. 12
( 41 .)


stitutions. Quite a lot of additional processing power was potentially
available from similar sources, and university systems could run DES-
CHALL as is, without any additional modi¬cations needed for things
like working through ¬rewalls.
To address the needs of the computer lab managers who wanted to
participate with all of the machines at their disposal, we would need
to provide the software not only to test keys, but to manage the key-
Organizing DESCHALL 73

testing software automatically on the dozens or hundreds of machines
that a lab would contain. Making these processes automated would
help the lab managers contribute a great deal of computational power
without a lot of time and e¬ort.
Something else that would help to inspire lab managers to contribute
their systems™ idle cycles would be to let them see just how much pro-
cessing power they had at their disposal. If we let them see how many
keys they were testing, we might ¬nd that various lab managers would
start to compare their results with others, using some friendly rivalry
to induce them to throw still more processing power at DESCHALL.
At this point in the project, we had searched less than two tenths of
one percent of the available key space, which despite being a long way
from the conclusion, was far ahead of the other e¬orts. The progress
of all three public e¬orts is summarized in Table 2. Jillson sent a note
to the DESCHALL mailing list with his observations on our relative
standing in the contest.

Keyspace Millions of Keys Average Time
Group Completed Tested per second to Find Key
DESCHALL 0.162% 165.000 7 years
SolNET 0.013% 50.704 23 years
DES Violation 0.058% 43.000 27 years
Table 2. Status of DES Challenge Groups

Shortly after Jillson wrote his thoughts to the DESCHALL mailing
list, I read his message and re¬‚ected on our standing. Not having seen
any tra¬c on the European DES-Challenge mailing lists for about two
weeks, it was becoming clear to me that the group that won RSA™s
48-bit RC5 contest was simply not going to write the software needed
to solve RSA™s DES Challenge.
Thinking about the comparison among the three active public
projects, I was pretty happy to be working with DESCHALL. We were
clearly the front-runner in the contest, and our rate of 165 million keys
per second seemed impressive enough. It certainly sounds like a lot
of keys. A more useful, and sobering statistic, was the time needed
to ¬nd the right key at our current testing rate”seven years. If we
were to succeed, we needed to get a lot more people to participate in
Needle in a Haystack

Thursday, April 3, 4:01 P.M.
The Ohio State University, Columbus, Ohio

Justin Dolske sat in his lab and and tried to think of ways that he could
encourage more people to participate in DESCHALL. Dolske knew that
we needed to ¬nd a way to describe the nature of the problem we were
Looking for the right key out of all possible DES keys is a big
job, basically the equivalent searching for a needle in a haystack of 72
quadrillion strands. Dolske grinned as the obvious question presented
itself: “How big is the haystack?”
“Figure that a strand of hay is a cylinder ten centimeters long and
two millimeters wide,” typed Dolske. “Then assume that the hay is
packed densely. Finally, let™s assume that a haystack is shaped roughly
like a sphere cut in half. After crunching the numbers we see that our
haystack is roughly two and a half miles wide and over a mile high.”
Dolske hit the “send” button and shot a copy of his observation to
the DESCHALL mailing list for other participants to see.

Friday, April 4
Carnegie Mellon University, Pittsburgh, Pennsylvania

CMU graduate student Bridget Spitznagel updated her Web document,
Frequently Asked Questions about DESCHALL and the DESCHALL


e¬ort at CMU. She thought putting the magnitude of our problem in
monetary terms might be fun.
For her calculations, Spitznagel assumed that U.S. paper currency
bills are six inches wide, two and a half inches tall, and one one-
hundredth of an inch thick. If each possible key were worth a penny,
the entire key space would amount to one square mile of $100 bills that
was twenty-two feet thick.
Put another way, if potential keys were pennies, we™d be looking for
one penny out of over $720,575,900,000,000 worth of pennies.

Friday, April 18, 5:01 P.M.
Virginia Polytechnic Institute, Blacksburg, Virginia

After reading some of the messages posted to the DESCHALL mailing
list two weeks earlier, computer science undergraduate student Alex
Bischo¬ started thinking about keys. He wondered, “What if cryp-
tographic keys were like keys for door and car locks?” An image of
mountains of little metal keys suddenly popped into his head.
Then he started some calculations. Assuming that such keys are two
inches long, if you laid them end to end, you™d have a line of keys long
enough to circle the sun 3894 times.
Visualizing just how many combinations we would need to try gave
us pause. To try so many possible keys, our DES key-cracking system
was going to need a lot more clients, since they would be doing the real
work of the DESCHALL project.
Spreading the Word

Tuesday, April 1, 10:15 P.M.
Megasoft Online, Columbus Ohio

I continued thinking about the challenge before us after reading Car-
leton Jillson™s message. If the way to defeat DES was to get more key-
cracking clients running, we needed to let a lot more people know about
the DESCHALL project and to convince them to run our client soft-
ware. We had to ¬nd the right people and we needed a compelling
message to get their attention.
Building on that initial awareness would be the hard part. We were
all pretty sure that once things got started, we could get some critical
mass of participants and then wait for one of the clients to ¬nd the right
key. We didn™t know just what would constitute critical mass, but we
knew that we were nowhere near it. At the rate we were going, we would
take eight years to ¬nd a DES key. We needed thousands of clients”
that would mean hundreds or even thousands of new participants.
To bring our message to a large number of people, we looked at
the media, with particular emphasis on the news outlets that were re-
porting on computing technology. Early conversations with writers in
the media were helpful. Once they understood what we were doing and
why anyone would want to ¬nd DES keys, they often expressed inter-
est in our project and wanted to be advised in the event of any major
milestone (in particular, once someone found the right key). Through
those conversations we learned that we didn™t have time to educate peo-
ple about cryptography, how DES was used, and cryptographic export
policy. Reporters need to know what happened so they can give their


readers the facts. We quickly learned to adapt our message to get their
attention ¬rst and to ¬ll in the details afterward. A typical story pitch
might go something like, “The government standard for cryptography,
used to protect the nation™s ¬nancial systems is vulnerable to attack.
I think your readers might like to know how a group of researchers,
engineers, and students are using their computers to demonstrate how
weak it is.” With that as a basis, many reporters would want to hear
Not all DESCHALL participants were talking to reporters, though.
Some of us were simply looking for ways to raise awareness among
people we encountered in our daily online activities. Many of the DES-
CHALL participants were active on a system called Usenet. Usenet
works much like e-mail, except that instead of being a one-to-one com-
munication mechanism, Usenet is many-to-many. Instead of writing an
article and addressing it to a person, authors will address it to a news-
group, and servers all around the world will carry that article in that
newsgroup. Thus, people all over the world with similar interests can
read articles that people have written and post their own for others
to read. Usenet would prove to be an e¬ective way for DESCHALL
participants to draw attention to the project.
Signature blocks have long been a part of Usenet articles and e-mail.
The basic idea is to de¬ne some block of text that will be automatically
appended to your message, rather than making you retype your name
on each message. Before long, people started adding more information
to the signature block, including contact information, thus creating
a sort of virtual business card. Pithy remarks were also included on
occasion, and some people even went so far as to create huge signature
blocks, with gaudy pictures made out of text characters, advocating a
dozen di¬erent causes. Taken to this extreme, signature blocks could
become the electronic equivalent of the bumper sticker.
Before long, messages showing up on the DESCHALL mailing list
were carrying signature blocks that advertised the project or provided a
link to the project Web site. As DESCHALL project participants went
about their business, their signature blocks advertising DESCHALL
started to spread. Usenet newsgroups, e-mail lists, and private corre-
spondence became graced with mentions of and links to DESCHALL,
usually with a simple tag like “Crack DES Now!” (Although we weren™t
technically attempting to crack DES itself”we were trying to crack a
DES-encrypted message”our experiences with the media helped us to
Spreading the Word 79

understand that opening with a long technical digression would not
catch and hold the reader™s attention. Brevity rules.)
Likewise, on their personal Web sites, participants began to describe
the project and their e¬orts to advance it. Invitations to join the project
were often extended on such Web pages. Oregon State University engi-
neering student Adam Haberlach and I made small graphical buttons
fashioned after the “Netscape Now!” buttons that graced so many Web
pages in 1997. In a problem akin to having a cupholder with no car to
put it in, the European DES-Challenge group that never made any soft-
ware had created a Web site and graphics. One particularly common
graphic was a “Crack DES Now!” button that came from that group.
Justin Dolske commandeered that button and put a copy on his Web
site for others to use. Since the European DES-Challenge e¬ort had no
software, it didn™t seem that they would need the promotional graphic
Dolske didn™t really have time to try to create new graphics of his
own. He had been drafting a “call for participation” document with a
brief description of the project and its purpose which was aimed at the
technically inclined who would be most likely to understand the project
without any explanation. Dolske™s call was posted to Usenet where it
would be seen by others involved in cryptography.
The increasing mentions of DES and DESCHALL online helped
us recruit new participants who, in turn, encouraged others to join

Thursday, April 3, 2:30 P.M.
Megasoft Online, Columbus, Ohio

A critical aspect of the promotional e¬ort was to stress the importance
of the DESCHALL project to others who weren™t cryptographers and
might not even use computers much themselves. To ¬nd a way to relate
DES security to the concerns of a typical American citizen, I called my
own bank, KeyBank, introduced myself as a computer scientist work-
ing on a security research project, and asked to speak with someone in
the bank™s information security group. The person who answered the
phone took my name and number, promising to have someone call me
back. Shortly thereafter, my call was returned, and the bank represen-
tative and I engaged in an interesting discussion about cryptography,
speci¬cally the use of DES. Although the bank o¬cial did not want

to share details of how DES was used in the banking industry, he was
willing to verify for the record certain vague statements like “DES is
heavily used in the ¬nancial sector.” He expressed serious interest in
the project, wished us success, and said that he would be watching our
progress from “a safe distance.”

Tuesday, April 8, 7:22 A.M.
Loveland, Colorado

Among the hats that Rocke Verser wore throughout the day was that
of editor. Justin Dolske and I worked with Verser to create a press
release that would help more DESCHALL participants to talk to the
media with con¬dence. Draft after draft, the press release got improved.
Finally, Justin Dolske, Rocke Verser, and I had something we were
reasonably happy to share with the rest of the project participants.
Many of the newcomers to the project were very enthusiastic, but
did not have the kind of background in cryptography needed to frame
the discussion in the right context for reporters on their own. Part of
the motivation for our press release was to provide the less technical
participants with a simple fact sheet that would help them to make
the pitch to their local media outlets. Once the release was put on
my Web site and posted to the DESCHALL mailing list, participants
began calling local media, pitching a story about the project, with
a connection that would be of interest to local news organizations”
someone from the immediate community participating in a nation-wide
Hoping that if we addressed tech-savvy media would help us ¬nd
still more participants, I sent a draft of our press release to the tips
contact address at News.com.

1:30 P.M.
CNET Networks, San Francisco, California

Courtney Macavinta, a writer at CNET™s News.com found the an-
nouncement of the DESCHALL group™s formation of interest. Given


. 12
( 41 .)