The Top Seven Questions Elon Musk Should Have Been Asked

The missed opportunity of the Rogan and Munro interviews

Image credit: Munro Live

Recently, Elon Musk conducted a pair of interviews with Sandy Munro¹ and Joe Rogan². In each, he was asked some softball questions, and he provided some entertaining and enticing answers about future products. But there are a number of important questions Musk should have been asked on a variety of fronts. I ask those questions below.

“What is Tesla doing to support user rights and privacy?”

Tesla vehicles collect an enormous amount of data — perhaps more than any other manufacturer. The data available to Tesla runs the gamut: video and audio of user surroundings; same for the interior and its occupants; GPS location of the vehicle; driver telemetry, which can infer degree of driver ability and “misbehavior”; and UI/UX interactions. This data is extremely valuable and can be used for unlimited nefarious ends. Here are just a few scenarios:

• A driver having to operate the vehicle in a slow, indecisive manner in order to avoid insurance rate increases.
• A Tesla employee surreptitiously offering a specific vehicle’s location data for sale.
• A Tesla vehicle picking up audio from the cabin to be used in corporate espionage.

Further, the software-based nature of these vehicles means they can be easily manipulated by a malicious actor. And I’m not just talking about hackers — even if the vehicles are secure from an opsec perspective, Tesla itself or an entity with power over Tesla could force the vehicles to do things not in the user’s interest. Some scenarios include:

• A citizen of China having their vehicle deactivated because their government social rating is too low.
• Tesla deactivating vehicles because a future evil CEO uses safety as a pretext to force consumers into making purchases.
• The singularity deactivating all vehicles to prevent resistance to the robot revolution.

The above possibilities range from plausible to outlandish, but the point is that with so much data flowing, and so much control over the vehicles, the potential for misuse is unlimited. User rights and privacy is something that gets built in at every stage of design, which means that user protections should be built in at down to the hardware. This means hard cutoffs of GPS and networking, as well as user physical overrides to prevent remote deactivation.

We shouldn’t have to just trust that Tesla will always do the right thing.

“How will Tesla resolve automation and fun?”

Musk has directed a number of bold moves with regard to design lately. In the past we’ve seen dashboards and physical buttons removed from interiors. Upcoming models are going even farther, with gear shift and turn signal stalks removed. Even more controversially is the switch from a steering wheel to a yoke, which has gotten overwhelmingly negative comments online and has even been shown in a test video³ to be a jarring experience.

The changes are reminiscent of Steve Jobs removing buttons and ports from laptops. Some of these moves were prescient, but others ended up being backtracked. Granted, people don’t know what they want until they try something new. No one asked for an iPhone, because no one had the vision to understand how groundbreaking the technology would be. Someone had to make it, and only then would people understand. Innovation passes from unimaginable to novel, then from common to required.

Likewise, one could argue the same for Tesla’s design changes, which are oriented toward taking the driver out of the equation. Musk even said as much in the Rogan podcast:

“All input is error. Unless it’s a game.” ⁴

But driving fun has also been a mainstay of Tesla. The ballistic acceleration times speak for themselves, but only slightly less obvious is the excellent handling from the cars’ low center of mass, as well as the BMW-like steering feel. Musk even said at the unveiling of the Model 3, “At Tesla we don’t make slow cars.” Musk, who is a manual transmission enthusiast, understands the fun of driving and interacting with a machine. That is to say, for many, driving is a game. But Tesla seems to be diminishing the experience and pleasure of driving in some ways at the same time they are enhancing it in others.

Ultimately, there’s an inherent contradiction to dissuading people from engaging in a fun activity, and I want to know how Tesla will resolve this.

“What will Tesla do to better support the DIY community?”

There’s a lot of enthusiasm among Tesla owners to maintain, upgrade, and repair their vehicles. Many of the early adopters that have kept the company going are tech-savvy tinkerers and makers of various stripes. This customer base would love to modify and repair their cars, but Tesla for whatever reason makes this overly difficult.

Some of the issues are less pressing. For example, it would be great if Tesla provided the equivalent of crate engines to DIYers. That is, raw battery packs and controller units for the purpose of doing ICE-to-EV conversions. Currently we see a lot of salvaged batteries ending up in these conversions, and while the aims are admirable, one always has to wonder at the condition of these parts. Another suggestion might be to open up ports and power in Tesla vehicles to allow for modding⁵. Plenty of people would prefer to have buttons, instruments, and heads-up displays. Why not smooth the path to let developers and manufacturers interface with non-critical systems to make their cars fully their own?

More pressing though are the travails of DIYers that try to repair their own vehicles. One look at the travails of Rich Benoit⁶ is enough to dissuade any but the most determined from attempting to work on a Tesla. This is a company that will even go so far as to blacklist VINs that are considered salvage, in effect offering parts only for cars specifically known to be working. How can this be legal in states with right-to-repair laws? At the least, this works against Tesla’s mission of converting the world to EVs by keeping repairable cars off the road and keeping total cost of ownership artificially high. Democratizing access to EVs — and their components! — is a crucial step in effecting the transition away from fossil fuels.

“What is Tesla doing to support repair shops?”

This is a similar issue to what DIYers are facing, but instead of consumers being prevented from making the most of their vehicles, instead we have small independent repair shops facing an insurmountable barrier to entry.

Tesla may think this works in their interest by keeping repairs in-house, capturing the revenue from the work while preventing bad press from shoddy repairs. But this also works against Tesla by making repairs take a very long time. The word is out that an issue with a Tesla may result in several months without a car, meaning that for consumers without a car to spare, the risk of being without a car must be factored into the purchasing decision. This seems like a real self-own when one considers there are many professionals eager to get into the space and take the load off of Tesla’s repair centers — all Tesla really needs to do is give access to adequate training and resources, such as diagnostic software and repair guides. You know, stuff that already exists for training their own technicians.

This issue becomes even more important as older electronics and batteries wear out and need replacement. Doing this safely should be a viable option without having to reverse engineer everything, as we are seeing with certain high-end repair shops⁷. Currently Teslas enjoy chart-topping depreciation, making it easier to justify their high price by citing their high resale value. But if these cars cannot be repaired cheaply, they will begin to move from the realm of Toyota down into comparisons with marques like BMW and Audi, which after a dozen or so years lose the bulk of their value when repairs become necessary but uneconomical. Leases will effectively become more expensive and the Tesla brand will be tainted.

“How are you going to handle China’s dark side?”

There are several issues with China, which is ruled by the world’s most powerful autocracy. The first issue regards intellectual property. Does Musk really think Tesla’s IP will be safe there? It’s not hard to imagine that one day when Full Self Driving is finally working suitably, the software and training parameters could simply be stolen and used in Chinese vehicles. The same is true for Tesla’s battery technology. Advances in each are hard-fought, but once developed can merely be copied. What will he do then?

Another issue is, what happens when Tesla is pressured by the government to do unethical things? The unthinkable in the democratic world is the future in authoritarian countries, and it’s not hard to come up with examples. I mentioned above the idea of shutting down vehicles operated by people with insufficient government social rating. Here are some more possibilities:

• AI to listen in cars and spot unharmonious speech.
• Provide lists of people whose driving locations suggest involvement with pro-democracy groups.
• Trapping dissenters in Teslas and delivering them to the police.

What will Musk do when China demands these features? Will he refuse and pull out of China? Will he dare to complain on Twitter? Will he find a way to justify compliance? Or does he not see a problem with any of this?

This leads me to a bigger question, and it goes right to the heart of Musk’s worldview. He has remarked that he admires how things get done in China. This of course happens because the people have no say in how their government operates, and they have no recourse to resist powerful factions when they are exploited. That is, things get done because those who suffer cannot resist. So let me ask directly: Elon Musk, do you think authoritarian government is preferable to democratic government? Given that Musk is the world’s richest man, who holds the key to both Earth’s environmental salvation and our escape to the heavens, as much as any one person ever has, the answer to this question may determine whether mankind’s fate over the next several generations is one of joy or horror.

“Why isn’t Tesla being more proactive in addressing safety-critical defects?”

Musk makes a big deal out of how Teslas are the safest cars in the industry. I can’t deny, this is an earned boast, and not an easy one to accomplish. The vehicles have many innovative physical and software-based features which are forcing competitors to respond. But Tesla has had a few missteps as well, which leads me to wonder exactly where the line is on their commitment to safety.

One issue regards the use of cast aluminum suspension components that were known to fail⁸ in the Model S. It’s appealing to use cast aluminum here because of the reduction in weight for a large unsprung part, paired with the cost advantage over milled billet. But car companies don’t go this route because the casting methods often lead to trapped gas bubbles which weaken the part. Indeed, the early Model S would occasionally suffer sudden, catastrophic failures of this part, which would lead to loss of control. But Tesla never moved to ameliorate this issue for customers who might have it.

At the time, Tesla was constantly on the verge of going broke, so it’s easy to understand why Musk would not move to recall their most popular model. But more recently, Tesla profits have been solid while its market cap has soared. It’s around this time that Tesla was forced by NHTSA to perform a recall over a safety issue with a predicted 100% failure rate: the failure of the touchscreen interface⁹. This affects 158,000 vehicles, and at a cost of $2,000 per car to repair (my own wild, high-end guess), this would cost Tesla north of $300 million. Quite a chunk of change! Except of course that at a market cap currently over $700 billion, making all their early adopters whole would cost less than an intraday market cap fluctuation. So why didn’t Tesla choose to do right by them?

“How much safer is Autopilot, really?”

Tesla has declared Autopilot to already be safer than human drivers¹⁰, but this claim comes with a number of caveats that go completely undiscussed. Broadly, there are two main issues.

The first is that Autopilot is only usable under certain favorable conditions. It’s hard to discuss this precisely as the capabilities of the feature keep expanding, but at various times Autopilot has only been available on highways, for use in suitable road conditions with clear lane markings, and in fine weather. To be fair, Autopilot keeps getting better at handling tougher conditions, but this only leaves increasingly tougher conditions to human drivers. As a result, Autopilot-driven miles are easier, and probably less accident-prone, than miles driven by humans. For a one-to-one comparison of human-to-computer driving, the relevant statistic cannot be merely the accident rate of humans vs. computers. Instead, the human accident rate used for comparison must be solely for miles where Autopilot could have been engaged.

The second issue is that Autopilot hands over control to a human when it comes upon a situation that it cannot handle, such as an impending accident. Granted, Tesla gives a five-second window in their Autopilot safety statistic — that is, if an accident happens within five seconds after Autopilot self-disengagement (as I understand it), it counts against the computer. But this five-second timeframe implies that that a human is expected to swoop in during a critical moment to save the vehicle within just five seconds.

Not only does this seem unreasonably short to me, but the fact of the matter may be that transitioning from computer to human in a critical moment may be less safe than if the human had just been driving the whole time, and this needs to be captured in the comparison. Moreover, an accident that could not have been avoided by either the human or computer, but which occurs after the five-second window, will get attributed to the human.

For all these reasons, I believe that the Autopilot statistic should more clearly indicate its safety by including accidents that occur significantly more than five seconds beyond disengagement, while the human-driven miles used for comparison should be ones that were only in Autopilot-capable conditions.

Conclusion

It’s great to see Musk delivering tidbits about new products with Joe Rogan and talking design shop with Sandy Munro, but in four hours of interviews it would have been better to see some tough questions asked. Hopefully the above will give some ideas to the next person lucky enough to get an hour of Musk’s time on video.

Notes

1. Munro Live: Elon Musk Interview: 1-on-1 with Sandy Munro
2. The Joe Rogan Experience #1609
3. The Tech of Tech: Barbed wire test… Is the Tesla yoke better or worse for driving?
4. Rogan #1609 68:54
5. Cybertruck Truck Guy: Top 15 Tesla Cybertruck Features and Accessories We Want
6. Rich Rebuilds: How Tesla rewarded me for telling the truth
7. As Teslas Age | Gruber Motors
8. Discussed in Edward Niedermeyer’s Ludicrous: The Unvarnished Story of Tesla Motors
9. Ars Technica: Tesla’s touchscreen trouble: NHTSA asks for recall of Models S and X
10. Tesla: Tesla Vehicle Safety Report