IEEE Spectrum December, 2007 - 3

FORUM
"Datacenterscan
requireupwardof
25000kilowatts,
enoughpowerto
servebetween
10000and
15000homes"
-TomSchaeffer
GooGle Goes
Greenish

and abundant." While growth
and economic development are
crucial to communities, electric loads such as these come
with system and environmental costs that you never seem
to find in a Web search.
Tom Schaeffer
IEEE Member
Longmont, Colo.

The article "The Greening of
Google" [October] outlined the
company's plan to locate a data
center in Oregon near "cheap and
abundant" hydropower. This is
presented as if that were somehow a positive thing consistent
with its conservation efforts.
While admittedly renewable,
hydro is a limited resource that in the october issue's Spectral
the company has elected to use
Lines there's a picture showing
up, most likely based on the "Google's Sergey Brin and Larry
"cheap" part of the quote.
Page plugging in a RechargeIT
Data centers can require
hybrid electric car."
upward of 25 000 kilowatts,
You failed to mention that
enough power to serve between
once the car was charged, they
10 000 and 15 000 homes or could've driven it to the airport
between 50 and 80 typical
to fly on their private Boeing 767
commercial customers. Most jetliner, which as we all know,
of that load is just for cooling
is a very fuel-efficient mode of
the waste heat from the elec- personal transportation.
tronics. The hydro resource
Joseph Katz
Google will consume must be
IEEE Fellow
replaced with some type of
Stony Brook, N.Y.
base-load generating unit that
probably won't have such a
noT as hard as we
positive public relations value. ThouGhT
Unless Google pays a "tap" fee
in Feng-hsiung hsu's article
equivalent to the tremendous "Cracking Go" [October], he
costs associated with build- lists the number of possible
ing all the facilities required
game positions for chess and
to generate and transmit this
Go at 10 120 and 10 170. I'm peradditional energy, these costs
plexed by his number for chess.
will be paid for by others in the A simple analysis of Go indicommunity. In return, the local
cates that, as an upper limit,
community might benefit from
not accounting for certain
as few as 20 new jobs.
"impossible" configurations as
The focus should continue
restricted by the rules, there
to be on energy efficiency and
are three possible states (black
conservation rather than "cheap
stone, white stone, no stone) in

IEEE Spectrum | December 2007 | NA

each of the 19-by-19 interstices.
So that comes out to 3 (19 x 19)
possible game positions, or
about 1.74 x 10172-close enough
to the 10 170 number given
by Hsu. If I apply this same
analysis to chess, however, I
fall quite a bit short. There are
six unique white pieces (pawn,
knight, rook, bishop, king, and
queen) and six unique black
pieces, as well as the possibility of no piece in a given spot,
on a grid of 64 squares, for a
possible 13 64 game positions,
or about 2 x 10 71 . Even this
number greatly overestimates
things, because there cannot,
for example, be 64 white pawns
on a board! The gap between
chess and Go is much wider
than Hsu indicates.
Ben Thompson
State College, Pa.
Senior Editor Philip E. Ross
responds: As Thompson and
several other readers noted,
there was an error in our table
comparing chess with Go. It
turns out that the best estimate of the number of possible
game positions in chess is 1044 ,
76 orders of magnitude lower
than the number we listed. We
regret the error.

Basic calculaTions

Thanks to IEEE Spectrum
and Kenneth R. Foster for
rekindling some old memories with the review of the new
HP 35s calculator [Resources,
October]. I was a Ph.D. student back in 1975 and somehow
managed to find the money to
buy a shiny, brand-new HP 35.
It made me the envy of most
of the other graduate students
in the school of engineering
at the University of Auckland.
This was in the days when programming a computer involved
a stack of punch cards, and you
got the answer the next day-
if you hadn't made a mistake
in your Fortran. I seem to
remember the calculator had a

distinctive, almost sweet smell
to the plastic.
The HP 35 worked f lawlessly for several years and
made it with me to Ottawa,
where I had taken a job with
Miller Communications in 1979.
Unfortunately I dropped it one
day, and a big ceramic IC cracked.
HP wanted $200 to fix the calculator. A new unit of similar
performance (but without the
nice smell) was about half that
price. Still, I have regretted the
decision to buy a second-rate allplastic IC calculator ever since.
I have not managed to kill it,
though, even with numerous
drops onto concrete.
Philip E.D. Wakeman
IEEE Member
Auckland, New Zealand
Your article reminded me of
two calculator-related incidents in my past. First: some
years ago, all the engineers in
my company were given free TI
calculators. I can't remember
the model number. Mine was
stolen, and I had to buy my own
replacement. Second: I won an
HP sweepstakes for subscribers to an HP engineering magazine. The prize was the HP 19B,
a business calculator!
You can't win for losing.

Bob Schuchman
IEEE Life Member
San Diego

correcTion

In the October issue, the voltages
were reversed in the "Quantum
Tunneling Creates Fast Lane for
Wireless" diagrams.

Letters do not represent opinions
of the IEEE. Short, concise letters
are preferred. They may be edited
for space and clarity. Additional letters are available online in "And more
Forum" at http://www.spectrum.ieee.
org. Write to: Forum, IEEE Spectrum,
3 Park Ave., 17th floor, New York, NY
10016-5997, U.S.A.; fax, +1 212 419
7570; e-mail, n.hantman@ieee.org.

www.spectrum.ieee.org

http://www.spectrum.ieee http://www.spectrum.ieee.org

Table of Contents for the Digital Edition of IEEE Spectrum December, 2007