The Third Annual International Computer Problem Solving Contest results. Donald T. Piele.

When the bulk of this year's returns for the Third Annual International Computer Problem Solving Contest arrived at the University of Wisconsin--Parkside, the mailroom clerk scratched her head and wondered, "Why are people from all over the United States and several foreign countries sending letters and packages to a mathematics teacher here? More than 100 pieces have arrived today alone!"

We began to wonder what we had gotten ourselves into. Opening each return, scanning the results sheet, and reading the comments enclosed took two hours a day. The stack of program listings and sample runs from teams who reported solving four or five problems correctly was steadily growting--especially in the Elementary Division. All of this occurred, of course, as final exams at the University began; grades were due soon in the registrar's office. Our work for the next two weeks was cut out for us.

One week after the deadline for returning the results, we began to grade the solutions. We spent several days reviewing sample runs and reading program listings. All solutions to a particular problem were compared, and an extra 1-5 points were awarded for ease of reading and simplicity of design. In a few cases, we made comparisons by typing and running the programs. To distinguish between the top two or three results, the programs were compared side by side. Of course, all judgments in style and ease of understanding are subjective and, therefore, difficult--a bit like trying to rank works of art. Elementary Division

The top ranked team in the Elementary Division was a pair of sixth graders, Ethan Straffin and Brad Perkins from Todd School in Beloit, WI. Ethan and Brad worked together to solve all five programs using an Apple II computer. Their solutions were judged to be the best among the twelve teams who solved all five problems correctly. Kent Weber, the Beloit contest director, reported that Ethan and Brad finished the problems in 52 minutes. Aside from the two-hour time limit, the time it takes to solve the problems is not considered in grading the solutions, but it was interesting to know how quickly this pair worked.

This was the first year that a team from Wisconsin had placed number one in any division of the contest. Since Beloit is only an hour's drive from the University, I presented the trophy myself at a special session of the school board to honor outstanding achievements by students in the Beloit School district. After the awards ceremony, I chatted with both young men.

The first thing that Ethan, the team captain, wanted to know was where we had taken points off from their programs. I couldn't answer that because awarding points for design and style is a subjective judgment for all the top teams who solve a particular program. We then talked about what he likes to do with his computer. Ethan, who recently upgraded his equipment from a Vic 20 to a Commodore 64, was working on a Monopoly simulation program with graphics displays of the playing board, rolling of the dice, and detailed overlays of properties when a player lands on them.

Although Brad Perkins, the other half of the winning team, does not have his own computer, he can use the "Apple Orchard"--32 Apple computers in a central facility in Beloit for teaching computer literacy to 2000 children in grades 3 to 6. After school, Brad and other computer club members can use the computers. Brad also likes to write programs that use graphics.

Twelve elementary teams solved all five problems, and among the top ten teams the difference between the final score was only 6 points. A complete list of the top ten teams appears in Figure 1. Junior Division

The first place team in the Junior Division consisted of one competitor, Robert T. Adams, a ninth grader from Woodson High School in Fairfax, VA. Robert and three other teams solved four problems correctly, but his well-constructed and efficient programs put him in the top spot.

During the past year Robert competed in several mathematical contests: the Virginia Mathematics League, Washington Metropolitan Competition, and Atlantic Regional Competition. He likes soccer, football, English, computers, and his mathematics and computer science coach, Andre Samson. "Mr. Samson gives us problems that really make us think," said Robert.

The problems for the Junio Division were apparently too difficult this year. We did not receive enough returns with four or more problems solved correctly to be able to rank beyond the top four teams. The top four teams that correctly solved four problems appear in Figure 2. Senior Division

The Senior Division winner this year also entered as a one-person team: John Rompel from Piedmont High School, Piedmont, CA. John is the first contestant to have his name engraved on two winning trophies; he won the Junior Division title in 1981, the first year of the contest.

Both John and the second place team solved all five problems--an outstanding accomplishment given the difficulty of the last problem. He said it helped that he had worked on a similar eight queens problem before. His solutions were very elegant, well-documented, and easy to read. John, who ha a Radio Shack lookalike at home, has been programming professionally for four years for a large software and consulting firm. He loves to work with different operating systems, and he programs in most low and high level languages, including Pascal, C, Fortran, and assembly for the Z-80, 6502, and 8086. A junior this year, John plans to skip his senior year to enter the University of California at Berkeley in the fall, majoring in computer science. A complete list of the top nine teams appears in Figure 3. Statistics

Each contest director was asked to return the results form from his local contest. The form, which summarizes the local results, helps us evaluate the difficulty of the problems. We used these returns to compile the overall average results (Table 1).

As the statistics confirm, the junior division problems were much too difficult this year. Last year, the junior division problems were too easy, so we increased their difficulty. Apparently our adjustment went too far. The senior division problems were just slightly too hard. Had we started out with a simpler first problem, we could have avoided this; we learn something every year. Feedback

The comments that we received this year were very informative and useful. In general, directors of contests with stronger and more experienced teams didn't mind the difficulty of the problems as much as those who were new to the contests. Junior division teams were especially hard hit, and some directors thought that there was not enough difference between the junior and senior levels. Many people believed that we should have at least one elementary problem--a confidence builder--that most entrants could solve.

The next major concern was the wording of some problems. Problem #2--De Bug was not stated as clearly as it should have been. A less-than sign was missing in the problem, which caused the output of some teams to look different than our sample runs. It was easy to think of the output in a slightly different way too. Most judges understood the problems here and accepted the different output.

A few judges complained about the difficulties of grading a simulation problem. The problem is that one can't be sure the program is correct by looking at the sample run because the runs are all different. Finally, some directors thought that senior problem #3, Factorial Power, was too tricky. It asked the student to compute the right-most non-zero digit in the decimal representation of N! without actually computing all the digits in N!.

Many very goods students kept only one significant digit in their computations, which proved to be insufficient. Two teams pointed out afterward that even our sample solution, which kep two significant digits at each step, would not work for all numbers between 1 and 500. Our solution gave the correct results for the test numbers required in the problem, but failed for a few numbers between 1 and 500 that we did not try. It was noted that it is necessary to keep these significant digits at each step to solve the problem correctly. We out-tricked ourselves on this one. Recommendations

The response to the contest this year was tremendous. It is very satisfying to know that people find our contest interesting and useful in their computer program, but a little scary. We thought we were just having a little fun with computer problem solving challenges, and now we get calls from organizations that are planning to sponsor a statewide contest for thousands of students! We estimate that at this year's 465 contest sites worldwide, about 10,000 students participated in our contest.

Consequently, we think that we must try to improve the contest for all the students involved while keeping our emphasis on problem solving. In addition to trying to eliminate tricky problems and making the problems easier to read, we are considering a few improvements for next year:

* Make two problems very simple to ensure that more teams solve at least one problem correctly.

* Keep two problems at the intermediate level.

* Keep the fifth and final problem difficult so that only the very best teams are able to solve it.

* Provide an algorithm to generate random numbers for all simulation problems so that correct sample runs to give expected results. This will make them easier to judge.

* Establish the last Saturday in April as the official date of each year's contest. The preceding Friday will always be the alternative date for schools that are unable to hold the contest on a Saturday.

* Send registration forms and information in December to all contest directors who participated the previous year and to others who ask to be placed on our mailing list. Conclusion

Outstanding teams were commended locally, usually in newspaper articles or with a trophy. Because of limited time and resources, we can judge only the very best results from all over the world. The winning international team members have their names engraved on one of three traveling trophies, which they keep for one year. (A trophy was added this year for the elementary division.) Placing in the contest is a significant achievement; we enjoy watching it happen and publicizing the results.

Finally, we wish to thank all the local area contest directors, contest duplicator, judges, and team members for helping to make the contest possible this year. As many contest directors told us, "Wait 'til next year!" Post Script

Next year's contest will be held on Saturday, April 28, 1984. For those who cannot hold the contest on a Saturday, the alternative date is Friday, April 27. To obtain a registration form, send your name and address to: Donald T. Piele, University of Wisconsin-Parkside, Kenosha, WI 53141.