It's the ongoing maintenance, just like how calibrating cars to exact speeds is expensive, so too are radar and laser guns. So the burden is on the radar/laser guns because there are fewer of them. And the calibration cannot be performed on site, as the unit your testing against also has to be calibrated to a much higher degree and you're not going to throw it in the back of a truck and drive it around risking damage. And in the event of someone disputing, it's up to the state to produce proof of calibration rather than the individual proving it was not. It's an expensive process, as anyone who uses specialized tools (not limited to law enforcement) would tell you.

Now, it's possible that the calibrations aren't done, and people don't question them and get tickets. But you want the ones that don't constantly make mistakes. 20ish years ago DC put up cameras and it was sending tickets across the entire US because it wasn't reading plates correctly. They had to hire more people to cross reference any matches with the make and model of the car just to double check, and that was a lot of eyes that needed paying. Even though a lot of it is automated today, you're still paying a premium over putting a camera on a stick.

It's not about monthly costs. It's about up-front costs. These systems can be millions, maybe tens of millions to start, and are typically a huge, multi-year contract to a third party company that manages the system. Plus hiring human staff, maintenance and installation of a huge amout of it systems, training, tying in the car registration databases... On top of the fact that they're politically unpopular. Most people don't want to be tracked everywhere they go, even if the trade off is slower drivers on the road. There's a lot of headwinds before monthly costs are ever even a consideration.

Speed limits aren't an "everyone gets to be right/wrong" thing at all. It's numerous new things added as band-aid and one-item "fixes" to specific situations. It's also why the US tax code is such a mess. No one cares to overhaul it, and it's just a huge jumble of singular tweaks. It's like when you let a group of people take 10 years to write a story, and every week 5 people get to add or edit 1 sentence. Will you get a harmonized story with coherent plot? Likely not.

Radar guns don't report the speeding to anyone. A radar gun sitting on the side of a county road will write a ticket... how? Magic? It's a tool. A tape measure doesn't measure your height for you, you have to read the number and know the context. Before radar guns cops had to base speeding on a guess, or their own speedometer, and had to testify in court to a more subjective standard.

We don't control speeding because we mostly have passive systems that fine people for violations. And anyone determined enough will speed anyway, fines be damned. Many people speed as a flex and pay fines like a status symbol. You can't engineer 200km or 200 miles of highway to keep people at 65mph or below. Engineering solutions are limited to pedestrian areas or slow speed areas. The systems we use to fine people have the added societal costs in that they take taxes to set up, and are surveillance state tools.

In most places in the United States speeding is a criminal offense. Each element of the crime must be proven beyond a reasonable doubt. And the accused has the right to confront the witnesses against him in court.

I sort of understand the argument about cops needing to hold the radar gun, but I also don't really get what they're contributing to the legal validity of the case.

The biggest thing they contribute is proof of identity. When they pull you over and run your license, they can testify with pretty strong credibility that it was you driving the car.

To prove speeding you have to prove that the car was going too fast and that someone in particular was driving.

Having a cop signing the ticket also gives the government a convenient witness who can be subpoenaed to come to court and to testify with personal knowledge of the incident. That satisfies the confrontation clause. Whereas in jurisdictions that have tried pure camera enforcement, the only witness they can usually bring is a low level camera company representative who has no personal knowledge. This leads defendants to demand a lot of discovery into the automated processes of the camera system, because that is what is really accusing them.

Conclusion: US jurisdictions do a lot better with camera enforcement when they create civil infraction liability under state law. But then some forms of enforcement, like license points, may be taken off the table.