Looking to the future

Charles Stross recently posted a transcript of a talk he gave to an engineering consulting group. You can find the rather long transcript here. It is a very good read, and he brought up a number of points.

One of the points was that 'driverless cars' are a near inevitability. This is a view I also hold, and I also think I may live to see it. My first talk about this topic was the second post ever on this blog, "PRT Will Save The World", which was a critique of advocates of the same idea trying to bring it off right now. Stross makes this point:

Once autonomous vehicle technology becomes sufficiently reliable, it's fairly likely that human drivers will be forbidden, except under very limited conditions. After all, human drivers are the cause of about 90% of traffic accidents: recent research shows that in about 80% of vehicle collisions the driver was distracted in the 3 seconds leading up to the incident. There's an inescapable logic to taking the most common point of failure out of the control loop — my freedom to drive should not come at the risk of life and limb to other road users, after all.

I agree with this point. Humans are the most unreliable factor in automobiles, so once automation has reached the level that it provides human-equivalent safety the drive to yield all control to automation will be strong. There is a fairly long road before we get there, though, as the first iterations of an auto-driver has to contend with not only road conditions but human drivers in the environment. It is much easier to plan a system based on the assumption that all vehicles are the same and all vehicle controllers are the same and probably networked; this is what Personal Rapid Transit advocates have been pushing for.

PRT can only be done right now if a separate system of roads is created along side what we already have. The build out for that will quite reasonably take 20 to 50 years before a reasonably complete system can be built in a metro area (never mind interstate traffic). In that time automation will continue. By the time the separate PRT road network is built out, it is highly likely that automation has gotten to the point that the existing road network can be used. In other words, we're about 20 years too late for PRT to pay off in any meaningful way.

The problem represented by "I get in my car and go to point B" has its own factors. Right now each driver is a selfish driver operating under a set of rules, meaning that they do not actively cooperate with other drivers yet still maintain (reasonable) compliance with driving laws. Once in a while they may deign to yield to a turn signal for a lane change, but that's about as egalitarian as it gets on the roads. Also the observation window for that driver is limited to what they can directly perceive as a driver, plus whatever news comes over the radio about traffic problems. More modern in-car nav devices have wireless connections to local traffic centers to provide near-term access to congestion data. This supplemental data stream is by necessity rather coarse grained.

In a future where human drivers are very few and far between with automation handling the details of getting from point A to point B, the problem changes markedly. The driver in this case is the in car automation. Because we're talking about 20-50 years in the future, I'm going to presume a wireless network between the cars themselves and to a central traffic authority. With GPS, which most new cars already have thanks to OnStar and similar services, the central traffic authority can gain very highly detailed view of what traffic is moving across the roads. The peer to peer network in the driver's environment can communicate with other vehicles on the road much more richly than can be done now through signals.

All of this means is that the driver in this case is cooperative not selfish. The peer to peer network can be used to notify cars behind the vehicle of sudden traffic hazards, lane change intent, emergency vehicle presence, traction hazards on the roadway, and other such data. The central traffic authority can route traffic around blocking points in a way that greatly enhances the total throughput of the roads. The network possibilities make the driver much more aware of the total environment it is driving through, which better enables correct decisions.

Then you have the fact that an automation system can have more precise sensors than a human driver, such as truly accurate range finders, and reaction time. If traffic suddenly needs to slow down to handle a deer on the road, the signal to "break" can back propagate from the deer MUCH faster than can be done with mere breaklights and right foot moves. If the car ahead of the driver sends the break signal, the driver can start breaking itself in less than a seconds, which is much faster than a human could do it.

The presence of a selfish driver, such as an actual human behind the wheel, can throw a smooth flowing cooperative system out the window. This is why after a certain point humans will be discouraged from driving, probably by insurance companies at first and then by law. As Stross says, "They're going to redefine our whole concept of personal autonomy". The social impact of this is going to be great, especially for those people earning their first driver's license this year, people who are most likely to see this system come about.

It is possible, though not too likely except in the 50-70 year range by my guess, that the concept of the neighborhood vehicle pool may exist and private vehicle ownership may fall off. We've got too much invested culturally to give up private ownership of a stable of vehicles, so if this happens it'll be a ways down the road. And if it happens, socialist Europe will probably be where it takes off rather than capitalist America.

On the other hand, such a system will better enable rental companies to rent things like pickups for weekend projects. $50 for a one day rental certainly beats out the $350/month loan payment to have that vehicle on semi-permanent standby. An auto-drive system would allow the rental companies to dispatch vehicles to residences without having to send a driver along, like Enterprise does now in certain areas. Once the system is more fully deployed, private ownership of vehicles may well decline.

This level of tracking of movement may well run smack into certain knee jerk privacy concerns, which is something else Stross addressed. In fact, that very topic is the main point he is trying to make about the world in 20/30/50 years hence. Within 50 years it will be possible for people to record every waking moment of their lives, and it is entirely possible that such activity be mandated for certain classes of citizens such as sex offenders and those under probation. The 'lifelogs' as he calls them will be searchable the way internet information is searchable, leading to levels of exposure that I find mind-boggling, but kids those days will be all too used to. So having everywhere you drive tracked will be nothing because every step you take is tracked in some way already.

It is an oft quoted statistic that there are more closed-circuit TV cameras in England than there are people, or will be within 10 years. This is probably exaggeration, but the point is that there are a lot of CCTV cameras in England. Couple those with sophisticated image processing, we're not quite there yet but we're real close, and you can get a tracking system of fair quality. This is something that can be in place within 10 years.

How we as a society will cope with this level of tracking is something we can't predict from here. That level of tracking will prove beyond all possible doubt that each and every one of us is a criminal in some minor way. Do you cross the street only with the light and at corners? Do you follow the speed limit ALL the time? Do you trespass by cutting across a lawn on your walk to the bus? Do you always move your car for snow emergencies? Do you always respect 20 minute loading zones? These questions will force some changes in society, hopefully the reduction of some of these laws. Won't know until we get there.

In defense of Gifted & Talented

I was a Gifted & Talented student for a number of years. I think I started in the program in 3rd grade, and continued through at least part of Jr. High. Because of that I have a first and view of what G&T programs provide.

Gifted & Talented is in essence special ed for smart kids. That might seem a wee bit dichotomous, but it is pretty close to the truth. Gifted and Talented programs are designed to keep said 'smart kids' engaged in education. Engagement is a particular hazard of Gifted and Talented youth.

I knew many kids while I was in school who were very smart, but rarely tested above the 70% mark. At the same time I was worried if I had a test come back with less than 90%. I was engaged in the process, they were not. To the gifted or talented youth, school can very easily seem intolerably boring so they detach from it. Native smarts are not enough to pass tests, you have to care enough to at least look at the material once to have a chance of doing well.

Engagement is the prime challenge for gifted and talented youth. These are kids who are smart enough that they can be told once, maybe twice, before they fully grasp a concept (assuming engagement, of course), where their classmates may be struggling with the 'tell me three times' method and still coming out with only partial comprehension. The continual back and retread required to bring the slowest students up to speed acts as a disincentive to keep paying attention to the class.

The method employed by my Gifted and Talented program was to have a special class once a day or a week where the G&T kids got together and did activities just for us. The hope was that by providing something to look forward to by going to school, that enthusiasm would rub off on the REST of school. That worked for me. It didn't work for some other kids I knew, but you can't help EVERYONE.

Another thing that helped me was attending an Open school. By having grades 4-6 in the same educational environment, 4th graders reading at a 6th grade level have a full class where in traditional education it'd be the three people in the class of 25 who read at that level. The traditional education model, three kids out of 25 reading at grade+2, isolates the 'smart kids'; and one thing that anyone who lived through largish classes at that age can tell you, standing out is a BAD thing socially. The Open school model also allowed students who are reading at grade+2 to also have classes with students with math skills at the same grade level.

Which brings up another point, in traditional educational settings gifted and talented students all too often end up with the "teacher's pet" nickname. This also provides a disincentive to doing well, as it is yet another way kids isolate 'outsiders'. I've seen three ways of coping with this problem:

1. Hole up. Don't answer questions, and try to keep people from find out your test scores. Sometimes, this means purposely blowing parts of tests to keep from 'acing' them. If you don't stand out, they can't taunt you.
   
2. Armor up. Assume a facade of uncaring, or downright smugness at how well you're doing. You're doing better than everyone else, and that's THEIR problem, not YOURS.
3. Clique up. If there are other smart kids around, pack with them. It's a lot easier to defend against the forces of stupidity if you are in a group of your fellow smart kids.
   

Gifted and Talented classes provide a way to meet other kids who are as smart as you are. This helps reduce the isolation experienced by these students. In my case this was a big help.

I said earlier that Gifted and Talented is special ed for smart kids. This is quite distinct from Special Ed, note the capitals, which is designed to try and get greatly under performing kids to at least grade-level. G&T programs also are not funded by Special Ed levies, which puts them right next to languages, band, and theater on the budgetary chopping block come time to 'trim the fat'. The G&T program came before the School Board several times while I was in the program, and each time was saved from outright extinction. The school district I grew up in is now even more cash-strapped than it was when I went through, so I'm fairly certain that the G&T program is all but scrapped.

Gifted and Talented helped me a lot. In 3rd grade I was reading at grade level. By first term 4th grade I was still reading at 3rd grade level. By the end of 4th grade I was reading at a 5th grade level. I was reading (and writing, by what I've been told) at a 12th grade level by the time I hit High School in 9th grade.

In 2nd grade I had exactly one other kid in my class that I knew was about my level for smarts, and that wasn't enough incentive to try and keep up. In 3rd grade he transferred to another area and I was The Smart Kid and all that entailed. That was the year I was entered into the Gifted and Talented program. In 4th grade I moved to another area, a 4-6 once, and met the same kid I had in 2nd grade. He found a social group there that I was introduced to, many of whom were also G&T students or otherwise working ahead of grade-level. This, plus the G&T classes helped catapult me from the rut I was in.

Gifted and Talented programs are a valuable service. In this era of focusing so strongly on keeping students from falling off the bottom, we have forgotten that kids at the top fall off too. A sparkingly smart student who is a D- student because they are disengaged will get absolutely nothing from a 'remedial' class that helps those with learning disabilities. In fact, it may hurt them even more. This is a tragedy.