Intellectual Horsepower

Joe Doakes from Como Park emails:

The thing I like about Richard Fernandez is that he thinks about systems, about consequences, about the unseen effects. You never get that in the mainstream media or day-to-day political reporting.

If we consolidate everything into one system, we’ll all be equal and everything will be fair. But what if that single system crashes?

From the comments, a great analysis of why public transit is doomed, and always, always requires massive subsidies:

“Private Subways and buses were able to charge premium fare prices because the alternative to their service was the enormous cost of owning a horse — purchasing a horse, stabling it, feeding it, exercising it and watching it depreciate as it aged — all in the context of a crowded inner city. That the rise of the automobile industry began the process of making private subways and buses uneconomical. The affordable, convenient alternative to owning a horse shifted from the subways and buses to owning an automobile. The car was much cheaper. This will never change. Today’s mass-transit planners are permanently stuck in the 19th century — providing a service that fewer and fewer people want at increasingly high prices — and they have no idea why their projects are going bankrupt, because they have no idea why the great transit systems they are trying to recreate went bankrupt, or were even created in the first place.”

Yep.

Joe Doakes

23 thoughts on “Intellectual Horsepower

  1. The 21st century transport technology to watch for and invest in is automatically piloted vehicles. This technology will eliminate accidents, allow transport for all of the non-drivers, and replace general purpose cars owned by drivers with on-demand fleets of purpose built vehicles, electrically powered, light, fast, and efficient.

    In 30 years, driver-less vehicles will dominate urban transport, and intermodal passenger travel will probably use either magnetic levitation or ground effect flyers, not heavy railway cars, little different from their 19th century forbears. We’ll rip out those expensive high-speed railway lines for the scrap value of the steel. The only thing of any value will be the corridor itself.

  2. Drinking the liberal koolaid a little early in the morning Emery?

    “This technology will eliminate accidents, “

    I know from monitoring fully automated production lines that the best you can hope for is to stay within six sigma levels, but you will never be able to “eliminate” anomalous incidents.

  3. Joe Doakes is making a hugely important point.

    Look at the S.F. BART system. I LOVE riding it. Tourists can totally ditch their cars out their. But…

    Even though they have HUGE ridership they are going broke in every possible way you could measure it. Human resources, capital repairs, general repairs. WHY? Because it’s highly centralized and HAS to be managed POLITICALLY.

    When people use cars, the bus, and bikes this type of thing isn’t eliminated, but it’s much, much less of a problem.

    The question is why aren’t cars more effective and affordable.

    Mitch: you should get a BART expert on. I have one suggestion, too.

  4. “This technology will eliminate accidents” – spoken like a true engineer. Funny, whenever an engineer tells me something preposterous like this (and yes, I have been told this about machinery that could in fact kill the operator if the “infalible, fool proof” system failed) I have on occasion asked if I could get a commitment from the engineer to kill himself if its failure did kill an operator. Suddenly their certainty in systems becomes faith & belief and legalese.
    There are already plenty of ‘driverless’ vehicles operating in transit systems, factories and hospitals (even homes if you want to count Roomba’s) and there are a lot of attendent problems. In fact, in an accident last week, a ‘driverless’ rail car in San Francisco operating on an automated track killed two rail inspectors.
    Although I believe this part is mostly right “driver-less vehicles will dominate urban transport”, it will be difficult to program these vehicles to operate in densely populated areas. People walk in front of ‘driven’ busses and trains now. Can you imagine how often a driverless vehicle will stop and start due to a person or a squirrel or a leaf moving in front of the sensor (and the continuous shock loading and moment peak energy draw on the propulsion system of the vehicle) causing the vehicle to have to stop, reset and re-start again every few minutes?
    But hey, we can all be like Obama on this one. After driverless vehicles kill scores, if not hundreds of people we can stand back and say “no one is more outraged than me about how these systems have performed!”

  5. FWIW there are A LOT of smart people that think driverless cars will work. More on roads / highways and less on streets I think.

  6. The “Driver” is the person or entity in control of the vehicle. “Driver-less” doesn’t mean it has no Driver, it means none of the passengers are in control. So who is in control? An on-board computer? The central computer downtown? NSA?

    That’s the heart of Fernandez’ article. If the on-board computer fails, one car crashes. If the central computer fails, metro-wide gridlock. Which is worse?

    So which Driver system would a sane society favor – individuals or collective? And what does that say about our society that we don’t favor that?
    .

  7. BART is just the latest and most prominent example.

    Look at any public transportation system in the world and you’re going to find a populace living under the constant threat of a union cartel that has the power to freeze life in place unless its ever increasing demands are met. It’s a textbook example of extortion, and it’s everywhere put in place and defended by leftist power mongers.

    Emery, you had better watch your step, pal. You keep on with the POD talk and a certain failed pornographer named Ken Avador (nee Weiner) and his motley crew of lunatics is gonna start a stalker blog on your ass. How’d ya like a 300 lb. lesbian digging through your trash every morning? Weiner can make that happen.

  8. Lemme see, do I want to be on a road populated with “driverless” cars, or the insane lady putting on her makeup while eating her breakfast burrito passing me at 80?

    I suspect that automated cars will lessen the number of accidents, especially during rush hour, but they’ll never eliminate them. Controlled chaos and swarming will still exist and even the best systems will fail at times. But that chaos will be far less than what we’ve got these days on my morning commutes.

    And no, I don’t think that central control of the vehicles is practical. The vehicles will need to be pretty much autonomous for pure safety reasons. The best case situation might have the vehicles communicating their destinations to a central server to adjust traffic patterns, but no central system can be viewed as safe if its failure will cause all the vehicles to malfunction.

  9. OK, I enter things on Mapquest, my car takes me there….but I’ve still got to pick a parking spot, figure out how fast I want to go. Suffice it to say that reports of the demise of the steering wheel are highly exaggerated.

    And, speaking as a QE by trade, it’s worth noting that “six sigma limits” are actually calculated at 4.5 standard deviations, and in practice most companies are lucky to achieve 3 sigma unless the spec limits are ridiculously wide. Drive the move (pun intended) to driverless cars politically, you’ll be lucky to get there alive.

  10. There is growing concern and awareness of the risks and nutritional value of genetically-engineered foods known as Genectically Modified Organisms (GMOs). The same should be true of socially-engineered Government Modified Organizations.

  11. Michelle Malkin has a column out today detailing the huge success of another Gov’tMO, the electronic medical record. Apparently spending tens of billions of dollars has succeeded in destroying coherent moves to what should be a no -brainer.

  12. BB, letting the government get involved in highly technical projects that don’t directly involve intentionally killing people is a bad idea. Even if does involve intentionally killing people the projects fail more often than the succeed.

    Speaking from having worked from within the Federal system, the only thing they can direct is failure in an IT system. I’ve mentioned NASA and the fact that being an engineer there really means that you’re a program manager at best and that drives folks who actually have a technical clue out of the agency. If you ever have to work on a federal project, be a contractor. They’re the only ones who can get things done in finite time.

    Honestly, I’m surprised the Ocare website is working as well as it is considering the intense politics behind it and the nutty federal regulations. The next administration’s Inspector General will no doubt find tons of regulation violations in the procurement for this disaster that made it only a Katrina-esque debacle rather than an IT Chernobyl.

  13. Holy cow. The website has 500M lines of code. By comparison, Windows 8 is about 80 MLoC and Curiosity went to Mars on about 0.5 MLoC. And the NYT is claiming that at least 1 MLoC need to be fixed for the site to operate “properly”.

    The industry average for fully debugged, deployed systems is around 0.5 defects/KLoC. Meaning we can’t expect the website to have fewer than 1e6 code bugs to track down even when they get through the obvious bugs that are screwing the site right now.

    Grossly oversized (5-20x what an entire banking system takes), constant change orders, and architected by idiots? Yeah, typical government IT. I think I may want to reclassify this as an IT Fukushima #1 with strong potential to becoming an IT Chernobyl.

  14. Nerd,

    Being in the Twin Cities IT racket as I am, I’ve had a contact or two with some visibility into the “MNCare” project.

    And, er, what you said.

    Issue of personal professional interest: according to my contacts, the project seems to have retained no “User Experience” people. The entire design of the user’s interaction – the way everyone deals with the system, from 20-somethings to 80 year old luddites – has apparently been left to programmers and business analysts. Which means you’re going to get a “UX” that makes perfect sense – to programmers and business analysts.

    It’s an observation, not a conclusion. But I’ve heard nothing from anyone in the racket to correct the observation yet.

  15. It’s worth noting that Windows 8 is not only a lot smaller, but there are significant portions of Windows that do not need to work–e.g. “Solitaire”–for the system as a whole to work. No such luck with this system.

    Regarding government engineers; point well taken. There are some good ones–I’ve done the source and first article inspections for the DOD, NASA, and their contractors in a former job, so I know there are some–but there are too many who simply can do a checklist from a 40 year old document. Sigh. It’s worth noting, though, that had Obama simply called up some quality specialists from the DOD or NASA, they would have been able to give him the tools–FMEA and the like–that he needed to get this right. That’s the level of the guy’s incompetence–it just boggles the mind, really.

  16. Another couple of thoughts; even without user experience people, nobody saw that 500 millon lines of code was going to cause problems? say what? Seriously?

    Also, 500 million lines of code also means ample room for all kinds of disaasters with the data taken. We’re not just talking about a Chernobyl for getting people signed up, but suffice it to say that those who have would do well to pull their credit reports routinely, if you catch my drift.

  17. @Seflores
    A lawyer might be able to make a lot of your arguments, but you are fundamentally wrong. If one faults the car designer, the road designer, and or the legislators, what they’re being accused of is not designing their element of the driving system with sufficient attention to the many failings and weaknesses of the vehicle’s pilot. That case can and indeed is made in courts of law. But arguing that designers should have made more allowance for the limitations of human drivers is very different than saying drivers aren’t at fault. The fact remains that vehicles crashing due to mechanical failure causing loss of control, or blind spots and high speed limits which do not allow room for a vehicle to stop, are exceedingly rare. People don’t notice hazards, they don’t react appropriately to hazards, they drive too fast. Relative to things like failure of brakes or failure of a steering mechanism, drivers cause orders of magnitude more accidents than the very reliable machines they are driving. Judged by the quality standards of the vehicles they are in, the driver is the only unreliable element in the entire system.

    Cars will soon have lane and spacing maintenance as a standard feature. Navigation systems that take control on city streets will be the next luxury feature within 5 years. They’ll be common in 10. The first adopters will be long-haul trucking companies. I read of a study showing that if even 10% of vehicles have sensors and automation to maintain correct spacing with the vehicle in front (adaptive cruise control, a fairly common luxury feature these days), the entire rush hour traffic flow is significantly faster which generates higher capacity. The idiotic drivers’ actions seem to be influenced, positively, by the calm and sensible computer driving beside them.

    I welcome any and all automation of the driving process, not because the automation will be perfect, but because it cannot fail to be better than the majority of drivers currently on the road.

  18. “…but you are fundamentally wrong.”
    What tipped you off? When I wrote this: “Although I believe this part is mostly right” then cut and pasted your comment “driver-less vehicles will dominate urban transport”?
    A concise summary of your 6:27 comment might be: (effect Scooby Doo, old man criminal voice) “Damn humans. Always meddlin’ with my perfectly engineered systems & processes.”
    Funny – Here I thought I was agreeing with you that automated vehicles are the future – it’s just that all automation has a human element that you can do your best to design for but can’t completely eliminate as a factor. People used to do all kinds of crazy things to bypass the seat belt warning indicators for God sake. When I was young, I thought ‘The Man’ (you may know him Emery, hepcat that you are, as ‘Mr Charley’) was forcing my fellow ditch diggers (yep – it’s how I put myself through college) to do dangerous work without the proper protection due to his greed for the almighty dollar. Then a guy on another crew bypassed the safety on his dump beds hydraulic lift because he didn’t like how slowly it operated. A dump bed loaded with a ton of debris came down on him while he was trying to clear a jam and he was cut in half. That place had many safety problems and 99% of them were ‘workers’ bypassing the safety systems both automated and those established by code and common sense for no other reason than they didn’t want to be told how to do their job. Didn’t stop their survivors from sueing though.
    “A lawyer might be able to make a lot of your arguments…”. Yeah – they kind of have. And don’t kid yourself Emery, the thought of liability and risk management has killed many ideas in the crib. I kind of make my living in the world of automation and potentially dangerous machinery if you can’t tell.
    “The idiotic drivers’ actions seem to be influenced, positively, by the calm and sensible computer driving beside them” better put as: I for one welcome our new robot overlords. And you think Republicans have contempt for their fellow man.
    “…the majority of drivers” – if this were true wouldn’t our accident rates be much, much higher? Honestly Emery, too often you come to comment here with such a chip on your shoulder that you don’t read and process what others have written in support of your comment.
    Final question: Why you mad bro?

  19. @selflores: I didn’t mean to come off as sounding ‘that’ critical.
    Your comment is not without merit and you appear to have a sound grasp of the issues in automated systems. Re: /..‘driverless’ rail car in San Francisco operating on an automated track killed two rail inspectors…/ This reminded me of the train accident in Spain this past July. http://www.cnn.com/2013/07/25/world/europe/spain-train-crash/#

    Automation is a tricky thing. Computers are much better than humans at keeping something where it is supposed to be, like the speed at a speed limit or a vehicle in a lane. They have an infinite attention span, and always try to follow the rules, two things people are not good at. On the other hand, computers are limited by their inability to recognize that something unusual is happening, like an instrument that gives a false signal. They can only react to things that the programmer has anticipated and spent the time to write a response for. And they can also be hyper-sensitive, for instance shutting down a system if a signal jumps into a danger zone for a fraction of a second because of some (normal) noise in a measurement system. The best automation systems almost always pair automation with a well-trained operator, but hitting the right mix of caution and freedom to operate is always difficult.

    I design chemical plants and design the control systems as part of that. The safest plant is neither the one with no operator input or no automation. Hitting the best balance for optimum safety (while still running) is always a challenge. It’s never as simple as kicking the operator out of the control room. Yes, trains are simpler, but there’s something to be said for having a driver watching the track ahead and scanning the instruments to detect little faults before they become big faults. That being said, constraining his ability to adjust the speed would have been a good move here.

    In the world of dangerous chemicals, we do lengthy studies called Hazard and Operability Studies or HAZOPs which examine what bad things could happen if temperatures, pressures, flows or levels get too high or low, and then ask whether we have sufficient safeguards given the risks. Giving the operator the ability to run the train too fast around the curve, risking derailment and dozens of deaths, is a big risk which, given the single human point of failure, has a fairly high likelihood. That would never be an acceptable design in a chemical plant. But private companies have higher standards (we’re required to by regulation). With public utilities, the operator and the regulator often report to the same boss, who is a politician who only answers to an easily distracted public once every few years

  20. Emery, I think that the distinction here is that there is a HUGE difference between poka-yoke systems at a chemical plant to error proof things and having 150 million collision avoidance systems in a scenario where you’ve got rain, smoke, and debris throwing off the radar/laser sensors–and a police union that will get understandably uneasy at that much laser/radar energy bouncing around and possibly interfering with their speed traps.

    My guess is that you’d have to get the system down to $1000 or less before it passes a cost benefit analysis (cost of life lost vs. cost of systems), so we’ll expect the next Democratic President to insist on it when it costs a mere $10k apiece.

  21. It was my understanding that the streetcars never did pay for themselves out of ticket revenue, or even pay their own operating expenses. The streetcar operators made their money by speculating in land along their right-of-ways.

    When we made insider trading illegal, they all went bust.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.