Historical Basis:  www.USNA63.org
- being an accounting of some of your Hostmaster's long journey into cyberspace, forward to a documenting of Y2K preparations -

Daddy, what did you do in the CyberWars?

Our Hostmaster, Bill Kennedy, has a long and storied background as a bit butcher.  Lou Simpleman extracted the following account from BK early in the Year 2000 -

Lou:  Bill, I told my son that you probably were one of the first in the universe to get a domain name.  He wants to know when you got it.

Bill:  The memory fades, but I think I was in the first 100, probably ~86-87.  I _think_ wlk.com is older than IBM.COM, I'm quite sure it's younger than DEC.COM and ATT.COM.  At that time there were three guys doing it on a volunteer basis and the fee was $500 to register, $250/yr thereafter.  It actually wasn't exhorbitant because there was a lot of hand tuning that had to be done to keep things working.  There were only three root (top level) name servers and the Internet didn't exist as such.  There was, of course, arpanet, but it was connected by 56Kbps [sic] digital lines and 1200bps [sic] modems.  Mail wasn't handled on a domain name basis until some years later.  It traveled via UNIX uucp links and addressing was done by the destination machine name.  I had two "backbone" connections, cbosgd and rutgers.  My mail address was cbosgd!ssbn!bill or rutgers!ssbn!bill.  Each originating and destination site maintained address maps.  We specified site!user in our mail and the mailer mapped the modem path for us.  If I wanted to send something to Bob Tracy (none of you know him) I specified okccrsh!rct and the mailer routed it as petro!ihnp4!okstate!okccrsh!rct.  My machine made a modem call to the machine named petro who called ihnp4 and so on.  Delivery times were measured in days.  It was ~36 hours to get from here to Bob in Oklahoma City.  My machine only called petro (San Antonio) after 1800 and before 0700, his machine only called ihnp4 (AT&T near Chicago) twice a day, but once it got to the phone company, it went pretty fast.

Once arpanet interconnected to nsfnet (National Science Foundation) and the Domain Name Service (DNS) software got developed things started to speed up. The addresses stayed similar because there were so few Internet connected sites, but once you got to one, things traveled in a blur.  My nearest neighbor by then was the University of Texas in Austin.  To send to Bob by then was othello!cs.utexas.edu!okccrsh.att.com!rct.  Othello isn't the right name for my Austin neighbor, I've forgotten it, but once it got to UT the mailer rewrote the address to rct@okccrsh.att.com and forwarded it immediately to an AT&T gateway who sent it along to the Oklahoma City site without the intermediate stop at okstate.edu.  Cycle times shrank to just a few hours. When I became Internet connected in 1992 it wasn't unusual for mail to be delivered less than an hour after it was sent, but still not the 5-10 seconds commonplace today.

Lou:  Bill, some of that original equipment should be in Boston's Computer Museum, or in the Smithsonian - did any of it survive?

Bill:  The ssbn machine has long since been retired and doesn't even have a survivor machine.  The last of the original collection left here yesterday [21 January 2000].

Forgive the bulk, but it's a mildly amusing account.  I once had a luggable road warrior machine I used for my off-site programming contracts.  I was doing some work at a Sun Microsystems factory when the keyboard died.  I knew what was wrong and knew how to fix it but had no time.  We had a deadline just four hours away.  Since I was in Silicon Valley, I went to a computer retailer and bought another luggable road warrior (a 6MHz 80286 fire breather :-) and finished the job on time.  The funds I used had been budgeted for floor covering in the addition I had built onto the house.  The machine, therefore, was named "carpet".  It had numerous motherboard and chassis upgrades but kept the name.

Carpet was the only survivor of the old namespace until yesterday.  Since its duties had been assumed by a laptop, carpet was pretty much a surplus machine.  I used it to check things on the '63 web site, but that's about all it did.  Tim Harvey and Tamara were over for a visit, he had been having reliability problems with his laptop and thirsted for a desktop.  Carpet was a nearly perfect fit since it's transportable (has a carrying case and handle) with a built in LCD display.  You all would have enjoyed being a fly on the wall watching Bill and Tim getting Windows 98 installed in time for them to haul it back to Austin :-)  Each time we'd make some kind of blunder we'd ask each other "Where are the kids when you _really_ need them?".  I'm like a hog on ice with peecees, Tim's no better.  We got it done, carpet (whatever he names it) has a good home, an era in Kennedy computing has closed.  The first incarnation of carpet was 1990...  Yes, it was nee carpet.wlk.com :-)

Editor:  Look here for a summary of www.USNA63.org registration information, April 1999, and here is the actual Domain Registration Agreement.

Y2K Wrap-Up for USNA63.org
[Trevor cartoon linked fromthe Albuquerque Journal, 01-05-2000]
Bill Kennedy, '63                          webmaster@usna63.org

Bonafide Y2K Incident on Web Server
11th Hour Y2K Panic:  Hostmaster's Shopping Experience in Texas

Y2K Preparations for USNA63.org:
    Server Summary
    CERT Advisory on Y2K
        Expectations (malicious) during Y2K
        Year 2000 Computer Viruses and Hoaxes
        Y2K Status Reports

Bonafide Y2K Incident on Web Server
"...I have a weather station accessible via the local area network.  It records current conditions every hour on the hour and copies the observation to the web server, the link is on www.wlk.com.  A few minutes ago [on 1 January 2000] I noticed that the weather station thinks it's 100 years old today.

"The software that came with it is pretty bad, so naturally I muttered dark thoughts about the manufacturer.  I needed to go in and manually clear out year-to-date rainfall, so I fired up their stupid program and did so.  They also offered to set the date and time to agree with the peecee, so I let it do that too.  The report of current conditions over the LAN continued to show January 1, 1900, but their stupid program insisted that it's January 1, 2000.  Guess who wrote the program that collects the observation over the LAN?

"I went to line 521 in my program and changed it from:
        wx_date[2], wx_date[0] + 1900, time_buf);
to read:
        wx_date[2], wx_date[0] + 2000, time_buf);
recompiled it, and now the observation is correctly in the twentieth century.   [I'm of the opinion that we've just started the last year of the century.]

"There is self defense available.  The weather station itself only retains a two digit year.  Admittedly I should have sent myself a reminder to change the program along with the reminder to clear year-to-date rainfall, but I didn't.  Nonetheless, that's a confirmed Y2K bug sighting, the only one I've seen thus far.  I expect some at the end of February, but we appear to be out of the woods for early January."

Y2K Panic in San Antonio
[dateline:  Thursday 30 December 1999, San Antonio, Texas]

Panic Shopping in SA:   I needed to get denture fizz at WalMart and light shopping at the local supermarket on the way out and back from SA respectively.  I was not only astounded by the length of the checkout lines, but also floored by what people were buying.  There were heated conflicts going on in the WalMart aisles over the dwindling inventory of small electric generators.  Both stores had people with shopping carts laden with 5-10 gallon water containers and the very few bottles of water that had been swept from the shelves.  The HEB (supermarket chain) house brand water label clearly states that it's San Antonio city water but people couldn't get enough of it.  While I was locating the items I needed, the PA system announced that a shipment (18 wheeler trailerfull) of water had arrived from San Antonio and before I was out of the store it announced store apologies: it was gone.

My list had a can of Sweet Sue Chicken & Dumplings on it.  I keep a can of it and one of Dinty Moore Beef Stew in reserve for when I'm sodium starved.  The only can of Sweet Sue on the shelf had a rim dent that breached the can and there was no Dinty Moore (I didn't need one) to be seen.  It was as though locusts had been there, dry goods were severely depleted despite the efforts of the store to restock the shelves.

That's not the climax of my report.  I had only seven items, so I went to the 'ten or fewer' checkout aisle.  The guy in front of me and the guy behind me each had ten six gallon containers on their carts - I felt odd with some butter, onions, potatoes, and TV dinners in my basket.  My total came to $15.16.  I didn't have a ten, so I gave him a twenty, a five, a nickel, dime, and penny.  He put the money away, tore off the register tape, smiled and wished me a happy new year.  I pointed out that not only had I given him a twenty, but also he had keyed in that I had and that I was due $10 change.  He flushed and said he needed manager approval to 'correct the Y2K bug'!  I showed him the register tape where it was clear that my proffer and change were correct, but he would not produce the change until a manager was summoned to determine whether or not it was a Y2K bug.

ATM Run:  I needed cash in addition to the other SA items, so I went to the USAA bank to use the ATM.  It's nice being near San Antonio because USAA is in strangling, grenade, tactical missile, and apology range.  Guys, you wouldn't have believed this any more than you might believe the WalMart or supermarket stories.  The parking lot was full, cars were orbiting waiting for departures.  All eight drive through lanes were open and there were no fewer than four cars queued for each.  Both ATM lanes were open and I was able to squeeze between the fourth car in the rightmost lane and the curb to become the fourth car in the leftmost lane.  I wondered, aloud, if the lemmings were right and I wasn't,  but withdrew my usual $200 wallet replenishment and went on.  Driving home from the supermarket I tried to think it through.

I keep a couple of gallons of water on hand, but so does anyone who's on a well and has bat brains.  I have a 20KW emergency power plant, but it's three years old and not purchased for Y2K.  I'll confess I stopped by the liquor store and bought a little more bourbon than I otherwise would have, but that was because _he_ is on just-in-time inventory with _his_ vendors and could offer no assurance both their computer systems were immune to paralysis.  I sigh and shake my head at the guys ahead and behind me spending $75 each for water containers.  What a sad commentary it is that our press, for being too lazy to find and report the facts, has convinced the lemmings to run and shown them the way to the cliff.  Yeah, I'm that cynical.  Had you been with me today, you would be too even if you weren't twelve hours ago.

Webmaster reports:

The Truth About the New Millennium:  So what millennium are we in, anyway?  Given our background, how can we deny that the Naval Observatory's position re  the end and beginning of the millennia is accurate?  According to the U.S. Naval Observatory, the end of the second millennium and the beginning of the third will be reached on January 1, 2001. This date is based on a calendar created in 526 A.D. by Dennis the Diminutive, the head of a Roman monastery who forged a common calendar from the divergent dating systems of his day. Rather than starting with the year zero, the calendar begins with the date January 1, 1 A.D. ... on the other hand, I have a little less faith in their work now, since the only national/government area Y2K problem I heard about was, yes, at our Naval Observatory, where their Web site year rolled over from 1999 to 19100!  This embarrassing moment you can view via the screen captures made shortly after the roll-over, all captured at this collection of Y2K Misteaks.   The site includes several humorous Y2K situations, one of which is the Naval Observatory's site gaffe.  (and take a look at Case Western Reserve University's page...).

Here is my personal position:     I accept that the Romans worked without a year zero, but the argument is null and void once we went to arabic numerals on 'our' calendar.   By putting 2000 on the calendar rather than MM, we imply that zeros exist, and that 2000 years have gone by [thus two millenia] since the zero point that we are counting from, nominally the birth of Jesus of Nazareth.

        So not from a calendar-historical viewpoint, but from a calendar-empirical viewpoint, we are beginning the new millennium.   I captured the following definition to clarify for myself:

I received a nice summary e-mail on 4 January from my Y2K buddy Tony at our Credit Union: My response was to refer Tony to Ed Yourdon's 'backpedaling' -  Ed Yourdon admits there were no bank runs, riots in streets, chaos in general - but he still won't admit that the problem was solvable.   He even tiptoes around the suggestion, 'corporate cover-up'.

Y2K Preparations for USNA63.org
Also see:  CERT Advisory on Y2K

Timekeeping:  The operating system used by the web server, Sun Solaris 2.5.1  has natural immunity to 01/01/2000 because the fundamental time keeping is done in seconds since 00:00:00 GMT 01/01/1970 and the system clock operates in UTC (GMT). Our server has some additional immunity in that the system clock synchronizes to radio clocks from the NIST (National Institute for Science and Technology; formerly National Bureau of Standards) using WWVB and Naval Observatory using GPS.  That has no Y2K significance other than it can't get the wrong time.  Here is a display of the timekeeper, I'll briefly explain it:
Timekeeper Display for USNA63.org Web Server
   remote time source 

The currently selected time source is WWVB, twenty one microseconds fast ('offset') with a jitter ('disp') of a hundred ten microseconds.  The GPS clock [=REFCLK(30,1)] is a millisecond and a half slow with jitter of eighteen microseconds.  The other entries are for other time servers.  The wustl is the Naval Observatory GPS clock at Washington University in St Louis, the two dec entries are clepsydra.dec.com in Palo Alto, CA; atlas and zeus are local.  Note that "fast" and "slow" are not used in the same context as we might use for determining the error from the correct time.  They mean by how much the system clock disagrees with the time reported by the radio clock or neighbor time server corrected as much as possible for network propagation delay.  This has nothing whatsoever to do with Y2K other than to present the accuracy of timekeeping by the web server.

Y2K Problems:  Although the Solaris operating system is immune to the rollover and the system clock is accurate, that doesn't mean that Y2K problems don't exist.  There is a collection of utility routines that convert from the system clock time to calendar date and time for GMT and local timezone.  Programmers that wrote the myriad of support programs were often unaware of or too lazy to heed the documented behavior of the conversion routines.  The year is the number of years since 1900.  Thus, when 2000 arrives, it will change from 99 to 100.  A known non-compliant program on the web server is the software license manager.  It will log events as 12/31/99, then 01/01/100.  The software system owners have elected to not correct the condition.  This is the sort of thing we may expect by any remaining undiagnosed (or ignored) Y2K problems.

Server Summary:   Any Y2K problems that have not been found or fixed on the web server will not harm anything, they'll just look silly or stupid.  You may be certain that your hostmaster will be up early Saturday morning looking for things that appear silly or stupid.  This is the least of our concerns.

Embedded Systems:  The more dangerous potential problems will occur in unattended embedded systems.  Imagine, for example, the safety controller that is programmed to shut down an operation if an event doesn't occur within an interval.  When the year rolls a non-compliant system could mistakenly interpret the event in the prior year as having happened a hundred years before and initiate a safety shutdown.  Another only slightly more benign example might be traffic signals operating weekday schedules on weekends and vice versa.  These are the most likely consequences of Y2K and they won't be discovered until they are observed.

Site Summary:  The bottom line for the web site is that even in the event of such things as are forecast by the lunatic fringe, it won't crater.  The site's local power derives from the Bandera [Texas] Electric Cooperative, Inc. which wholesales its juice from the Lower Colorado River Authority.  The LCRA has had over half of its generating plants operating in calendar 2000 for the last three months.  [Their big concern is not what happens at midnight on 12/31 but rather the consequences of retarding the clocks 120 days in mid-January!]  But the operation of your web site is not dependent on local power:  There is adequate emergency generator power and fuel reserves for thirty days or more.  If the telephone company is up and running, so will be your web site.

Let me know if you have any questions.

CERT Advisory on Y2K

Year 2000 (Y2K) Information

   The CERT/CC has recently published web pages to assist sites in dealing with Y2K-related security issues.

Background on CERT:  The CERT (Computer Emergency Response Team) Coordination Center is part of the Survivable Systems Initiative at the Software Engineering Institute, a federally funded research and development center at Carnegie Mellon University. CERT was started by DARPA (the Defense Applied Research Projects Agency, part of the U.S. Department of Defense) in December 1988 after the Morris Worm incident crippled approximately 10% of all computers connected to the Internet.  Gene Spafford's crew at Purdue was responsible for the initial tooling and post mortem on the worm.  It cooperates extensively with law enforcement.  Their primary duty is to vaccinate before the fact and diagnose after the fact.  It's also a clearing house where vendors can report newly discovered vulnerabilities and publicize remedies.  Hackers (complimentary use of the term) also use it to report discoveries and suggest remedies.  They've been able to spot a virus embryo before it could spread and have distributed antivenom to stop the spread.  They've also developed some really effective evidence gathering technique that has put some criminals behind bars.

    Gene is one of the old timers of the net.  His security beginnings were when there was no Internet at all, just a loosely affiliated group of cooperating sites with 300bps [sic] modems.  I'm pretty sure the CERT was his idea, but if it wasn't, his DNA abounds in it.


   Y2K FAQ - Our Y2K FAQ has been extensively revised in collaboration
          with participants in the International Y2K workshop held in
          October 1999. The FAQ includes information to help sites
          determine whether a failure is Y2K related or an attack. The
          FAQ is available at


Expectations During Y2K - "Cyber Infrastructure and Malicious
          Expectations during the Y2K Transition Period," also a
          collaborative effort by members of the Threat Analysis Working
          Group at the International Y2K Workshop, discusses potential
          activities associated with Y2K and offers recommendations. This
          paper is available at

Year 2000 Computer Viruses and Hoaxes - This web page lists the
          approximate number of reports the CERT/CC has received on
          viruses and hoaxes that reference Y2K. There are also links to
          anti-virus vendors having Y2K virus and hoax web pages. Our
          list is available at

Y2K status reports - During the Y2K event, the CERT/CC will publish
          regular reports on our web site to inform the community of
          activity being reported to us by other response teams and
          sites. Those reports will be available at

posted:  58th Pearl Harbor Anniversary

historical update:
17 June 2000
 Navigate to: 
Plan of the Day