February 14, 2024

The USA is moving towards considerable tariff regimes on Chinese parts and using this to block import of complex products. Like cars.

The Biden administration is considering restrictions on imports of Chinese “smart cars” and related components that would go beyond tariffs to address growing US concerns about data security, according to people familiar with the matter. (FT)

The issue here is supposed to be about blocking the Chinese state collecting critical information on USA infrastructure using data collected by these cars' sensors. It might not be such a ridiculous notion, but banning certain "kinds" of sensors seems like a strange way to do this.

The way that the regulators explain this to the public is by saying they should be scared of the Chinese spying on them directly. So, continuing and leveraging the anti-China sentiment in preparation for the war the state thinks is "inevitable".

The question we might want to ask is what exactly does the data look like that the USA state is collecting through corporate proxies?

The move to electric vehicles is speeding up the transition to built-in surveillance systems. Especially for workers. Cameras and sensors phoning home and uploading all sorts of information on driver habits, collecting information for AI systems in the move to self-driving, and the use of data to direct discipline toward workers they deem disposable for minor infractions.

There is a feedback system between consumer data surveillance and workplace surveillance tech. So much so that testing in each area informs in the implementation in the other.

While the self-driving car is something of a sci-fi dream right now, there is an expansion of self-driving technology "inside the gate". The AI systems are being partially trained with consumer-focused technology and by drivers subjected to bolted-on tech "upgrades" sold to the companies on the promise of more control over workers and reduced insurance rates.

The point here is that it is not that the chip is made in China and might be used to collect information on critical infrastructure or the less likely hacking of your car. It is that there is a chip uploading any of this data on an ongoing basis anywhere.

This is part of the regulation of AI. It is the collection of data and the regulation of that process, storage, selling, and use of that data which we should be focused.

Making transportation safer and easier is a laudable goal. So is ensuring that production of inputs to transport infrastructure and products remains endemic. However, how we talk about that regulation can either seek real safety or can be about jingoism. Here we see pure jingoism and it should not be confused with a pro-worker position.

Industry groups in the USA and Canada who have been pushing against collection and dissemination of data for decades are now suggesting that the regulators do not have enough data to craft regulations effectively.

In the trucking and rail sectors we have the unregulated introduction of accident avoidance technology. These include AI driver assist tech in trucking and automated AI-supported "Enhanced Train Control" systems that can automatically take control over trains to slow them down in certain circumstances.

There is no question that the introduction of certain technologies can help with safety in transport. However, when left on their own, transport companies are less interested in the safety generally than they are with specific aspects of safety that increase their profit rates. Employer groups consistently oppose regulated systemic solutions with known technology such as wayside detectors and electric breaking systems. Instead, these companies push for the introduction of "safety" technology that seeks to replace workers such as automated inspection portals in an unregulated way.

The argument against the regulation of this part of the industry is the same as around AI: not enough data and the speed of change in the digital sphere.

Part of this is not untrue, of course. Regulators have been underfunded for a long time and their capacity to both understand and even enforce current regulations is severely diminished. However, the answer here is not to continue "self-" regulation, it is to expand capacity.

One easy way to start in this direction is the collection of data. The response from left-wing politicians to any comment about "not enough data" should be the demand for the release of corporate data to regulators, and then the release of analysis of that data to the public.

Some of this is already happening with the NRC, Transport Canada, and NRCan releasing more of their studies to the public through the new federal government websites (see above). But, it is haphazard and not centralized. Each government department, using the same basic technology for web-facing data, are using that technology differently to release data.

We need this data for trucking and rail industry so that we can make determinations of current safety technology implementation, but also so we can better plan and track the shift to greener technology.

New data on some derailments points to wheel fatigue as a major issue. There an argument here to be on the committees looking at this or at least paying attention to the outcome of this work. If derailments are a result of wheel fatigue, then sensors are not the solution to this, maintenance work is.

Analysis of Broken Rail Derailments

Figure 3 shows a useful example of seasonal broken rail derailments from a 2013 paper [10] that used 10 years of broken rail derailments and 2 years of service failures. The authors revealed that:

• Unsurprisingly there are considerably more broken rails (service failures) and broken rail derailments in the winter season compared with all the other seasons;

• However, the ratio of derailments to broken rails is eight times higher in the summer compared with the winter.

The second finding was cause for a separate discussion and the creation of an ICRI working group called Broken Rails Modeling (discussed in Section 3.8.4). Some members cautioned the reliability of the results, since the numbers of derailments used in the analysis was so few. For example, the number of broken rail derailments in the summer months might be as low as one per month over the entire class 1 industry in North America. Also, it was noted that, in the interests of time, derailments are often attributed AST-2023-0003 Version: 1.0 National Research Council Canada Page 25 to the easiest to find culprit, such as a broken rail, even if the root cause might have been a car, wheel or track issue.