4 ms·

Frankly: given recent news (hacked jeeps, software used to cheat on emission tests) I think it is time to _require_ _all_ software used on vehicles that transit

Frankly: given recent news (hacked jeeps, software used to cheat on emission tests) I think it is time to _require_ _all_ software used on vehicles that transit on public roads to be open sourced and available to the public. (for independent testing and verification in the name of public safety)

lucb1e 11y ago
Vehicles aren't the only things running software that might need verifying. Far from it. While I wholeheartedly agree that it would be an amazing step, I also fear the huge backlash from the car and IT industry. Doing this more gradual probably works better, perhaps requiring open source (I'm not talking freedom of modification or redistributing or anything yet, just viewing source code) for software older than x years (like the patent model a bit, 20 years or so). Of course Windows 10 wouldn't have to be open source just because Windows is older than 20 years, but Windows 95 would be. Anyway, that is just one of the many possible ways. I just think we should be weary of people stuck in the old ways and not limit the move to just cars.
- cyphar 11y ago
  > Doing this more gradual probably works better, perhaps requiring open source (I'm not talking freedom of modification or redistributing or anything yet, just viewing source code) for software older than x years (like the patent model a bit, 20 years or so). In that case, the car software is still proprietary software so you've gained very little from making the source code public. If a vulnerability is found, you can't write a patch for it and you have to wait for BMW to "get around to you". Car software should be free software upfront.
  - lucb1e 11y ago
    > Car software should be free software upfront. Yes! Just like medical implants (and the server-side counterparts that control them), and my password manager, and actually any software that is in between me and my password manager (OS, drivers, perhaps a browser), and private stuff like my email client/server, and instant messaging client/server, and things that track my vitals like those smart watch-like wristbands, and any non-domestically developed software the government uses, etc. Except few people aside from a few "open source radicals" will think like that. For every person I know who will fully support this motion, I know five who will object and another one or two who aren't really sure, and I already have a biased social circle because I do believe in this cause. As it stands, we never get to peek under the hood. A system like this would at least make people aware that "gosh, if we implement this cheating now (or whatever), we have to make sure to cover it up before we have to go public with it. Better document extensively where it's implemented so we can remove it thoroughly". Which is a bigger step than just one manager going "do it". And we would be able to see trends over the past years and reflect necessary changes in new policy. Perhaps two decades too late by then, but as it stands, we have naught. Then once people get used to the idea that in 20 years, when the software is commercially useless anyway, perhaps we can open source it sooner. Or more companies will go "look at us, we publish after 5 years for transparency without loosing our cutting edge!" A bit of wishful thinking perhaps, but I'm sure some will, and everyone would be forced after a certain time anyway. But once again, this is just one way of doing it. There are almost certainly better ways I didn't think of yet. We should just be thinking about it in the first place.
- Qwertious 11y ago
  "Perhaps requiring open source (I'm not talking freedom of modification or redistributing or anything yet, just viewing source code)" The term you're looking for is "source-available", which means precisely that the source code is available for viewing, and does not imply anything about conditions or lack of.
  - lucb1e 11y ago
    Right, I heard that term before but didn't think of it. Thanks :)
benwerd 11y ago
Or at the very least, available to the safety testers! https://medium.com/building-an-open-startup/untestable-unsafe-and-on-the-freeway-8cd58c0f2984#.msla6xjvz https://medium.com/building-an-open-startup/untestable-unsaf...
boobsbr 11y ago
I think this problem is impossible to solve, and it's the same on faced by electronic voting. How can you guarantee that the software running in the car is the same one provided by the car manufacturer? A more detailed explanation: You can't guarantee that software provided by the manufacturer and the dump you could retrieve from the car – through an API call, or through forceful means – are the same as the one which is running in the car because you can't know what's going on the silicon inside epoxy packaging of the myriad of chips in the board Even if the hardware is completely open sourced, the manufacturer could use a different/modified microprocessor packaged/labeled/branded just like the one specified in the schematics, so that it could internally store and run the shady code from the manufacturer, while giving you a perfectly legit and signed dump as the one provided by the manufacturer. Although I do agree that opening the software and hardware to public scrutiny would be a massive improvement on the actual situation.
- nathan_long 11y ago
  > it could internally store and run the shady code from the manufacturer, while giving you a perfectly legit and signed dump as the one provided by the manufacturer. If you had the right to replace it with your own copy, you could make it blink lights in morse code at various points in the software to prove it was executing your code. Of course they could still do sneaky things, but it would be harder.
- wepple 11y ago
  I don't think it's impossible. Manufacturers should provide the software for audit and assessment. Car tinkerers[1] can usually quite easily pull various firmwares directly off chip, and if the two diverge in a significant way the car company will have a LOT of explaining to do. [1] http://opengarages.org/handbook/ http://opengarages.org/handbook/
dognotdog 11y ago
At this point in time, having the software available would do more harm than good, considering that 90% (or more) of software development in the auto industry seems to be done by electrical or mechanical engineers with little to no training in software development, and even less knowledge about security and vulnerabilities. The auto industry does have a plethora of coding standards and software processes, but not enough institutional knowledge and foresight to produce secure software for connected devices. The fact that it hasn't been too catastrophic so far is more a testament to the technical limitations of the bus systems and ECUs than real security, but that is rapidly changing with the advanced hardware being put in cars today.

lucb1e 11y ago

Vehicles aren't the only things running software that might need verifying. Far from it. While I wholeheartedly agree that it would be an amazing step, I also fear the huge backlash from the car and IT industry. Doing this more gradual probably works better, perhaps requiring open source (I'm not talking freedom of modification or redistributing or anything yet, just viewing source code) for software older than x years (like the patent model a bit, 20 years or so). Of course Windows 10 wouldn't have to be open source just because Windows is older than 20 years, but Windows 95 would be. Anyway, that is just one of the many possible ways. I just think we should be weary of people stuck in the old ways and not limit the move to just cars.

cyphar 11y ago
> Doing this more gradual probably works better, perhaps requiring open source (I'm not talking freedom of modification or redistributing or anything yet, just viewing source code) for software older than x years (like the patent model a bit, 20 years or so). In that case, the car software is still proprietary software so you've gained very little from making the source code public. If a vulnerability is found, you can't write a patch for it and you have to wait for BMW to "get around to you". Car software should be free software upfront.
- lucb1e 11y ago
  > Car software should be free software upfront. Yes! Just like medical implants (and the server-side counterparts that control them), and my password manager, and actually any software that is in between me and my password manager (OS, drivers, perhaps a browser), and private stuff like my email client/server, and instant messaging client/server, and things that track my vitals like those smart watch-like wristbands, and any non-domestically developed software the government uses, etc. Except few people aside from a few "open source radicals" will think like that. For every person I know who will fully support this motion, I know five who will object and another one or two who aren't really sure, and I already have a biased social circle because I do believe in this cause. As it stands, we never get to peek under the hood. A system like this would at least make people aware that "gosh, if we implement this cheating now (or whatever), we have to make sure to cover it up before we have to go public with it. Better document extensively where it's implemented so we can remove it thoroughly". Which is a bigger step than just one manager going "do it". And we would be able to see trends over the past years and reflect necessary changes in new policy. Perhaps two decades too late by then, but as it stands, we have naught. Then once people get used to the idea that in 20 years, when the software is commercially useless anyway, perhaps we can open source it sooner. Or more companies will go "look at us, we publish after 5 years for transparency without loosing our cutting edge!" A bit of wishful thinking perhaps, but I'm sure some will, and everyone would be forced after a certain time anyway. But once again, this is just one way of doing it. There are almost certainly better ways I didn't think of yet. We should just be thinking about it in the first place.
Qwertious 11y ago
"Perhaps requiring open source (I'm not talking freedom of modification or redistributing or anything yet, just viewing source code)" The term you're looking for is "source-available", which means precisely that the source code is available for viewing, and does not imply anything about conditions or lack of.
- lucb1e 11y ago
  Right, I heard that term before but didn't think of it. Thanks :)

benwerd 11y ago

Or at the very least, available to the safety testers! https://medium.com/building-an-open-startup/untestable-unsafe-and-on-the-freeway-8cd58c0f2984#.msla6xjvz https://medium.com/building-an-open-startup/untestable-unsaf...

boobsbr 11y ago

I think this problem is impossible to solve, and it's the same on faced by electronic voting. How can you guarantee that the software running in the car is the same one provided by the car manufacturer? A more detailed explanation: You can't guarantee that software provided by the manufacturer and the dump you could retrieve from the car – through an API call, or through forceful means – are the same as the one which is running in the car because you can't know what's going on the silicon inside epoxy packaging of the myriad of chips in the board Even if the hardware is completely open sourced, the manufacturer could use a different/modified microprocessor packaged/labeled/branded just like the one specified in the schematics, so that it could internally store and run the shady code from the manufacturer, while giving you a perfectly legit and signed dump as the one provided by the manufacturer. Although I do agree that opening the software and hardware to public scrutiny would be a massive improvement on the actual situation.

nathan_long 11y ago
> it could internally store and run the shady code from the manufacturer, while giving you a perfectly legit and signed dump as the one provided by the manufacturer. If you had the right to replace it with your own copy, you could make it blink lights in morse code at various points in the software to prove it was executing your code. Of course they could still do sneaky things, but it would be harder.
wepple 11y ago
I don't think it's impossible. Manufacturers should provide the software for audit and assessment. Car tinkerers[1] can usually quite easily pull various firmwares directly off chip, and if the two diverge in a significant way the car company will have a LOT of explaining to do. [1] http://opengarages.org/handbook/ http://opengarages.org/handbook/

dognotdog 11y ago

At this point in time, having the software available would do more harm than good, considering that 90% (or more) of software development in the auto industry seems to be done by electrical or mechanical engineers with little to no training in software development, and even less knowledge about security and vulnerabilities. The auto industry does have a plethora of coding standards and software processes, but not enough institutional knowledge and foresight to produce secure software for connected devices. The fact that it hasn't been too catastrophic so far is more a testament to the technical limitations of the bus systems and ECUs than real security, but that is rapidly changing with the advanced hardware being put in cars today.