have a different shape between electric cars and gasoline cars.

One would expect the odds of fire in a gasoline car to increase with age as hoses/connectors wear out. Fire would result when those parts finally fail (whether by themselves or in a collision) in a way that sprays flamable fluid on hot parts (i.e. not every leak results in fire... it needs to leak onto something pretty hot).

But in an electric car, you'd expect the risk of fire to be much more constant over time, as the battery's impact resistance doesn't change with age. There might still be an increase with age, though, as accident rates rise with car age. But you wouldn't expect to see the fire risk rise as quickly as with gasoline cars.

So it's possible that a car like the Model S might have a slightly higher risk of fire when it's new (compared to new gasoline cars) and a slightly lower risk of fire when it's older (also compared to similarly aged gasoline cars).

This is why I get annoyed when Musk goes on one of his whining rampages about some recent coverage he doesn't like. He's totally happy to be the center of attention when the coverage is positive (however out-of-scale it maybe be with the importance of his company). But when the coverage is negative... suddenly he's like a 4-year-old who dropped his ice cream cone.

And the worst part is the Tesla fans who troll the internet forums to enforce the gospel.

Yeah, what did the Romans ever do for us? ;-)

Your generalization is a little too generalized. Modern natural gas cogeneration plants are 60% efficient. Coal doesn't come anywhere near that, though.

The other important distinction is angle. A ballistic missile is roughly headed directly towards the interceptor. A spy plane is roughly headed on a 90-degree angle to the interceptor.

If you imagine it like duck hunting, much easier to hit one that's headed directly at you, than one that's just flying by.

You didn't read my reply. I don't want wireless to work that way. I'm arguing that wireless is not a solution for allowing cars to travel that close together.

As far as stopping distance, you should leave high school physics behind and think about practical engineering. First, the sensors they will be using will have some inherent error within them. The car in front may also be swerving at the same time and it's front end will be dipping. Either of those will throw off any calculations of distance by multiple millimeters. If you're counting on matching that car's deceleration before closing a one meter distance, you'll simply need to do better than that. Even worse, what happens if the car in front of you has better tires/brakes than yours. Automobile deceleration rates can vary pretty dramatically, especially when starting at highway speeds (downforce varies by body style).

People love to talk about computers allowing cars to travel closer together. You can certainly let the computer trail closer than you'd want a human to trail, but we're not going to be driving at highway speeds with one meter separations. Computers simply aren't enough to keep that from being dangerous.

It's worse than that. If the system relies on a person sitting there doing nothing, but ready to take over in an instant... IMHO, that's worse than if the person just drives. My understanding of the current Google vehicle is that the drivers are often taking over for complicated parts of their journey, on a pre-emptive basis. I'm not that excited to have systems on the road, being driven by Joe Sixpack, where the driver has to (i) pay close attention while the car drives and (ii) make intelligent decisions about taking over control in anticipation of complicated situations coming up.

(For the record, Joe Sixpack includes me. I'd be terrible at paying good attention while the car drives itself. I don't even use cruise control because I feel like my reaction time -- getting my foot to the brake -- is slower when I use cruise control.)

The rear cars have to have some way to know to brake with maximum force immediately. Without that, the feedback loop will be too slow and they will close that 1m gap all too quickly. Or the alternative is for the autonomous car to slam on the brakes every time the car in front slows even a little bit.

Relying on wireless communications for that is a recipe for disaster, as wireless is simply not reliable enough. (You wanna bet your life that a WiFi, Bluetooth or similar connection doesn't get dropped at the wrong moment? I don't.)

Furthermore, if the front car swerves, instead of brakes, when it detects the deer, then a car following closely behind could be in a world of trouble, not have detected the deer due to the intervening car blocking the view.

You can't just say "fast computers" and suddenly ignore physical realities.

That may be, but one still need not blame the individual cop for that. That's really what my original comment was about.