To paraphrase Shakespeare, I am writing this obituary not
to praise Daniel H. Wagner, but to help us learn from him. Beginning
with his at-sea analyses of aircraft capabilities for the Navy's
Operations Evaluation Group (OEG) in 1952, his career dates back
to the initiation of operations research (O/R) as a recognized
profession. As his colleague in the mathematics department at
Brown University from 1946 to 1948, and later his partner in
Kettelle & Wagner (K&W) from 1957 to 1963, I shall start
by sharing with present-day INFORMSers some of the excitement
of those early days.
The excitement was real. It combined the "ach-phenomenon"
characteristic of the discovery of a new theorem with the belief
that this new discipline could revolutionize the non-scientific
world. At Brown, we had been studying math for math's sake -
real variables, Banach spaces, finite groups, and cohomology.
Beauty, rigor, and elegance - certainly not utility, were our
goals. I'm sure Dan would agree that his Ph.D. thesis - On Free
Products of Groups - had not the slightest application for O/R.
By the time Dan and I converged nine years later, we had tasted
this excitement of O/R - Dan at OEG, under the guidance of Jacinto
Steinhart; and I at A.D.Little, mentored by Arthur A. Brown and
George Kimball.
Our partnership was quickly given a challenge commensurate
with our eagerness. The Soviets had developed intercontinental
ballistic missiles (called ICBMs, to avoid confusion with a business
machine company), and the US was building the Ballistic Missile
Early Warning System (BMEWS), so that it could flush its strategic
bombers before they were destroyed by these missiles. BMEWS could
get individual radar returns from a space object, associate them
into target tracks, and if enough tracks appeared to be directed
at the US we were to launch SAC (Strategic Air Command, then
headed by General Curtis LeMay, who smoked big cigars). The problem
was that other things, most notably aurora borealis (The radar
was in northern Greenland.) and space debris, could also generate
radar returns. SAC decreed that a false alarm rate of one per
100 years was acceptable, and asked us for alarm rules that would
allow them to launch as many bombers as possible (before impact)
at the cost of that false alarm rate. Given the morass of data
on false returns, computation of a false alarm rate that any
alarm rule would generate was a daunting (not to say scary) task.
We made a key breakthrough - by parsing the behavior into (1)
the probability that one more report would trigger an alarm and
(2) the rate at which such reports would arrive. (Like so many
results, this idea was far less obvious before we found it than
it seems in retrospect.) Our alarm rules were in fact installed,
and were operational for many years. Fortunately, the cold war
has ended before those fateful 100 years elapsed!
Here I can't resist telling the story (well publicized at
the time) of the first BMEWS false alarm. It occurred the evening
of the first day the system went operational. When the moon rose,
it was plumb in the middle of the radar coverage. At a range
of some 275,000 miles, it should not have looked at all like
a Soviet missile (with typical ranges of one or two thousand
miles). But radars are saddled with a pulse repetition rate;
in our case this meant we could only measure the range modulo
3000 miles. For example, an object 275,000 miles away would give
a range of 2000 miles. So the moon generated a massive Soviet
raid. Fortunately, that was the week that Kruschev was at the
United Nations in New York, pounding his shoe on the table, and
the SAC general in charge used that fact to reinforce his belief
that the alarm was false!
In 1963, after 5 years, we split up. To avoid even the appearance
of acrimony, we never announced the reason. (We did announce
a general reason - that a partnership is like a marriage, except
it doesn't have sex and kids going for it.) In fact, we had a
fundamental disagreement about the extent of rigorous mathematical
defense for the results that should be presented to our clients.
Dan felt that I was too elliptical (inclined to omit much of
the rigor), and I felt he was too hyperbolic (insisting on details
only a mathematician could love). The 35 intervening years have
convinced me that Dan had a better point than I gave him credit
for. As professionals, we must set our own standards for the
validity of our work. If his were too high, mine were surely
too low. Of course, you are confronted with a dilemma when the
client buys your preliminary result and doesn't want to pay for
your subsequent rigorous follow-up!
When we split up, Dan formed Daniel H. Wagner, Associates
(DHWA) (and I formed Kettelle Associates, there being sufficiently
few Kettelles). His personal technical accomplishments in those
early years concentrated in submarine warfare - a report (jointly
with Tony Richardson and Ed Loane) on the "Theory of Cumulative
Detection Probability", and another (jointly with Loane)
entitled "Submarine-versus-Submarine Secure Sweep Width
Manual'. In those days, when Soviet attack submarines were the
principal threat to our carrier Navy, and when Soviet ballistic
missile submarines were threatening warningless attacks on the
national command authority; this stuff was central to efficient
naval operations. He also spearheaded work (for Morgan Guaranty)
on optimal coupon selection in bidding on new issues of municipal
bonds.
 |
|
Rear: Bill Browning, Tom Corwin, Frank
Engel, Bruce Scranton,
Stan Benkoski, Bill Barker
Middle: Burg Rhodes, Bernie McCabe,
Les Arnold, Barry Belkin,
Joe Bolmarcich
Front: Dave Bossard, Dan Wagner, Tony
Richardson, Larry Stone
|
Nevertheless, it is my (I believe well-shared) conviction
that Dan's biggest contribution to our profession (one that will
be replicated long after his death) was his injection of "pure"
mathematicians into the field of O/R. He recognized that the
rigorous thinking and the scientific maturity that can be found
in first-class math Ph.D.s are of great potential value to the
identification and solution of the operational problems that
are O/R's raisons d'etre. He devoted his professional life to
this conviction. As a result, we are blessed with scores of such
mathematicians. To illustrate, I can't help listing a few of
these known to me personally: Dave Bossard, Larry Stone, Tony
Richardson, Barry Belkin, Ed Loane, Tom Corwin (now president
of Metron, a Wagnerian spin-off), Bernie McCabe, and Joe Bolmarcich.
An extensive history of DHWA written 12 years ago illustrates
the productivity of his staff - some 150 reports to clients,
and another 91 journal and textbook publications.
A particular strength of the Wagner staff has been search
theory. Larry Stone's book; "The Theory of Optimal Search"
is still in print as an INFORMS publication. This strength led
to some remarkable adventures:
- The successful search for the H-bomb lost in the Mediterranean
off the Spanish coast in 1966, when a B-52 collided with a tanker.
- The successful search for the USS Scorpion in 1968, an attack
submarine that imploded 400 miles west of the Azores, and went
to the bottom at a depth of some 2000 fathoms.
- The successful search (guided by Larry Stone, who was then
with Metron) for the Coast Guard packet ship sunk off the coast
of South Carolina in 1857, with $400,000,000 of gold (from California)
aboard.
Not that Dan doesn't deserve praise for his O/R career. He
certainly does. But I led off this "obituary" with
the primary objective of assessing what the profession should
learn from it. It is my conviction that he was able to demonstrate
the truth of the early basic hypothesis on which the profession
was founded - that mature scientists, such as physicists like
Philip Morse, chemists like George Kimball, or mathematicians
like Dan himself, can bring a unique and valuable approach to
the solution of operational problems. I do not belittle the other
approach that has dominated the profession over the past 35 years
- graduate training specifically for the O/R profession. But
the signal success of the mathematics scholars brought into our
community by Daniel H. Wagner is a candle that the wind will
not blow out.
