Ed is an EE, PE, and author in Poughkeepsie, New York. You can contact him at [email protected].
You wake in a cold, shaking sweat at a dark, early hour. Today is the final exam of a course that's required for graduation. You remember registering for the class and buying the books, but somehow you never attended the lectures or did the assignments. Suddenly, it's the end of the year and you won't graduate because you screwed up.
The fact that you've been out of school for, oh, a quarter-century or so has no effect on the intensity of The Dream. It's the same heart-stopping jolt every time.
Sound familiar? Every techie I've asked has experienced some version of The Dream. It rarely involves a major course, but the class is always required and inexplicably overlooked. Each person I asked seems relieved to discover that someone else has The Dream, too.
Liberal arts majors have nightmares, but generally not The Dream. There's something different about an engineering education or, perhaps, about the people who become engineers. They know, deep down inside, with absolute certainty, that failure is their fault, that the universe isn't forgiving, that it's their responsibility to Get Things Right.
If people with that particular mindset might be a valuable asset in your business, read on. I have some bad news and, perhaps, a bit of good news about your hiring prospects.
Trend Lines
Common knowledge says that test scores have consistently declined over the years. It turns out that's not entirely true, as the very few valid long-term test histories indicate that scores bottomed out in the late '60s and mid '70s, then began a gradual recovery that continues today.
It is not clear whether the increasing test scores are the results of better learning, more specific teaching, or twiddling with the scoring. "Teaching to the test" can produce higher scores without actually giving the students any useful general knowledge. On the other hand, say critics, at least the kids are learning something.
The Scholastic Assessment Test, formerly the Scholastic Aptitude Test and best known as the SAT, provided a solid baseline up through 1995, when the mean score was arbitrarily adjusted upward by about 100 points (http://www.ets.org/). Recalculating all previous scores on the new basis introduced major biases into the historic record.
The SATs, as many of you know by heart, measure both verbal and mathematical skills on a scale of 200-800 points. By the early 1990s the average scores were in the low and high 400s, respectively, which indicated feeble verbal and weak math abilities. Recentering both scores to 500 boosted prior scores by unequal amounts, leading one to believe that earlier test takers had better verbal than math skills, exactly the opposite of the truth.
The U.S. Department of Education, through its National Center for Educational Statistics (NCES) administers a National Assessment of Educational Program (NAEP) long-term testing series (http://www.nces.ed.gov/nationsreportcard/about/trend.asp) to provide a baseline of student achievement in math, science, and reading over the years. It began in the early 1970s and continues today, using the same test questions and the same methods to produce comparable results.
The test scores vary over a narrow range, higher for the earlier tests, lower in the 1980s, and increasing in the 1990s. Although the scores do, indeed, show statistically valid differences, the variations are only a few points and the trend lines look depressingly flat to me.
In all of the documents presenting these results, the year-to-year variation of the scores gets major attention early on, while the absolute scores don't appear until much later in the text.
It turns out that there's a good reason for that.
State of the Students
My mother occasionally reminds me that she drilled the multiplication tables into my head in third grade and that I didn't like it one little bit. I may pass that family tradition on to our third-grader, but she seems to be catching on without major parental intervention.
The Brookings Institution published a report on mathematics education in September 2000 (http://www.brook.edu/gs/brown/bc_report/2000/toc.htm), which observed, "A more sensible goal is for all students to master arithmetic by the end of eighth grade, if not before." Uh, master arithmetic by eighth grade? I distinctly recall taking algebra in eighth grade, admittedly in an Advanced Placement course one year earlier than the usual schedule. Mom's drills paid off.
Dutchess Community College here in Poughkeepsie offers Math 091, Beginning Algebra, which the catalog describes as "Intended for students who must bring their mathematics proficiency to the level necessary for entrance into [Math] 100, 109, or 131." In mid December 2001, nearly all of the 16 sections available for the Spring 2002 semester were full.
Math 100, Intermediate Algebra, has 17 nearly filled sections. Strange though it may seem, you can earn three college credits while getting up to speed on high-school math.
One can reasonably infer a significant demand for 8th-grade math refresher courses at the college freshman level. Community colleges don't generally attract the highest performing students, but the magnitude of their problem should be fairly obvious. We'll see shortly why they're in this situation.
A slightly different series of NAEP tests produces what's called "The Nation's Report Card" (http://www.nces.ed.gov/nationsreportcard/), an assessment of student abilities in various academic fields. The math and science scores should interest high-tech employers, although the students' reading ability may also count for something.
The NAEP classifies the test scores into three Achievement Levels: Basic, Proficient, and Advanced. Separate 4th, 8th-, and 12th-grade tests in various subject areas generate a blizzard of numbers, but the Levels provide a common reference.
The numeric scores range from 0 to 300 for Science and 0 to 500 for Math and Reading, with cutoffs for the three Achievement Levels set separately for each grade and subject. Panels of educators and citizens determine what students should know in each grade, then divide the questions into sets that, when answered correctly, indicate mastery of the subject at that grade level. As a result, however, you cannot compare numeric scores across grades or subjects.
The Basic Achievement Level indicates only partial mastery of the subjects. In effect, this is a failing score because the student has not learned the material of the grade. It is not the lowest score, simply an indicator of a particular set of abilities or, more precisely, their absence.
At the Proficient Achievement Level, the student demonstrates "solid academic performance." The lower cutoff score represents the minimum a student should know about a subject in that grade. Higher numeric scores indicate higher proficiency and better comprehension, as you might expect.
Advanced scores lie above Proficient and demonstrate complete mastery of the tested material for the grade. By definition, relatively few students will achieve this category.
The most recent Mathematics tests, administered in 2000, show 2 percent of 12th-grade students score in the Advanced range, 14 percent in Proficient, 48 percent in Basic, and 35 percent below the Basic level. That means only 17 percent (rounded up from the raw numbers) of the student population has demonstrated mastery of the grade-level material they're supposed to know.
In the Science assessment, also administered in 2000, the scores follow a similar pattern: 2 percent Advanced, 16 percent Proficient, 34 percent Basic, and 47 percent below Basic. Only one in five students exhibits grade level or better competence and half fall below even partial mastery of the subject.
Although reading may be fundamental, only 40 percent of our students reach the Proficient or Advanced levels in those tests. The other tests are worse: Would you be surprised to discover that only 11 percent know their history very well at all?
There isn't a whole lot I can say about those figures. I encourage you to check the original reports for yourself for the disclaimers, caveats, and usage recommendations. As the saying goes, I'm not making this stuff up, no matter how much I wish I were.
The NCES also participates in the Program for International Student Assessment for 15-year-olds in math, science, and reading (http://www.nces.ed.gov/surveys/pisa/2000highlights.asp). To make a long story short, U.S. students don't score particularly well in those tests either, although the difference isn't as large as you might have been led to believe. The U.S. is very large and diverse in contrast with some of the higher ranking countries, a condition that introduces some difficulties of its own. In short, the higher scoring countries are not a whole lot better off than the U.S., at least as I read the numbers.
None of which can possibly excuse four out of five students being incompetent in math and science. Those falling below the Proficient, not to mention below the Basic Level will have a tough time becoming responsible citizens, regardless of which country they call home. They can't handle simple arithmetic, they don't understand how technology works, they can't apply the skills they were supposed to learn in school to the real world they couldn't wait to get into, and their lack of reading skills cuts them off from day-to-day learning.
Welcome to our future.
See It Live!
The good news, if any, comes from knowing that the best students are still very, very good. Those scoring in the Advanced Level, the top 2 percent, are essentially all the people who can implement our technology-laden future. If only we had more of them!
For the last several years, I've been watching the best and brightest college and high-school students from around the world, as well as some grizzled engineers and outright kids, gather at the annual Fire-fighting Home Robots Contest at Trinity College (http://www.trincoll.edu/events/robot/) in Hartford, Connecticut.
The contest's problem specification seems straightforward: Scratch-build a robotic device that can navigate through a model of a single-floor house, find a burning candle, and extinguish it, all without human intervention. As organizer Jake Mendelssohn puts it, "The only people who think this is easy haven't tried it."
The robots have relatively few, fairly sensible design restrictions. Most significantly, they must fit within a one-foot cube and extinguish the flame without using hazardous materials. Entries include fans on Lego block agglomerations, inert-gas nozzles protruding from precise cubic-foot blocks whittled of solid alloy, and balloons taped to Popsicle-stick-and-glue gizmos, with a few shapeless tangles of wire and plastic thrown in for good measure. There appears to be no correlation between complexity and success; see Figure 1.
Detecting a burning candle seems like it should be simple, but the real world interferes with every ought-to-work, classroom-perfect technique. Visible-light sensors lock onto daylight through open doors, IR sensors pick up residual heat left on a room wall by a long-gone candle, and UV sensors dutifully report camera strobes. Carbon-dioxide sniffers assume the previous entrant didn't hose the candle down with that gas. And so forth and so on.
The Saturday trial runs also demonstrate that navigating the house model isn't nearly as easy as it seemed during the previous semester's design phase. Robots freeze at the starting line, persistently ram the walls, pirouette gracefully before whirling in a hard spin, and wander aimlessly through the halls. You're guaranteed to see at least one robot enter a room, decide there's nothing burning there, and knock the candle over as it exits.
"I can fix that!" becomes the mantra as frantic last-minute code changes, mechanical tweaks, and sensor adjustments chase algorithms and mechanisms rattling down those "can't possibly happen" event trees.
All in all, it's a high-energy wonder to behold!
The Payoff
What does that have to do with education and embedded systems? Think of Hartford as a talent-scouting venue for embedded developers.
How long does it take to convince an applications designer that there's a difference between the project's specifications and the real world's demands? That textbook-perfect hardware probably won't survive its first power-on without a touch of experience? That there's more to sensor interfacing than just hitching a webcam to an Ethernet port?
The college students entering the Hartford contest come largely from engineering programs, the high-school students are paying careful attention to their science and math grades, and the elementary-school students (yes, there are some!) may have "techie" burned into their genes. They're growing up writing code and finding out what doesn't work, building gizmos and watching them crash, discovering that they can learn what they don't know, enduring lessons that simply can't come from open books and multiple-choice tests.
They pull all-nighters getting their robots working, only to fall sleep across a row of chairs just before their contest time slots. They engage in (mostly) friendly espionage, swap techniques and stories and tall tales, watch other teams' robots in action, then race to incorporate the latest discoveries. They cry in frustration, laugh in delight, and whoop in triumph.
You want iterative design, rapid application development, and fast prototyping? Here you'll find folks who know what those buzzwords actually mean, in an environment that makes failures both painfully obvious and easy to diagnose. With any luck, these students will become part of our future, too. We desperately need more citizens with these experiences, even if they never do any embedded development for the rest of their lives.
They probably don't know about The Dream yet, though. Let's not tell them until after graduation. Okay?
Reentry Checklist
The 9th Annual Fire-fighting Home Robots Contest occurs on Sunday, April 21, 2002, with seminars and preliminary test runs on Saturday. I'll present a Saturday seminar on analog gotchas for digital designers. See you there!
You'll find several other robotics contests, most notably the RoboCup soccer competition at http://www.robocup.org/. Find a contest near you and check out the state of the art.
If you're looking to broaden your mind on embedded systems topics, begin with Jack Ganssle's book reviews at http://ganssle.com/bkreviews.htm.
As if you didn't know it already, watching television isn't good for high-school seniors. There's a nearly perfect linear fit between test scores and hours of TV per day. Hint: The slope isn't positive. See http://www.nces.ed.gov/nationsreportcard/science/results/television-g12.asp for a chilling graph.
The problems with schools may not lie entirely within their walls. Even if you don't agree with John Rosemond's child-rearing and student-support advice, he's well worth reading. Start at http://www.rosemond.com/, then do a Google search for rebuttals.
Innumeracy: Mathematical Illiteracy and Its Consequences (Hill & Wang, 1988, ISBN 0809058405), by John Allen Paulos, shows the consequences of not knowing how numbers work. It was scary before I read the test scores and it's worse now.
Finally, take a look at http://www.boots-coots-iwc.com/references/16_How%20well%20control%20techniques_a.htm for some fire-extinguishing ideas that won't appear at the robotics contest.
DDJ
