- Feature Articles
- CodeSOD
- Error'd
-
Forums
-
Other Articles
- Random Article
- Alex's Soapbox
- Announcements
- Best of…
- Best of Email
- Best of the Sidebar
- Bring Your Own Code
- Coded Smorgasbord
- Mandatory Fun Day
- Off Topic
- Representative Line
- News Roundup
- Editor's Soapbox
- Software on the Rocks
- Souvenir Potpourri
- Sponsor Post
- Tales from the Interview
- The Daily WTF: Live
- Virtudyne
Admin
So Rob came Frist? I still don't get whether the article is saying his answer was wrong due to the lack of precision and his lower graded fellow students may have handed in better answers?
Admin
Wouldn't one of the other students show their correct slide rule calculation to the teacher, which would have unmasked the whole FORTRAN scam? Wouldn't the teacher have had the answer correctly from the slide rule ahead of time?
Admin
To explain: the professor assumed that the Rob's answer came from the school computer with a Fortran compiler correctly configured for scientific calculations, and therefore that Rob's answer must be more accurate than the hand-calculated answers.
Admin
You'd like to think so, but university lecturers can be astoundingly lazy.
My degree was in a building related subject, and first year project was to survey the two main towers on campus and answer the question "How far apart are they at the top compared to the base?". There was a design point to this, which escapes me now.
Campus folklore had it that in a previous year most of the year group had failed this exercise, and with it being the same wrong answer, fingers were being pointed at people for copying/cheating. As this drama unfolded, some wily undergrad got hold of some of the answers from previous years (via the usual method of hunting out older students and buying them beers).
He saw a trend that this final measurement had been increasing over time and it had only now got far enough away from whatever Professor LazyBones had measured in the 19th century to be obviously wrong, and not just a bit out.
Some of these students had ended up below the bar to pass the year, and most were understandably annoyed, so various appeals were kicked off. Unsurprisingly, the academics dug their heels in, but after much flapping of gowns they were "persuaded" they would need to go out recheck these measurements with the equipment the students had been given, and maybe with a newer theodolite that hadn't been bashed to bits by years of careless undergrads.
Concerns about "drift" in the theodolites turned out to be misplaced, they were surprisingly accurate and consistent, and told the story that the buildings were, in fact, in the process of tipping in opposite directions. Scripts were hastily re-marked, academics congratulated on discovering something so important in a timely (?) fashion.
The only evidence of this when I was there was the huge exercise to underpin the foundations Don't suppose that supports my point, but I'm bored shitless and fancied typing an anecdote.
Admin
I had a similar situation back in the day. I earned the first Computer Science degree offered by the university in the early 70s. One of my classes was on Sorting and Searching and used a textbook by Dr. Donald Knuth, who had developed a programming language called MIX just for this course. The instructor told us students that part of our grade would be based on how many times we attempted to run a homework program on the university's IBM 360. He said if it took more than two submissions that it would impact our grade. Since I was working a full-time job as a junior programmer trainee at a shop that had two (TWO!) IBM 370 machines, I was able to take advantage of my computer access. At the university computer lab I handed the computer operator a tape reel and said I was about to run a job that would call for this tape, and would he please mount it for me? Then I submitted a job to run an IBM utility to copy the MIX compiler to the tape, which I later loaded on the computer at my work. Then I was able to perfect my MIX program assignments with as many runs as necessary before running it twice on the university computer for my grade. Got an A in the course.
Admin
Having actually used a slide rule, I call BS. Even the biggest ones, with a magnifier, gave maybe three digits of precision. And that was if you were really good at guessing the distance between lines on a log scale.
Admin
Reminds me of one of my old high school teachers. We had one assignment that consisted, in part, of measuring the length of some lines with a ruler. Everyone in the class got it wrong by a few mm, although interestingly enough not everybody had the same answers. At first the teacher went on this long-winded spiel about parallax... because of course the parallax effect is enough to throw off everyone by several mm from just a foot or so away. Then someone had the bright idea of asking the teacher to measure the line again on the student's page, and that measurement was also "wrong", which set the teacher off grumbling about the entire class's rulers being mis-calibrated or something.
We finally came up with the right answer, which was that the copy machine didn't make perfect copies; the lines grew just enough in the copier to be a significant difference from the teacher's original.
Admin
I was actually expecting integer overflow. Something like 654*185 = 654185 = 1111101101101001 = 64361.
Admin
Nope. My math is wrong.
654*185 = 120990 = 11101100010011110 = 55454
Admin
I busted things the other way. The U computer (a 360/65) had a special job class for student programs running a few special programs like SPSS. Everything else had to run the "normal" class. The special class cost like 3 cents per run and the regular class, regular rates.
For our EE programs, it cost us almost 40c/ run - with a student account limit of $2.50 for the semester, you see where this is going...
Being an IT professional - by Junior year I had already worked two summers in a REAL IT shop, I figured out in about 2 minutes how to run our EE jobs in the special class. While I shared this with my class (and the professor who happened to be the department chair), I did not share it widely. So those of us in "Chuck's" class were able to easily complete all of the semester's assignments in all of our classes...
Made several friends that way and back-in-the-day, the drinking age in NY was 18...
Admin
In my high school programming class we once had a test. It was a multiple choice test. We took it one day. The next day in class the teacher was very disappointed in us. He said only two people passed the test. Then he called up the two people who passed it and realized the were probably the two lowest scoring students in the class. The top three students didn't pass. I felt pretty confident on this one and didn't pass....
After some review, he realized he used the wrong key for the answers. I think this test was one for a state certification too, so it was important that he made sure it was graded correctly before submitting the results.
Admin
Yes, I think you got it - his answers were wrong because he had extremely low precision (only 16 bits for floating point!), but because he turned in a program the instructor assumed his answers must be right and marked the other students' answers as wrong.
Also, a single 16-bit integer would hardly be sufficient for currency manipulation these days. That gets you up to $65535 dollars, or $655.35 if you need cents. Half that if you allow going into the red (negative).
Admin
Ah, the joys of inflation!
Admin
Bah! In that era, if it was worth doing, it was worth doing in APL.
Seriously, I worked in a university lab where we bought a PDP 11/70 with a floating point unit. My job was porting all the FORTRAN programs that ran on the IBM 360 down at the computer center to the 11. Stuff that ran on the 360 in an hour took all weekend on the 11, but we didn't have to pay the computer center's $500/hr bills. I hear they raised rates on everybody else when we moved out to make up for the lost revenue from our grants.
And we paid for the 11 and hired a couple more postdocs.
Our stuff would probably run in a browser on a Chromebook today, in milliseconds. But that was then.
Admin
While I was earning my CS undergrad, I had to retake Discrete Math my senior year (notably a match class not a CS one). We were covering early encryption and ciphers, and I decided to use the class to brush up on my python. With my instructor's permission I had my laptop and ran my program to complete in class assignments. On the day of the final, the instructor asked to see me after class. Turns out she wanted to see my program's answers as a way to check her work on the questions she created for the final.
Admin
My boomer parents bought a decent house for less than that, now it's a restaurant bill ... well almost, but I do like a drink!
Admin
Yes, 4k 16 bit words runs FORTRAN. Now... one word integers did NOTHING to improve precision. This simply made integer one word (16 bits) which is 1/2 a real (32 bits). Otherwise, integer and real occupied 2 words each.
Save memory. The old timey FORTRAN had EQUIVALENCE, which allowed you to overlay variables.
So -- submitters code actually worked. And, on a machine with no floating point hardware to boot.
FredW
Admin
I have to admit I needed that explanation :)
I've heard a similar story at Vienna university (from the instructor of the lab). In the basic physics lab, one of the tasks was to measure the impedance of an antenna on the roof of the Physics institute from within the lab room. And for many years, students handed in the results as expected. Until one day, some student started complaining, that he couldn't reproduce the number; His measurement was far off the expected value.
Turned out, that the antenna had been removed during a renovation years before and never been put back, and that for the years since distraught students had just copied the value from the solutions of their seniors, and nobody had caught on.
I can't quite believe though, that it took years before a student pointed out the issue, so I suspect that "correct results for YEARS" may have been embellished a bit.
Admin
To Old Fart, Regarding Knuth Search/Sort with Mix
You realize that was blatant cheating.
I don't mean the homework, that was just using resources you had available.
I mean making the computer operator do all the hard parts of getting a tape with MIX. You should have contacted Dr D.K. and acquired your own legitimate copy rather than bootlegging it. Just think of the keepsake/souvenir you would still have if you had done that!
Hey, did you accidentally invent the term "mix tape"?
Admin
I don't think you should assume that a "word" was 16 bits back then. It was much more likely to be 18, 24 or maybe even 36 bits.
Admin
When I was in the University, the first year there was this basic programming course which was supposed to be for first-years only. The requirements for courses changed that year, and because of that there were 1,200 studens on that course (I think there were about 10,000 students total at the time). This created a host of problems, and while this was decades later than the original story, the public terminals of the school were basically occupied all the time during this time. We still had some text terminals, some in classrooms and some in the hallways. There was a 15 minute restriction on the hallway stand-up terminals, but I coded at least some of the assignments during that time, standing up.
That was not the cheat. There was obviously an automatic grading system for the weekly exercises, and IIRC this was the first time it was in use. It basically took the submitted C code, compiled it, ran it and used stdin and stdout for test input and output. It then checked that the calculated output matched the input. Some people did realize where it ran and just wrote a simple program which took the stdin, wrote it to a known place and wrote garbage to the output. Then they took the output, calculated what the output should be and did that same in reverse: read the input and wrote the correct answers.
This was of course discovered at some point and measures were taken. I don't know the details, because I did the course the proper way.
Admin
TRWTF is Robert not having the integrity to let his teacher know how and why his homework was incorrect and most others was correct. I wonder, how was "slinking off" good for Robert's mental health? Is he a psychopath?
Admin
If the class was about using programs, using the wrong program is better than using a slide rule.
Admin
My version is a COBOL class where your maximum marks went down each time you compiled your program. I realised that if you booted the PC before it had completed then that session wasn't logged, but you would have seen the first error, so you could fix that and try again.
Admin
So, wait, the professor just assumed that the student who used a computer had 100% accurate answers, and didn't bother to check them at all? He never considered the possibility that there might be an error in the program, or that the student might have entered incorrect data? Wow, that's as naive as people today who say, "Scientists said that such-and-such will happen based on their computer models. Therefore it's proven fact and if you question it you must be some kind of anti-science religious fanatic."
Admin
Psychopath then, manager now....
Admin
"If the class was about using programs, using the wrong program is better than using a slide rule."
As you can see from reading the story, the class was about numerical analysis. As you can tell from an internet search, that is a branch of mathematics.
Admin
Shut the fuck up
Admin
It's possible that he did consider the possibility and looked over the logic of the program to make sure it was correct, but didn't realise that the one word integer directive would reduce the precision unacceptably (he may not even have seen that it was in use).
Admin
Dafuq you talking about? I have a basic plastic 12" student slide rule (inherited it as I'm far too young to have used one in anger) and I can easily get 3 sig figs out of it. At the low end of the scale, or if you're good at estimating, you can get a fourth. I can't imagine that it'd be hard to get 3 even out of a 5" rule -- 2 would be if I made one by hand on paper.
The high precision ones usually do 5, then if you really need more, you can treat it as a base 10,000 or base 100,000 multiplication (takes about four times as long).
Admin
"Wouldn't one of the other students show their correct slide rule calculation to the teacher, which would have unmasked the whole FORTRAN scam?"
Rule #1: The professor is always right. Rule #2: If you believe the professor may be wrong, see rule #1.
Admin
With me almost the exact situation went the other way around. Wrote a complex program to dimension a DC motor and the prof wanted to fail me "since the electronic brain did all the work for you". So I rewrote the program to print out all intermediate calculations and transcribed them to paper by hand. And passed! "See? Now you did the work!"
Admin
I worked with FX, foreign exchange, calculation using double-precision floating point arithmetic. My calculations were correct because I simulated the less-than-single-precision of the fixed-point arithmetic used by the standard contracts, as originally specified for use with fixed-point minicomputers, as used for FX trading.
And they were probably using FORTRAN as well.