Texas Instruments seems to be locked into what, in aerospace, would be called a “coffin corner.” Standardized testing is eventually going to go over to the computer anyway, for reasons of administrative convenience. Any tool the student is permitted to use will be on the screen anyway, and there will be no use of paper whatsoever. No books, no test booklets, no scratch paper, no electronic devices, nada. That gives the invigilator a clear “bright line” to uphold, especially when people are taking tests in different subjects, side by side. The test-taker either has a clean desk, or he doesn’t.

In standardized testing as it presently exists, you need a calculator to reduce calculations on paper to numerical form, so that you can indicate which of four or five numerical answers is the right one. In an ideal computer math testing system, you don’t do that. You drag mathematical expressions around the screen with the mouse, inserting them into each other, combining them, etc., in short, doing something rather like programming, or playing chess. For example, in ninth-grade algebra, you can have a program which automatically generates problems of the form:

a*x+b=c*x+d

using a random number generator to get a,b, c, and x, and setting d to match. The program can sieve to insure that these numbers are chosen in such a way that intermediate results fall within desired limits.A sample session would look like the following:

Computer: 11*x+7=6*x+27, what to do?
Student: -6*x
Computer: 11*x-6*x+7=6*x-6*x+27, 6*x-6*x= (6-6)*x=0, 11*x-6*x=?
Student: (11-6)*x
Computer: (11-6)*x=?
Student: 5*x
Computer: 5*x+7=27, what to do?
Student: -7
Computer: 5*x+7-7=27-7, 7-7=0, 27-7=?
Student: 20
Computer: 5*x=20, what to do?
Student: /5
Computer: 5/5*x=20/5, 5/5=1, 20/5=?
Student: 4
Computer: x=4 for: 11*x+7=6*x+27.
[door opens, small piece of hard candy lands on desktop.]

With a somewhat more aggressive student:

Computer: 11*x+7=6*x+27, what to do?
Student: -6*x, 11*x-6*x=(11-6)*x=5*x
Computer: 11*x-6*x+7=6*x-6*x+27, 6*x-6*x= (6-6)*x=0, 11*x-6*x=(11-6)*x=5*x [GOOD, GOOD, GOOD]
Computer: 5*x+7=27, what to do?
Student: -7, 27-7=20
Computer: 5*x+7-7=27-7, 7-7=0, 27-7=20 [GOOD, GOOD]
Computer: 5*x=20, what to do?
Student: /5, 20/5=4
Computer: 5/5*x=20/5, 5/5=1, 20/5=4 [GOOD,GOOD]
Computer: x=4 for: 11*x+7=6*x+27.
[door opens, two small pieces of hard candy land on desktop.] (fair is fair)

What Texas Instruments might produce is a self-contained examination-room terminal, with as few potential vulnerabilities as possible, at the cost of reduced capability (no USB ports, keyboard and mouse plug in via PS/2 ports, no storage devices, etc.). This terminal plugs into a matching server, via a port which can never be superuser/root on the server. And so on and so forth. The system would be produced in comparatively small numbers, and would command a premium price.

The same hardware system could be used to test students in a foreign language. There would be a different program, and the student would use the mouse to position the cursor at various points in blank lines between lines of the foreign language, and type in translations. The student would not be not allowed to have a dictionary, but he could right-click on words to “concede” them, and cause glosses to appear. The penalty for conceding a word would depend on how rare it was, and on how many words the student had to concede altogether.

The new-model standardized test would not have a single multiple-choice question in it. For quantitative subjects, the testing program would make the student work problems and look over his shoulder as he did so. For verbal subjects, it could require the student to type things into the computer, and even if a human grader was not economically feasible, the student’s typed input could at least be fed into a program which statistically compared it to a rubric.