Also, Mark Weiser, former PARC head and UM College Park CS professor
died last week at a shockingly young age. Someone may have gotten to
http://www.ubiq.com/weiser.html and changed the tense already, but
there's no news of his passing there. He's mister ubiquitous
computing, and he offered me excellent advice about grad school back
at UIST94 (namely, work first :-). I always thought of him as one the
first people I wanted to review a munchkins thesis... sigh. ]
April 30, 1999
Rolf Landauer, Pioneer in Computer Theory, Dies at 72
By GEORGE JOHNSON
Rolf W. Landauer, who helped solidify the slippery concept of
information and bring it firmly into the mainstream of physics, died
on Tuesday at his home in Briarcliff Manor, N.Y. He was 72.
He died of brain cancer, his family said.
"Information is inevitably physical," the kindly but irascible Dr.
Landauer often admonished colleagues. The ones and zeroes coursing
through a computer may seem abstract and ethereal, but, he told
researchers at every turn, information is rooted in the real world,
and must be understood by applying the no-nonsense laws of physics.
"Rolf Landauer did more than anyone else to establish the physics of
information processing as a serious subject for scientific inquiry,"
said Dr. Charles Bennett, his colleague at I.B.M.'s Thomas J. Watson
Research Center in Yorktown Heights, N.Y.
A German-Jewish immigrant who came to the United States as an exile
from Hitler's Germany, Rolf William Landauer graduated from
Stuyvesant High School in New York and from Harvard University with a
degree in physics. He was a pioneer in the attempt to plumb the
limits of computers. In a universe ruled by the laws of quantum
mechanics and relativity, how fast and powerful can these machines
conceivably be?
Engineers concern themselves with practical limitations of squeezing
more and more circuitry onto tiny chips. Dr. Landauer was interested
in theoretical limits: given that technology will improve
indefinitely, how soon will it run into the insuperable barriers set
by nature?
Until Dr. Landauer's landmark work in the early 1960's, it was widely
believed that processing a single bit of information, each 1 or 0 of
binary code, inevitably consumed some energy, placing a fundamental
constraint on computer power. Dr. Landauer showed, to many people's
surprise, that this was not true. As computer technology becomes ever
more efficient, calculations can be done with less energy per
computation.
But there is still a cost. At some point, the bits must be flushed
from the computer's memory so that the machine can be reset for
another computation. It is then, when the information is erased, Dr.
Landauer demonstrated, that a certain irreducible amount of energy is
lost. This result, now called Landauer's principle, has been
described as "the thermodynamic cost of forgetting."
While he initially assumed that all computers are saddled by this
loss from erasure, Dr. Bennett soon convinced him that it was
conceivable to design a reversible computer that would circumvent
Landauer's principle and operate with unprecedented efficiency. When
such a computer completed its task, one would essentially run it
backward, returning it to its initial state without discarding
information.
Though reversible computing has remained mostly theoretical, some of
the concepts are being used to make more energy-efficient laptops.
"Rolf was an old I.B.M. type who did things on the straight and
narrow," Dr. Bennett recalled. "I was a scruffy hippie. He endured
that. Many people don't like their scientific ideas overthrown. But
he was very interested and eager to discover what this all led to."
From his perch as an I.B.M. fellow in Yorktown Heights, Dr. Landauer
became a kind of critic in residence, firing off acerbic but friendly
E-mail missives when he felt a colleague or a journalist was
overselling a new theory or not being vigilant enough in weeding
sense from nonsense. "Landauer had a good nose for ideas sexy enough
to become fashionable but too grand and simple to be useful," Dr.
Bennett said.
In recent years, Dr. Landauer was especially skeptical of the new
field of quantum computation, in which scientists are trying to
harness the behavior of subatomic particles to make extremely
powerful computers.
These devices appear to work on paper, but Dr. Landauer doubted that
they would ever operate in the real world.
Current developments in the field won his grudging respect, but he
urged his colleagues to accompany their papers with the following
disclaimer: "This scheme, like all other schemes for quantum
computation, relies on speculative technology, does not in its
current form take into account all possible sources of noise,
unreliability and manufacturing error, and probably will not work."
Last fall, Seth Lloyd, a professor of mechanical engineering at the
Massachusetts Institute of Technology, was with Dr. Landauer at a
conference on quantum computing in Helsinki, Finland. During his talk
on the potential of the new technology, Dr. Lloyd projected a slide
bearing that disclaimer. The audience, recognizing it as vintage
Landauer, laughed appreciatively."I haven't put it in all my papers,"
Dr. Lloyd said, "but I use it in most of my talks. By requiring that
people be honest, he's done the field a great service."
Dr. Landauer is survived by his wife, the former Muriel Jussim; a
daughter, Karen Walsh of Dublin; two sons, Carl of Oakland, Calif.,
and Thomas of Columbus, Ohio, and three grandchildren.