Back in 1982, personal computers and workstations weren't prevalent, and system software concentrated on large timesharing machines. The technology used in primitive workstations was, well, primitive. The first wave of qualifying exams to consider such small systems was just beginning, and this is a spoof of those beginnings I did back then.
From inferno.lucid.com!lucid.com!rpg Mon Jun 21 15:40:07 1993 Received: from inferno.lucid.com by looking.clarinet.com id aa04560; 21 Jun 93 15:40 PDT Received: by inferno.lucid.com (4.1/SMI-4.1) id AA02522; Mon, 21 Jun 93 15:39:26 PDT Date: Mon, 21 Jun 93 15:39:26 PDT From: "Richard P. Gabriel" <rpg@lucid.com> Message-Id: <9306212239.AA02522@inferno.lucid.com> To: funny-request@clarinet.com Cc: tbc@col.hp.com, arg@lucid.com Subject: System Qual Status: RO
A friend noticed the following message pass through the wires recently and thought it looked familiar:
Dated: 2 September 1982
1 Pick from the following pictures the one which most accurately represents a
computer:
A. <a picture of a Cray-1>
B. <a picture of a S-1 Mark IIA>
C. <a picture of a DEC 2060>
D. <a picture of a 3 foot spool of coaxial cable>
Answer: D.
with the explanation:
I stashed this away while reading an ARPA bulletin board in school a decade ago. It was posted by a Xerox PARC hacker (you know, the guys who invented EtherNet) as a test called "the systems qual."
I wrote this fake system qual in May of 1982. The context was that I was at the CS department as a research associate at Stanford when several grad students failed the system qual which that year was given by Keith Lantz. These particular students were trying to do dissertations in timesharing. At the time it was possible to fake e-mail from anyone you wanted, and so I posted a message (reproduced below) from Lantz to "protest" the "computational correctness" that the distributed computing community was apparently trying to establish. I kept my identity secret for fear that Lantz - at that time a relatively powerful force in CSD - might retaliate.
I think now the heat is probably off, so you can attribute the message to me. To my mind, the header of the message is as good as the test itself. Note that the ^O is the standard ascii glyph for a lower case beta in the Stanford ascii character set we used at SAIL. Lower case beta was used by our mail system as a means to show the start of a message and to identify a line that would appear in the table of contents of the mail file - the idea was to indicate the sender and the header. Gene Golub was chairman of CS at the time; ``John'' is John Hennessey. The question marks in Sue Owicki's question are other SAIL characters corresponding to mathematical quantifiers, set membership, etc.
^O31-May-82 2046 Lantz@Mt-St-Coax (WireNet) Systems Qual Mail-from: COAXNET host SU-8080 rcvd at 19-May-82 2039-PST Date: 9 Apr 1982 1837-PST From: LANTZ at SU-COAXIAL Subject: System Qual To: "CSD.GOLUB@SCORE": cc: Reid at WIRE-CITY Remailed-date: 19 May 1982 2030-PST Remailed-from: Keith A. Lantz <CSL.LANTZ at SU-COAXIAL> Remailed-to: Mail-Server at SU-SHASTA-COLA Remailed-date: 19 May 1982 2030-PST Remailed-from: Mail-Server <MAIL.SERVER at SU-SHASTA-COLA> Remailed-to: Mail-Server at MT-ST-COAX Remailed-date: 19 May 1982 2030-PST Remailed-from: LANTZ-MAIL-SERVER Remailed-to: Lantz-Bit-Bucket at SU-SHASTA-COLA Redistributed-To: SU-BBOARDS@SCORE Redistributed-By: Lantz at Mt-St-Coax Redistributed-Date: 20 May 1982 Redistribution-from-Header <LANTZ at SU-SCORE at SUMEX-AIM at MT-ST-COAX> Forwarded-by: MAILSER@SCORE Forwarded-to: BBOARD at SU-AI Forwarding-Date: 20 May 1982
In response to the message sent 18 May 1982 1341-PST from GOLUB@SU-SCORE
It seems odd that you are upset that there are several people who have failed the systems qual for the last time, and from the amount of mail flying about the last few days I suppose we'll have to do something about this. Nothing can be done until John gets back in town (Thursday) so you'll just have to sit on the edge of your seat. But here is a sample of some of the questions we asked on the qual, just so you know we were fair:
1. Pick from the following pictures the one which most accurately represents a computer:
A. <a picture of a Cray-1>
B. <a picture of a S-1 Mark IIA>
C. <a picture of a DEC 2060>
D. <a picture of a 3 foot spool of coaxial cable>
Answer: D.
2. What is the limiting factor on the speed of paging on modern computer architectures?
Answer: the number of meters of coax between your 8080 and your floppy disk.
3. Name 100 advantages of personal machines over timesharing machines. Name 1 advantage
of a timeshared machine over a personal machine.
There was a lot of complaints about this question, and we admit it was intended as a trick.
Several of the people who passed spent over an hour trying to think of the advantage of timeshared machines.
4. What is the primary design consideration in designing a modern computer system?
Answer: How to maximize the ratio of coax to silicon.
5. What was the most important invention for modern computing?
A. ECL and high speed logic
B. Advanced cooling technologies
C. Video Disks
D. Cache memories
E. Coaxial cable
Answer: E.
6. What is the most important function of a modern operating system?
Answer: the mail server.
7. What is the most important measure of the sophistication of a modern operating system?
Answer: the complexity of the mail headers it produces.
8. What is the most reasonable power dissipation in modern computers:
A. equivalent to a 2000 megaton nuclear device (e.g. CRAY-1)
B. equivalent to the output of the Hoover Dam (e.g. S-1 MARK IIA)
C. equivalent to a room full of toaster ovens (e.g. a DEC 2060)
D. equivalent to a sexually satiated male mosquito in a room at absolute 0
(e.g. a single board 68000 connected to 90 miles of 300 ohm coax).
Answer: D.
(The next question is from Sue Owicki)
9. Define: A is `strongly hyperhyperimmune' if A is infinite and there is no recursive f such that
(Forall u)[W(f(u)) intersect A = empty] & (Forall u)(Forall v) u > v => W(f(u)) > W(f(v)) = empty].
A. show that if A is strongly hyperhyperimmune then A has no infinite retraceable subset.
B. show that if A is strongly cohesive then A is strongly hyperhyperimmune.
Answer: A - obvious; B - immediate corollary of A.
10. What are the design considerations in a modern display?
Answer: it must display 10^49352 points per inch and run at least at 2 baud (to support the new,
high speed 8080's out on the market).
11. Describe the new generation of `supercomputers'.
Answer: the MC68000 is...
12. Name the institutions where the most progressive computer systems work is being performed.
Answer: Bells Labs (C and Unix) because they are part of the phone company and, hence, like copper wire;
Xerox (Altos) because they have cornered the world coax market.
13. What units are used to measure the performance of modern computers?
Answer: TIPS - Thousandths of Instructions Per Second.
14. (Methodology) Why is it that large computers (e.g. Cray-1) are no longer of interest to systems people?
Answer: They run too fast to understand and to use coax effectively. Running one of these computers
on an ultra-high speed network (3 megabit net) would swamp it.
15. Where are the reliability issues centered in modern systems?
Answer: UHF connectors.
16. Define a `large program'.
Answer: A program that is more than 1/2 a page long or that has fewer than 10 lines of declarations
for each line of code.
17. How many programs have you written?
Passing answer: < 10
Failing answer: > 15
Conditional: 10<=x<=15
18. What is the largest program that you have ever heard that a real computer scientist has written?
Answer: A mail server.
19. Why is synchronization research better performed on small, modern computers connected by a network?
Answer: The coax slows things down so much that you don't have to worry about deadlocks.
20. Some people say that extremely large programs (> 500 lines) require very large computers.
How do you answer them?
Answer: If they knew what they were doing they wouldn't need a large machine. The fact that they write
such large programs means they are doing the wrong thing.
21. What should we do with people who believe in huge timesharing machines that run like hell
and who want to write gigantic programs?
Answer: Fail them on the systems qual.