Software Governance and Automobiles - Session 3a

Automotive Adoption and Governance of FOSS: Qt’s Challenges and Experience, by Alistair Adams.

ALISTAIR ADAMS: I just thought I’d throw up a slide since we had a question this morning. Talking about road deaths and somebody had said 34,000. It’s not 34,000, it’s actually gone all the way up to 40,000. So I just pulled this from Wikipedia and it’s a sort of number that I actually follow is what the road deaths are.

Where it says total deaths, it’s the third one down. That sort of light blue one there. You can see that sort of right about the turn of the century. It was coming down nicely, and then we had the Great Recession, nice and low there, and then it started climbing up again. You’ll see some people in the media say that while these peaks in deaths is due to distracted driving, I’ve also read other people say that it’s also due to the fact that after we came out of the recession, people were driving a lot more as well. So it’s probably a combination of the two. So I just thought that’d be interesting since that came up.

EBEN MOGLEN: The speaker is Alistair Adams, who needs no introduction, he is the angel of death.

ADAMS: There you go…

MOGLEN: But I’ll introduce him in a minute when the rest of the team is back in the room.

All right, so, thank you for returning after your lunch.

I guess you have sort of figured out the hidden message of all this, which is that I’m putting off self-driving vehicles for as long as possible. The elephant is in the room but still unacknowledged. I still want to deal with software that does things I believe software can do, unlike driving cars which I’m not at all convinced about.

The conversation of the morning having been about one aspect of the question of how to govern software, with respect to user modifiability in particular, leaves us asking some more questions about how do cars, as a particular place to put software, uniquely identify themselves one way or another? What is it that’s special about the problems of software in cars if we leave that driving part out of it?

So this next conversation is about the particulars of engineering systems that use software in automotive applications in particular. Alistair Adams was part of one of the great Nordic free software experiments. There was a period of time in which the Scandinavian world was the unique proving ground of ideas about free software and capitalism. Trolltech wasn’t necessarily my favorite part of that experiment because it wasn’t a GPL company in the beginning and the non-GPL licensing of Qt had vast implications in securing us diversity of desktops in free software operating systems, for example. But from Trolltech, Alastair has done the job of writing with Qt through a whole variety of places and times and is now, in addition to GENIVI and the sort of intra-industry consortium part of this story. Also an old time free software survivor in the automotive business. And it’s that which I wanted to capture even more than traffic deaths. So Alistair.

ADAMS: Yes okay, thank you, pretty honored to be here. I looked at the agenda on the overall topic and put together a talk which I hope talks to some of the diversity of the talks that you’re going to get here, so I’m going to give this a talk from what our experiences are in the open source in automotive and also talk about some of the economics of the open source software as well and how we deal with it.

So I’ll just say what is Qt, it’s pronounced cute. I don’t know how many people are familiar with Qt. Some are, some might not be– so I’ll just give you a quick overview.

It’s called Qt. The origins come from originally there was a toolkit called XT for building user interfaces, and that’s still around, a little bit. And so the original people liked the letter Q in the format and so it became Qt. And if you say Q t fast it sounds like cutey. And then you can abbreviate that to cute.

So it’s a cross platform application framework. So what that means is that it’s used for developing application software. So that’s things that the user sees that can be run on various software and hardware platforms, so that’s what the cross platform means. So that’s one of the nice things is you write software once, and it runs on Linux, Windows, Mac and embedded platforms and so forth. So you can do that with little change to the underlying code base.

I don’t want this to turn into a marketing thing, but sometimes just a few pictures just gives you an idea of what it does. So these are where people use it on the desktop. These companies are using it there for a cross platform. It’s used in tons of devices. It’s mind blowing, all of them. So if you came here on an airplane and it had a screen on it, chances are it’s running our software to do the display on that.

And then in the car… We’re in all the cars. All of the screens in the car, you’ll see Qt. Not necessarily everybody uses us, but quite a few car companies. So that’s the instrument cluster and that’s also the cockpit. We are not in the autonomous systems, which are controlling the car. So that’s a totally different piece of software. And there’s a lot of discussion of that going on here. So I just wanted to set the context for where we are.

So when I looked at the agenda for what we’re talking about, I just put it down here. Just to go through it. So the question was about how the software is governed. And I’ll talk a little bit about how we do that, and users exercising their rights in that software and so forth. And again that’s a tricky situation. I think Daniel has already spoken to some of that, have a look there. And so I’ll try to talk to some of these issues as we go, as I go through the presentation.

I think one of the things here is that Qt, we’ve been around for 25 years. As Eben mentioned, it’s gone through a variety of different changes from Trolltech, which had one set of licenses. It transitioned to Nokia, which had added additional licenses. And so what I’m going to do is just talk about what the motivations were for all these different licenses and why they changed and what happened. And I think that’s sort of insightful if you think about what the economics are of building open source software.

So, just looking at open source software, and automotive, and what’s the meaning of free, and I think the term “free” kind of gets a little bit controversial at times. Does it mean free as in beer or free as in the freedom to actually go change things? So I think that when you are looking at the open source and automotive, the two main ones here, and this is not the four down in the GNU Manifesto, but I think the first one is the ability to be able to inspect the code, and for reasons that have already been discussed, you want to be able to see how the code works. You want to be able to inspect to see if there are any bugs there, and see if there’s any malware. So it is one of trust.

And then again the freedom to be able to change the code so that if there is something that you don’t like in there, you want to be able to change. You want that freedom too. And if there are bugs, you want that ability to be able to change that, which sometimes you can’t do. That doesn’t necessarily mean that everything is free of charge as well. We’ll come to more of that later.

So open source is really good. And I’m going to talk a little real life example I had with some proprietary software, and this goes back to the days when I worked for Nokia. I worked for Trolltech, it was acquired by Nokia to build the user interface. You spent a lot of time in a company that’s trying to build mobile phones, and for those that followed, at some point there was this burning platform memo and then Nokia switched to using Microsoft as a platform for its mobile phones. And I was well involved in the project which became MirrorLink, which is a mirroring technology like Android Auto and CarPlay where you can project your smartphone into the vehicle. But to do that we needed an extension to the USB driver so you could run Ethernet over that. We needed access to framebuffer so we could actually send that over.

Well, we couldn’t do that. No one at Nokia could do that because we didn’t have access to the source code, and it wasn’t on Microsoft’s priority list to implement that. So that feature never actually got implemented. So that’s why open source is really beneficial to the industry, and that’s why having these freedoms to be able to modify the code is really, really important.

So what I’m going to do is just go through five different use cases here and types of economics of open source and how it works and what the pros and cons are. I think a lot of people, maybe more of the lay people, think that open source is people working in their basement for free in their spare time building stuff. Well, there’s a lot of that going on. And one example, I think Node.js started off like that one person who was sort of working away in his spare time and doing something, but then when that became mainstream, it rapidly became sponsored by another company.

And there are lots of small projects, lots of add-on modules for Python. I’m playing around with a little home automation system at home, and I’m using a lot of these modules that somebody is probably just written in their spare time. It’s fantastic. It’s great for that. And it’s also a great place for people in the industry to come out and learn what good software looks like and then to contribute to something and put out their trademark and get noticed.

And then you have things like open SSL. So open SSL had this thing called Heartbleed, which I’m sure everybody knows about. And this was code that was introduced in December 2011, and it was reported in 2014. And it turned out as somebody commented if you think about it, you’ve got two full time engineers who are writing, maintaining and testing 500,000 lines of code. And what I’m trying to get at here is that here you have the industry that’s relying on all this, but it’s something that severely under-resourced. And another quote from somewhere said that it’s not really…someone said the mystery is not that a few overworked volunteers missed the bug, the mystery is why it hasn’t happened already. And as someone said, it also may have resulted from failed economics.

Thinking of that, it turns out that someone estimated that 50 percent of the servers in the world are relying on open SSL. So what the message here is is that, yes, open source is great, but you still need people behind it to maintain it and open source is good, but you still need people to maintain it and keep it going. Otherwise, you end up with problems like this happening.

Then so you look at some of the other really, really big open source projects that are very successful here. And this is taken from the Linux Foundation, where they did a survey looking over a 12 month period, which were the largest projects. So they came up the top 30. And these are 14 of them, and most of these are probably very familiar names. So if you look at some of these, you’ve got Google’s doing three, you got React from Facebook, and other big ones from other big companies, and then you stop to think, well what’s the motivation for these large companies to invest in all this software and give it away for free? Well, in the case of people like Google and Facebook, they’re making tools that they want their engineers to use to be more productive, and by putting these tools out into the open source, they’re getting feedback from other people, other developers who are using them. Other developers will use them and learn from them. That in turn gives them more developers who know those tools. They get bugs reported that improves them, and then they also have a bigger source of engineers that they can recruit from who actually know these tools and can work for them. Then other ones, these companies also have different business models around it. So I think when we’re thinking about the open source and the economics, it’s important to understand what’s the motivations, the economic motivations behind all of these.

So a lot of other ones, another way that people make money from open sources is by the consulting, training, and support. So just to quote the bit there it says they do well because Hadoop is so complicated there’s lots of demand for training, consulting, and support. So the question then is well this is a business model that probably works for software. It’s really hard to use.

Now our philosophy in building Qt is that we want to make something that’s really easy to use. It’s a well documented framework, and it’s fun to use, and in fact at one time one of the quotes we get back from many people is that writing software C++ in Qt makes it fun again. So that kind of business model doesn’t really work for us.

And then there’s the moral obligations around the open source. So on opensource.com, they’ve got the phrase, “You got the penny jar, so you give a penny, take a penny.” And then I was at a Linux Collaboration Summit and Jim Zemlin said, “Well, the fantastic thing about open source, is you take a lot and you give a little.” Then associated with that is the copyleft licenses, where if you do make changes, you are supposed to make your changes available to everybody else so that they can benefit too. And that’s all great, but in our experience, there is a fair amount of corporate avoidance of that. There’s a lot of corporations that will just take the open source and not contribute anything back. Part of that is the legal teams inside those companies aren’t familiar with how to contribute things back, and there’s a lot of people who believe that IPR is their secret sauce for everything.

And then the the last one I’ve got the list is the the consortium approach to building open source, and in automotive, we’ve had two alliances here, and I’m sure Jeremiah knows all about the GENIVI one. And so to start off in really fantastic ideals, the point being is that a lot of the IVI software in there is commodity. Everyone has to build it. Everyone needs a Bluetooth stack. Everyone needs a media player, so why don’t we all just do it once and share it between all of us and reduce the cost of everything?

Well, the reality of that is partly due to the business model and the relationship between the Tier 1 suppliers and the OEMs is that the OEMs don’t actually write the software. The OEMs are probably the ones would really benefit from this, but they don’t really write the software, though that is starting to change. And the Tier Ones really were super afraid to give up, to contribute things, because they think, well if I contribute something, is everybody else going to contribute from it, or will they gain more from it than I will? The feeling was that that was their secret sauce, so to me I think GENIVI, its biggest success really was raising the idea of open source awareness within the automotive industry. It’s goal to creating a baseline of software for a stack, I think didn’t quite make it out there.

It’s funny I was just in a conversation with my boss yesterday. We were talking about this. Like you take Bluesy, for example, you’d expect that Bluesy is this open source Bluetooth Stack, and you would think everyone’s using it. But everything I hear back from people is no, you can’t use it, because the interoperability is not there. You would have thought, it would make sense for the whole industry to get together to do this but no everybody uses a proprietary stack instead.

The diagram there is one from a few years back, but I think it’s still valid in the GENIVI ideal, but the concept there was that you got the 80 percent-15 percent-5 percent rules. So what that means is 80 percent of the software is pure open source that you can take, the 15 percent is open source that you might modify and the 5 percent is your custom code, and that would be what is sort of in a generic GENIVI system.

So, like all things, there’s competition. GENIVI is very European-centric. And then the Japanese came along and say, well, we need something of our own. And so they created this thing called AGL, whereas GENIVI is very special occasion based, AGL said that code is a spec.

I should probably step back. I think that one of the reasons that GENIVI went specification-based is that you did have this conflict between people who say, well I’ve got my stack, and someone else has got their stack, and OK, so why don’t we just define the APIs for that so that we can have commercial components in there as well?

I kind of like the idea that code is a spec. It’s a bit of the Linus Torvalds mantra which is “Show me the code.” What’s interesting there is a funny story there. They set their membership fees at a half a million dollars a year. And I heard from one of the GENIVI board members at the time that they all sort of looked at each other and thought, these guys are nuts. There’s no way they’re going to get that. Then all of a sudden five members turned round. So five members bring you in two and a half million, and GENIVI had 130 members bringing in three point two million, so… It was rather interesting.

So from that they are managing to actually start finding companies to actually build components for it. But I’m still left with the sense, are the OEMs really serious about using it internally? And I think the thing that worries me is a lack of direct requirements coming from the OEMs into that project. Again is it moving fast enough?

And then there’s a lot of work that GENIVI’s done that they’re not picking up. So there’s an audio manager that GENIVI has that, as I understand it, is actually shipped in quite a few production vehicles. My experience as a software engineer is that something that shipped is way better than something that someone’s redesigning again. But as we all know software is another form of religion. So, you can’t always persuade people of different things.

So I’m not convinced that a consortium is the best way to build open source for cars because the two we’ve got here are not really convincing me that they’re being very successful.

So, with that said, I think it might be interesting to talk about the history of where Qt’s been and the business models that we’ve had. So it started off as a company called Trolltech in 1993. Eben sounds like he knows more about the history of it from the very beginning than I did. But we can come to that. But anyway, we ultimately ended up being a commercial and GPL GNU-licensed model. So that’s GPL, not LGPL. Really, the philosophy for that was very, very simple. If you’re making money building a product that uses Qt, then you should be paying us. If you’re going to build the product you’re going to give away, then you can use Qt for free. And if using GPL, it basically means that you’re linking to those libraries. That you have to give away, you have to make open the source code for your product as well.

So that worked really well. Our biggest competitor there at the time was GTK. And the way I’d like to put it is that whether or not we won the order depended on whether or not it was the engineering manager that made the decision or the purchasing manager who made a decision. If the purchasing manager made the decision, then it went to GTK because that was free. If it was the engineering manager, they chose Qt because it was a little better than GTK at the time.

So, along came Nokia. They had a bit of a messed-up ecosystem with their Symbian phones, and it was really hard for people to write code with it. We had these beautiful little diagrams where we said four lines of code you can do what the Symbian does in 20. And this goes back to what I was talking about earlier. What are the motivations for it? Why does Google and Facebook and so forth have all their projects open source? Why do they give it away? It’s a similar reason why Nokia did. They added the LGPLv2.1 license. The whole goal there was, we want as many developers as possible using Qt because the more developers using it, the more bugs they find, the more performance improvements they find. That’s a big benefit to Nokia in the end. And then when Nokia gets its smartphones out and its act together and starts shipping phones, there’s going to be a lot of developers out there who know how to write apps using Qt for the Nokia phones.

And there’s another reason, there is that the revenue that we were bringing in, which I know was about 25 million or something like that, it was a rounding error on the balance sheets. An even funnier story there is that we were brought into the devices division of Nokia, and they had no mechanism for recognizing that revenue. So all of that revenue went to some completely different group. I would say that by doing that they effectively killed GTK. I mean, GTK is still around, but it’s nowhere near as… its less than what it used to be. And the other thing is a number of paying customers halved pretty quickly, which is kind of what you expect.

So, as Nokia phones are no more. Although they are reselling under some other brand. And they basically sold everything off to Digia, which is a Finnish services house around about 2012. And they still had the commercial and the LGPLv2.1 license. And then the attitude was please, please pay us. Can you please? Yes, we need the money. And some did, some didn’t. But there was limited funding to be able to drive Qt forward and improve it and make it better.

Now this little story here: If you ever met an automotive purchasing manager, their job is to drive the cost down to zero. And I kid you not, that’s really true. We had one deal where we were trying to negotiate with an automotive company, and we were trying negotiate the money, and we gave into everything, and all their demands of what they said, and then they said, well you mean to say we can actually use this for free? And then they said, yes, so they’re using the open source version for free.

Now, the problem with that is that’s part of philosophy. If every company does that, then Qt as a software tool is just going to wither and die. And then it becomes of no use to anybody. Interestingly enough there’s one Tier One we worked with, and they said we’ve made a strategic decision that we’re going to standardize on this product, because we think it’s a great product, and we’re going to pay for it because we want to invest in it. We want it to keep going. We want it to keep improving and getting better. So it’s just interesting, different companies around the world have got different viewpoints on things.

And I think it’s also important to understand there’s a difference between the engineering managers who might believe what I just said about wanting to make that investment, and then the purchasing people who’ve got different motivations. They’re totally rewarded on how low a cost they can get for things. They don’t care about the future proofness of the company.

OK. So that was the Digia time. And those were the problems that we had, and we want to make the product better. We want to improve it. Want to invest in it. And one of the things we did then is we did a switch to LGPLv3, and that kind of ties into Tivo-ization discussions that we’ve had earlier, because as Daniel says there’s an aversion in the automotive industry to using LGPLv3, so we’re shamelessly running off that one, and it’s helping, for that respect.

So that’s pretty much what we’ve got. We’ve got this combination of the open source and the commercial. And there’s a couple of other things to that. We’ve also got additional commercial in the add ons, which is also another standard way of people making money from open source projects. But we’re also tied down by this thing called the KDE Free Qt foundation, and that comes back to Eben’s original story, so… One of the big benefits that Qt had right on the early days was that the KDE foundation, KDE wanted to use Qt as the UI framework for their Linux desktop, and they didn’t like the original license, and I think that’s when it came to GPL. And so there is an agreement between Qt and the KDE Foundation, which has actually survived from Trolltech to Digia too. What’s now been renamed as the Qt company, and they somewhat limit what licenses we can use, but basically we have to stick to this LGPL license for most of the core components of the Qt framework, and also we’re required to do regular releases, so if we don’t do any release within any, within a 12 month period, then they have the right to fork it and release it all under DSD.

So that’s sort of on the dirty money side that some people don’t like talking about. On the more positive side, on the governance is that we do treat this like many open source projects, so there was when Nokia created the LGPL V2.1, we were looking for lots of contributions. And we created what’s called the Qt Project, which still exists, and that defined a governance model for how we run. So that is documented on a Wiki site. Look’s pretty good. All the code that we have is all developed in a completely open repository. So we’re following more of the… If you’ve read The Cathedral and the Bazaar, we do more of the bazaar model where even everything that’s in the middle of being developed is out there in the open.

When you are committing codes, there’s a bunch of approvers for using Garrit, and you can find that list of approvers everywhere and there’s a mechanism for who becomes an approver and who isn’t. All the bug tracking systems are in the open, although there are some that are commercially sensitive, so we can hide those. List of the maintainers, and there’s a public developer’s mailing list. So all our communication that we have amongst our developers and all the other developers happens on this public mailing list. There is an internal developer mailing list, and there’s virtually no traffic on it. So it really is everything’s done in the open, so in that sense we are developing everything open. So you got free access to code and inspection, what’s going on.

So wrapping up here… I think the open source is a fantastic benefit to everyone. We’ve already talked about that today. The innovation that goes around it is pretty amazing. Everyone benefits from it. I think the point that everyone has to recognize though is that good software doesn’t come for free. People are working for it. They need to be paid. So, somehow, you’ve got to figure out some way of paying them. There are different business models for that.

You’ll find that big corporations give away their software, and they have different motivations around that. A lot of it is around, they want control as well. And if they can do that, that’s great. I think it’s important to recognize that some of the reason for giving it away is a control one too. And then the business models can vary depending on the type of software. Some build on services, some build on add-ons, some just give it away because they want to engage their developers.

And there are corporations there who just take the software and don’t contribute back. But, there are also other ones that do contribute. So, and then just to wrap up, so we’ve chosen to use this dual commercial strategy, and so far it’s working for us. So, that’s my talk. Thanks everyone!

MOGLEN: Questions?

If the consortium model isn’t the way to make software for common needs across the manufacturers, and if it’s going to be hard for small dual-licensing companies to avoid being driven to price zero by large purchasing manager oligopolies in the OEMs, how is this software going to be made?

ADAMS: Well I think it depends on the software, but I mean, I think for us so far, I think you’re right. Well, we gone with the LGPLv3, that’s stopping them driving it to zero. But you guys are now coming up with strategies where you can use it.

MOGLEN: You see! This is my question.

ADAMS: So this is where it gets tricky.

MOGLEN: Is it really that this software is going to be made by making people afraid of free software licenses? Is that how it’s going to work?

ADAMS: It’s not the ideal one.

MOGLEN: No, I don’t want to be your enemy. I mean, I want this software to be made, and I want it to be free, and I don’t want the one to be in conflict with the other. The advantage that comes from that uniform UI design is beneficial to everybody because when we get into car of type A, we don’t have to relearn how everything works on the screen. It all looks kind of uniform. And while it may be desirable to make it possible for people to be aware whether they’re driving a Volvo or a Ford, there’s also reasons why you wouldn’t want to make it impossible for people to drive one another’s cars, assuming people are going to drive cars.

Isn’t the standardization of ergonomics a business model in itself? There are all those design elements, all those pieces which are particular to the people making the toolkit. Why isn’t that a sufficient basis upon which to be part of the consortiums that makes things? Because we provide human interfaces, which everybody can understand.

ADAMS: I think we’ve already got two different user interfaces on phones right now, which are different. And I think there’s a benefit. If everyone – we had this discussion earlier– if everything is standardized, then you don’t get any innovation around it, and letting it… Well, first of all, I’ll say we, with our toolkit, we don’t define what user interfaces looks like. We give everyone the ability to define it however they want. So, right now I think it’s great that each OEM can go around and make their decisions on what they think the best user interface is. And I think we’re starting to see a little bit of innovation now especially now that some of them are actually building it in-house.

Let me just… I don’t know how many people know how that works but as I illustrated earlier, the OEM’s are typically spec houses. They write the specification of what the software looks like, should be like. They send it off in an RFQ and they get various bids come. They award it to a Tier 1. And then 6 to 9 months later, the software comes back, and then you start to put it in a car.

You take a look at it, and you say well that button is the wrong size and that color should be tweaked in shade. And then you say to the Tier 1, can you go change that? Then they come back and say, well, that’s a change request. You need some money for that. And so you have a negotiation, and you pay for it, and then you go back and forth like that, and after a while you say, well, we got to ship this car, and you’re going to ship it regardless of what the state of the software is.

And that to me is one of the reasons why the software in these cars is so bad, from the user point of view, and if you look at the way that you design user interface software, often it’s highly iterative. You sit down with your designers and software side by side, and you try something out, and you see how it works. And then if it doesn’t work, you go back and you iterate it quickly. But if you’ve got a nine month design cycle, you can’t iterate it.

So, we’re starting to see OEMs bringing this stuff in-house and doing some amazing stuff. And one of the systems that was shown at CS this year was, we had a lot of people coming around and saying, that’s the best system we’ve ever seen. I think things like that are changing, and certainly help that way.

MOGLEN: And then the business model for Qt is it helps you to be more rapid and adaptive in your design when you bring it in-house.

ADAMS: Yes.

MOGLEN: Oh, well…

ADAMS: There is that.

MOGLEN: Well, then why don’t we just GPL3 the whole thing and not worry about it? You’ll make a fortune. It will be fine.

ADAMS: Yes, well we have to adapt as things change.

MOGLEN: Indeed.

ADAMS: And everybody does.

MOGLEN: Anybody want in?

AUDIENCE MEMBER: Hi. Thanks for your talk. Can you talk a little bit about– as I understand it, you have, you lead the automotive sector inside Qt.

ADAMS: Yes.

AUDIENCE MEMBER: Yea, okay, so, I… Do you have any kind of visibility on the communities that form around, as I understand in reality, it’s more about dashboards?

ADAMS: Yes. it is, yes.

AUDIENCE MEMBER: OK. So do you have any visibility on the communities that form around hacking those dashboards for customization purposes? Do you support them? Do the automotive industry supports them as well? Is it mostly an underground thing? Does it exist?

ADAMS: I don’t… I’m not aware of any that are hacking that, to be honest. Maybe that’s my… If there are any, I’d be interested to know, and maybe I’m not looking in the right places, but I’m not aware of anybody hacking them, to be honest. I know a few people have hacked into the Tesla systems, but that’s more to see what’s in there and how did it work. I don’t think anyone has actually gone in and changed anything.

AUDIENCE MEMBER: Can you just provide some commentary on how liability laws apply to open source?

ADAMS: Well, that’s a very interesting one because when we go into a commercial agreement with automotive companies, they’ll ask for some liability. On the other hand, some companies have been very happy just putting our software in there, under the LGPL2.1, as is, so we’re not involved in that from a liability point of view at that point. So, and, so clearly car companies are quite happy doing that. They don’t worry about it.

AUDIENCE MEMBER: I find it, if I can contribute an idea to the conversation, I find it interesting that people seem to have a different view when we’re talking about the software from everything else that is in the car. I mean it’s not like people cannot change the mechanical parts of the car by themselves today. Right? And those same people are also driving the cars. There’s already laws that cover reckless behavior. So I think software– as software evolves and as we learn how to change it in a controlled way, I think my gut’s vague feeling is just that it’s not so different, right? If you change your car, if you change the software of your car in a reckless way, it’s similar to changing your car, the mechanics of a car, in a reckless way, as well. You can cause damage.

AUDIENCE MEMBER: So, if I understand correctly you are talking about the dashboard especially like the infotainment part, but how about the back-end of the driving algorithm? So, do you think there’s a possibility also to have an open source platform for that kind of driving algorithm to merge every car manufacturer or software vendors efforts?

ADAMS: Sorry, you meant for autonomous driving?

AUDIENCE MEMBER: Yes, autonomous driving.

MOGLEN: If you wait just a few minutes, we’ll bring you the people whose goal that is, and we can talk about that in our next session after these two. So, we will in fact be talking about it. More for Alistair? Well, thank you.

ADAMS: Okay, thank you.

Previous: 2-gpl3-panel | Next: 3b-mcguire | Contents