> And in every one of those stories there are ten people asking why this is a story because battery fires are less common than gasoline fires. System working as

> And in every one of those stories there are ten people asking why this is a story because battery fires are less common than gasoline fires. System working as intended. It's still FUD that they could rightfully perceive as hurting them. They're taking enough risks as it is, I can imagine that they don't need more criticism of their safety. > But suppose they don't. You can't have it both ways. Either someone allows modifications and whatever hypothetical bad reactionary laws that causes will come to pass, or nobody does and the laws are never passed but the result is the same. I'm not just saying it's about the laws (Tesla has had a hard enough time tangling with these, and has been specifically singled out by regulators on occasion, such as in Germany right now), but also about their brand. They're sensitive to criticism, and, therefore, to the possible sources of criticism - whether justified or not. > Uber is going to make a self-driving car and then sell them to the public so that members of the public can lease it back to Uber and Lyft? I'm saying that Uber is developing self driving cars (or rather, the self-driving systems) which they appear intent on using for their own fleet, rather than selling as a kit or whole car, which they could have done. Tesla has been selling cars already, but it's going to be selling a new kind of car which has certain Uber-like properties. There are a range of ownership options and this is not an inherently illegitimate one. > Well first of all that's false, having them fix the problem saves you the trouble of doing it. But even identifying problems doesn't work without understanding. 99% of the fix is thoroughly understanding the problem. If the system is a black box then the user doesn't know how it's supposed to work. So you simultaneously get spammed with false positive bug reports from users who think there could be a problem but aren't allowed to understand whether there really is, and lose true positive reports because weird behavior gets written off as "it's a black box" and the last three things they reported were false positives. Sure, understanding the problem often helps more than simply reporting "there is a problem". But they could allow people to inspect the code and diagnostics without allowing any changes, the first does not necessarily imply the second. And if you really allow people to get into the code deeply enough to make the kinds of changes that fix problems in the system itself (rather than some sort of API for strictly limited input) then you've potentially given a large amount of your system to your competitors.