Innocent people taking a train are dead. Many are injured. The government had an answer in 2008: spend billions of dollars and wait for years. There's a better answer: Build a smartphone app, with some cloud software, a couple sensors and cameras, and engine cab remote-control harness. It would be faster, cheaper and more effective than the existing partly implemented "solution," and lives would be saved.
The Crash
Here's the story of the crash in a nutshell:
Eight people were killed, and 43 still hospitalized days later.
Reactions to the Crash
The basic reaction has been typical all-politics-all-the-time. Here's the Reuters story:
Later in the same story, you learn that the engineer was driving at more than twice the speed limit for that part of the track, and that the accident would not have happened except for his error. But that's a detail, I guess.
Technology Could Have Prevented the Crash!
Then it turns out, we know how to prevent things like this! But according to the experts, it just hadn't been installed.
This PTC ("positive train control") sounds like wonderful stuff. It turns out it's been around for awhile. Everyone seems to agree that it would go a long way to solving the problem of crashes like the Philadelphia one. So what's gone wrong?
Government-Mandated Positive Train Control
Here's a good summary of the issues and problems of the wondrous PTC solution, which was mandated by Congress in 2008. It was declared by Congress that it must be completed by the end of 2015. It won't be. And the cost? The GAO estimated somewhere between $6.7 billion and $22.5 billion.
A brand-new system dreamed up by government bureaucrats in a short period of time -- of course it takes billions of dollars and many years to implement! Of course it's a completely custom system, relying on railroad-only technology that will be generations behind the general computer industry before it's even deployed! Of course everyone assumes you can spec out a never-built-before system and get it right the first time!
This is amateur-hour technology, and it is ... killing! those of us unfortunate enough to be in the wrong place at the wrong time. This is a near-perfect example of bureaucratic "innovation." It is an example of the "what not how" problem of regulation: what should happen is simple declarations of goals (don't murder people) instead of gruesomely detailed directions for how to avoid murdering people. The bureaucratic approach mandated by Congress has already resulted in incredible expense and multiple avoidable deaths, just as its similar approach to computer security has resulted in some of the worst security breaches in history.
The Modern Approach
There is a better way. It leverages modern computing, devices, networks and software. "Experts" will pooh-pooh the approach, saying that anyone who proposes it doesn't understand the harsh and peculiar railroad environment. That's what experts always say in situations like this, standing on their little technology island, protecting their "expertise" and their jobs, until modern, high-volume technology gets the job done. Then, without further comment, they retire.
I won't lay out the whole approach in this post; this blog has lots of the core ideas, and so do lots of modern computing technology people.
Just as mapping software on a phone can track your location and speed when you're in a car, it can do it when you're on a train. Why shouldn't lots of people have this app? Why not publish the complete map of all the train tracks? Most of it already seems to be available to consumer mapping programs -- they just need to be tweaked to allow travel on rails instead of on roads. Yes, there are areas where track maintenance is taking place where trains shouldn't go -- just like with roads! Mapping software already exists to avoid such routes -- just use it! Yes, there are switches -- how about adding them to the maps, and making whatever controls them upload their state to the cloud? Yes, there are other trains to be avoided -- how about the apps all upload their positions to the cloud, and give a view to where other trains are? Yes, there are things you should pay attention to when you're not looking at the app -- navigation apps already handle this through audible alerts or talking to you.
These simple steps, which could be built iteratively and deployed in weekly cycles, would go a long way to solving the problem. There remains the problem of overriding the train controls in case something terrible happens -- but if all the conductors have the app and they have access to the engine car, many of the potential bad things could be avoided. The potentially tricky issue of automated speed control could then be addressed -- but after all, airplanes are largely run by auto-pilot, why shouldn't trains? If auto-pilot works for vehicles that go hundreds of miles per hour, miles in the air with no tracks, surely it can't be too hard to make a version for relatively slow vehicles without steering controls, whose only variable is speed!
While the government is mandating and regulating, billions of dollars are being wasted building systems that will be obsolete before they're installed, and meanwhile people are being killed and injured. There is a better, faster, cheaper way. Its cost to build is likely to be much less than the cost to simply maintain the PTS. So let's do it!
Comments