Index Home About Blog
Subject: Re: [SGI] Help!!! Have I missed anything? [really: stock weirdness]
From: (John R. Mashey) 
Date: Oct 04 1995
Newsgroups: misc.invest.stocks

In article <44ub90$>, (Paul
Favret) writes:

|> Can we hear from the guy who works at SGI as to why the drop?

No .. I don't know why the drop, and have long since learned to avoid
over-interpreting the stock market.
There are tons of post-hoc speculations, most of which don't make much
sense ... except that long history of stock market weirdness has
convinced me that anything is possible. (see below).
I don't know of any SGI announcments offhand that would have caused this.

In any case, readers of this newsgroup must be aware that for a public
company, anybody who knows anything has to be very careful what they say,
especially during a "quiet period", which generally includes the last
month of a fiscal quarter through shortly after the quarterly results are
announced.  Hence, as usual, most of what is passed around comes from
people who don't know :-)

Weirdness: I can't remember if I've mentionend this here, but I just ran
across an old copy, so I'll post this as cautionary/amusing note, called:

Or How I lost 15% of a Company's Stock Value in a Few Hours."

1) When working for MIPS Computer Systems, Inc, I once did an interview
with Bill Bulkeley, a fine technology writer for the Wall Street Journal.
We talked about why the then-common MIPS-ratings didn't make much sense,
led to apples-oranges comparisons, and why we had to do better, mentioning
the SPEC benchmarks.  He'd called me because he'd run across a paper I';d
written on this topic.

2) On June 20, 1990, the article appeared prominently on the front page of 
section 2 of the WSJ.  I read it and was quite happy: Bill had written a nice
clear summary that captured what we were trying to do with SPEC & debunked
the widely-used, but useless, MIPS-ratings.  He quoted me accurately (which
sometimes does not happen)  It was 5 paragraphs long (included at end):

3) Then I got into work, and was *not* so happy:
        a) MIPS Computer Systems stock had dropped ~2 points, ~15% that
        b) There was on ongoing firestorm of analyst & investor phone calls
         asking why the stock was down, and what was the problem.
        c) I had phone calls from our CEO and CFO wondering what on earth
          I was doing.

So ... what had happened?

4) It turns out that someone other than Bill supplied the headlines,
and what they had supplied was:

        MIPS Keeps Slipping as Speed Standard

5) And people saw it, punched the SELL MIPS button, probably without
reading the article, and then, since the price dropped quickly, it
must have shown up on other traders' computer screens, and when they looked
for recent news, they probably got nothing but "MIPS Keeps Slipping"
headline, so they punched SELL to get ahead of anybody else.
That the article was about MIPS-ratings, and not MIPS Computer Systems,
was irrelevant :-)  If you don't read the article, but just look at it,
you of course will see MIPS Computer Systems in bold ... so the article
must have been about problems at MIPS...

6) All of this blew over in a week or two ... but it was an "instructive"
experience, and an example of why you don't want to go too crazy over
occasional mad gyrations in stocks, especially in the (volatile) high-tech

7) The remainder of this is the entire actual article:

William M. Bulkeley, Wednesday, June 1990
MIPS Keeps Slipping as Speed Standard
 The common measure of computer speed, MIPS, is an acronym for millions of
instructions per second, but cynical customers joke that it means
meaningless indicator of processor speed or even marketing instructions
for pushy salespeople.
 Now, a new suite of measurements called SPEC, which is being widely used by
makers of engineering workstations, makes clear just how meaningless MIPS can
be.  The 10 SPEC benchmark programs were designed to provide consistent
speed comparisons among computers doing various types of jobs.
 John R. Mashey, a vice president at MIPS Computer Systems, Inc. a Sunnyvale
workstation maker, recently compared vendors' MIPS speed claims with their
reported SPEC performance.  It turns out, users have a right to be cynical.
While Mr. Mashey says the vendors had legitimate reasons for their MIPS claims,
the ratings could badly mislead customers.
 In the study, Mr. Mashey concludes that one of his company's workstations
rated at 20 MIPS "shows similar integer performance to an International
Business Machines Corp RS/6000 rated at 34.5 MIPS, and is twice as fast as a
Sun Microsystems Inc. 12.5 MIPS workstation.  A Digital Equipment Corp's
BVAX translates into about 90% of its claimed MIPS, but a Digital RISC
workstation is about 83%.
 Overall, Mr. Mashey says that actual performance as measured by SPEC ranges
from 54% to 96% of companies' MIPS claims.  In floating-point operations,
which reflect scientific performance, differences range from 34% to 135%.
"There is no such thing as a MIPS rating that means much," he concludes.

-john mashey    DISCLAIMER: <generic disclaimer, I speak for me only, etc>
DDD:    415-390-3090    FAX: 415-967-8496
USPS:   Silicon Graphics 6L-005, 2011 N. Shoreline Blvd, Mountain View, CA 94039-7311

Index Home About Blog