Certainly appears to be the case, looking at X owners experiences at the Tesla Owners forum it looks like they're delivering cars that just would never make it out of the factory in a normal manufacturer.

The fact that the car is totally awesome makes up for this for most people, but it is a very different attitude to most major manufacturers.

In think autonomous will ultimately make roads safer, but how long that will take is the question.

I think the problem is that Tesla is treating an automobile as a software product, and letting the public beta test it. They have been proclaiming how their cars have been updated with autopilot, while in fine print saying don't trust it and pay attention while the car is driving.

The auto industry doesn't work like that, at least in America, for what I would think are obvious reasons. No longer is beta software something mostly innocuous, where if it goes wrong it is annoying, but rarely more. Now the safety of not only the driver, but others on the roads are affected.

Putting a feature like autopilot into a car and not expecting it to be abused is naive. It needs to be bulletproof if they are making the claim the car drives itself.

There is a related but separate factor, that of driver competence reduction precisely as a result of frequent exposure to 'automated' self-driving. The driver, who is currently expected to take manual control at the very moments of difficult driving and unexpected situations, is actually less competent to do so as the proportion of driving time under automated control increases.

This 'Glass Cage' article describes the lessons of highly automated aircraft and the correlation effect on competent pilots without recent exposure to manual flying that rapidly degrades their abilities.

Combined with the human control transition time problem, I anticipate more 'self-driving' car negative outcome news events as the cars get better and more people use the auto-pilot modes more intensively.

As an aside, I find it remarkable that Tesla (the brand) is seemingly able to avoid deeply negative press despite documented crash and near crash events, quite different from the 'sudden acceleration' media frenzy that Audi (and other brands) have suffered through in the past. Perhaps the 'press' (such as it is these days) is mostly happy (profitable) using Tesla as a positive news story.

This is interesting, as I largely agree, but from a radically different perspective.

Certainly Teslas are "computers on wheels", as they described their cars. And certainly that is a unique view in the automotive industry. Tesla does releases "beta" products and lets users test them, and does so openly. They actually advertise this quite a lot.

Per se, I think that is good, actually, and also unavoidable considering how the industry is obviously evolving. Even without Tesla pushing in that direction more than anyone else, such change in the industry was/is easily predictable.

That does not mean less safety, though. Equating "beta" to "unsafe" is wrong, I think.
When it comes to safety, I consider Teslas significantly safer than virtually all competitors' cars. Statistics on Autopilot Teslas seem to prove they ARE in fact safer. If then you consider how irresponsible many Tesla owners are being, I am impressed at how good the Autopilot is.

Also, I don't think other less software-minded manufacturers are bringing cars to market that are less "beta" then Teslas. I'd think the opposite, both when it comes to safety and when it comes to other features. My 2003 Audi A3 had a poorly designed clutch that would let 1st gear stuck (manual transmission) and traction would be there even with clutch pedal all down (!!), or, it cut my hand (blood on it) because of poorly metal finishing on the shift lever (I almost ended up in the other lane because of the sudden pain and surprise). And, these are two of the most popular issues with that "traditional" car, if you go and read forums of those years.

I think software for self-driving cars is a prime example of software that can't be developed in the public-beta, agile, MVP, evolution-y kind of way which web development has made so fashionable. A half-arsed Facebook is a lot better than nothing and entices users in, but a half-arsed self-driving car is worse than nothing, as human nature inevitably leads to people trusting it more than it deserves, despite the fine print.

Peter

I wonder how much testing they did of this on real roads before releasing it. I have to imagine it wasn't a fraction of the amount that Google has been doing.

On how safe it actually is. I agree with Vanity Fair.

http://www.vanityfair.com/news/2016/07/h...=social_twitter

I'd agree with that too.

This is another angle on the autopilot story.

Someone on twitter linked that story and said, "so does this bring our net Tesla fatalities back to zero now?"

I saw a fantastic talk about the interplay of ethics with this (and other internet-of-things stuff) at Open Source and Feelings in Seattle a few weeks back. When it's online, I'll link it.