From: jpbonsen@Athena.MIT.EDU
Message-Id: <9304150301.AA17265@eos-17.MIT.EDU>
To: rdshydur@Athena.MIT.EDU
Date: Wed, 14 Apr 93 23:01:00 EDT


------- Forwarded Message

Received: by access.digex.com id AA27114
  (5.65b/IDA-1.4.3 for kfl); Fri, 18 Dec 92 21:19:09 -0500
Date: Fri, 18 Dec 92 21:19:09 -0500
From: "Keith F. Lynch" <kfl>
Message-Id: <9212190219.AA27114@access.digex.com>
To: UC482529@MIZZOU1.missouri.edu
Subject: Re: Welcome back to the Net...
Cc: kfl
Status: R

> Subject: Welcome back to the Net...

Thanks.  But I've never been off the net for more than a couple weeks
at a time in over six years.  Or for over a month at a time in the
past thirteen years.  It's hard to imagine what my life would be like
without it.

The recent direct TCP/IP connection of Digex to the net hasn't made
much difference to me.  I'm not willing to pay the extra I'd have to
to use FTP or telnet from Digex.  (I'm free to use them *to* Digex
from another site, but I've had no recourse to.)

One difference in my net usage from past years is that I now spend
most of my time reading netnews rather than mailing lists.  I haven't
unsubscribed to many lists.  Instead I download my mail unread, for
future review, as I have been for a couple years now.  Floppies are
cheap.

Of course I always scan my mail for instances of my own username,
which is how I spotted your message.

I have a very strong sense of history, and a great attachment to a
great many things.  I almost never throw anything away, unless it's
something I discard at once (e.g. banana peels, junk mail).  I still
have my 3rd grade notebook, twenty year old news magazines, paperback
books too awful to ever re-read, long-expired book catalogs, manuals
to a computer that the place I no longer work at no longer has, e-mail
dating back over 18 years, every message you or I ever sent each other,
all my remail, broken TVs I rescued from the trash decades ago and
have been meaning to repair, all my convention badges, core memories,
platters from crashed disks, old calendars, and much more.  Eric
Raymond described my apartment as a museum.

I wouldn't be surprised if this message (among many others) makes it
intact to the end of time.

I did eventually donate all my too-large clothes to charity.  Some
things are obviously of no further use.

I'm recently "on the net" in another sense.  A couple weeks ago, I
plugged our computer room ethernet into a router, thus connecting
to the One True Net.  It doesn't really work right yet.  I can't
send mail.  I can telnet to NIC, but not to any non-Milnet hosts.
SCO Unix seems to be very intolerant of large nets.  Or maybe I just
don't understand TCP/IP as well as someone who's used it as much as
I have ought to.

Our computer-room net is kind of wild.  It includes four different
kinds of ethernet, and, on all of them, four different kinds of
transaction (TCP/IP, DECnet, LAT, and DDP).

> How've you been?

I refuse to answer on the grounds that the authorities may be
listening, and if they had any inkling of how good I feel, they'd
certainly make it a class A felony.

My continued perfect health is a source of amazement to my coworkers,
who had assured my I'd come down with all sorts of things once I
started working in hospitals.  I do have a touch of sun/wind/rain/
snow burn on my face.  Not surprising considering I've biked over
6000 miles this year, at high speed, in all weather.

They say that if you want to have any respect for laws or sausages,
you should watch neither being made.  Well, the same is true of medical
care.  After seeing it from the inside, via my job in medical quality
assurance software, I intend to stay as healthy as I can.  Especially
after Clinton manages to gut the last vestiges of a free market in this
vital field.  Now is a good time to invest in gold, silver, seat belts,
helmets, and smoke detectors.

At a con, I asked a woman with a diabetes kit to test my blood sugar
after I pigged out on tons of sugary food.  My pulse was more than
twice normal, but my blood sugar was exactly the same as when I haven't
eaten in hours.  That's not supposed to happen.  I don't know what
it means.

I think it would be interesting to run dozens of harmless medical tests
on myself.  I really do think my metabolism is seriously different from
average, though not in a direction I've ever heard of.

> I've been doing reasonably well, myself.  Still eating, breathing.

That's what counts.  Also check for pulse and respiration.  Are you
still eating nothing but corn flakes and carrots?  (I'm kind of curious
whether one can survive such a limited regimen for as many years as I
had conjectured.)

> ... job as an Internet applications specialist for Campus Computing.

Please tell me more.  Have you done anything with SCO Unix?

> Is your job still keeping you busy?

Yes.  1992 has been a hectic year.  Our software has worked on two
platforms, DEC PDP-11 and DEC VAX/VMS, running two subtly different
versions of DEC MUMPS.  This year, I and one other person were
tasked to port our quarter million line application to SCO Unix
and Micronetics MUMPS.  (Keep in mind that MUMPS is so succinct that
I wrote a Huffman decompression routine in three lines, Conway's LIFE
in two lines, and uudecode, uuencode, and rot13 in one line each.)

The task was rather like rewriting all the works of Shakespeare in
modern English, without sacrificing the slightest iota of meaning
or nuance.  Better yet, like doing so while Shakespeare was still
constantly revising his work -- AQCESS is a moving target.  (It has
gone through seven revisions this year alone.)  Or rather, like
doing so while also leaving it readable to his original audience.
AQCESS has to continue working on PDP and VAX.

That was the easy part.

The Unix platform we chose was a *network* of Everex 40486 machines,
fed by a bunch of Xyplex terminal/printer servers.  Thus we had to
rewrite the system to support parallel processing.  And we had to
figure out how get the servers, which have no non-volatile memory, to
load their parameters from the Unix host at boot time.  (When I figured
that part out, both SCO and Xyplex had me fax them my solution!)

Before that, I had to get various peripherals working.  The ethernet
cards weren't supported.  The version of SCO Unix we had didn't have
bootp.  The disk controller wouldn't work with the disks we had.  The
SCSI address jumpers were undocumented on the Exabyte and Wangtek tape
drives.  One Wangtek silenty destroyed any tape fed to it.  A Mountain
tape drive, we ended up giving up on.

Instead of dealing with one company, we are now dealing with fifteen!

My coworker, Mark, is amazing.  He knows more about AQCESS than I would
have thought humanly possible.  It sometimes seems that every one of
those constantly changing 250,000 lines of MUMPS code is permanently
etched into his memory, along with its relationships with all other
lines.  Without him, I wouldn't have stood a chance.

Unfortunately, that's *all* he knows.  He doesn't know Unix, he doesn't
know VMS, he never *heard* of CP/M, the DEC-10, the net, or 99% of what
you and I and the rest of hacker/net/SF-con people takes for granted
as a common cultural heritage.  He's gay, and *I* know more about the
gay scene than he does!  He apparently has no interests outside work.
He doesn't read for pleasure.  At all.

I guess if I spent as much time studying AQCESS as I instead spent
studying physics, math, history, geography, chemistry, biology,
astronomy, medicine, nutrition, geology, law, electronics, operating
systems, networks, psychology, objectivism, economics, and politics,
and also threw in all the time I spent on the net, and all the time
I spent reading fiction, attending cons, commuting by bike, and
watching _The Simpsons_, I might be an AQCESS wizard too.

One of the biggest problems has been the DEERS link.  DEERS is an
interactive service provided over a bisync (!) line.  This has
previously been handled by an external protocol converter, which
our system communicates with in serial full duplex async ASCII.
We want to get away from the protocol converters, since they cost
over $6000 each, need expensive PROM revisions whenever there's a
DEERS software change, and are so unreliable that every site has
*two* of them, one as a backup.  It was decided that the 486 sites
would do this in software, using an internal bisync board.  I was
tasked to develop the software.

It would have been easier except for several minor details.  The
board cannot be directly written to or read from in MUMPS, but can
only be addressed from special routines callable from C or another
compiled language.  We didn't have C or any other non-interpreted
language.  (We ended up buying C.)  The protocol converter was totally
undocumented, and I had to carefully watch exactly what it was doing,
frame hypothyses, and perform experiments, to learn just what was
inside the (literal) black box.

Bisync itself was a major learning experience, with its mysterious
formatted and unformatted screens, attribute bytes, protected fields,
polling, keyboard lock, and wait states.

I had the C program communicate with AQCESS over a pair of named
pipes, which look to Micronetics MUMPS just like serial ports.
After weeks of work and testing, it was done.

Then I was told that it was going to be a different brand of card,
and I had to start over.  EDS had known this months ago, but hadn't
bothered to tell us.

Digression:  I work for NDC.  NDC wrote and supports AQCESS.  Somehow,
NDC's contract for doing so was won by EDS three years ago.  EDS then
promptly subcontracted it back to us.  We work closely with EDS, which
mostly does things like checking the inventory numbers on everything,
or holding pep-rally style meetings full of buzz-words signifying
nothing, and presenting awards to managers.  They are also hot on
the appearance of security.  Not the substance, just the appearance.
Last year, they installed locks on most of the doors in the building
where I work in Alexandria.  The locks require that one inserts one's
badge, and then keys a combination unique to that badge.  Neither a
badge nor a combination would let someone get through a door, unless
they had both, and the two that go together.  Just like ATM machines.
Also, not all doors respond to all badges.

These locks frequently refuse to open when they should.  Many of us
carry screwdrivers, to jimmy the door open, which only takes a few
seconds.

Last summer, when my badge started falling apart, I biked to EDS's
Herndon office near Dulles Airport, twenty miles from Alexandria,
one Saturday, to get a new badge.  I had never been there before.
(My original badge had been made in Alexandria, but they don't do
that there anymore.)

There was an immense steel gate across the road, topped with spikes and
what appeared to be electrified wire.  There was a guard shack there.
Nobody was in the shack.  There was no evident way to ring a bell or
get anyone's attention.  So I crawled under the gate, dragged my bike
under, and cycled up to the main building a quarter mile (!) past the
gate.  I let myself in with my card, and found myself in an immense
apparently empty building.  I biked up and down the long empty hallways.
I found the front desk, with dozens of video monitors, and a desk
like the bridge of Star Trek's _Enterprise_, facing an enormous
glass entrance-way decorated with thousands of eagles painted on.
Nobody was there.

Eventually I found the security office, with one person in it.  He made
me a new badge, even though I had no ID except my old badge to prove
who I was, my picture had come off the old badge, and my name wasn't
in any EDS directory.  And even though it meant leaving the security
office untended.  He was going to put my SS number on my badge, until
I showed him that my old one merely said "SUBCONTRACTOR" in the SSN
field.  EDS employees are required to wear their SSN around their necks
at all time while at work.

I wasn't surprised that I didn't see anyone else there.  I sometimes
have to work evenings and weekends, and I've noticed that there are
always NDC people around, but EDSers are purely 9 to 5.

One of the few "technical" functions that is purely EDS is the DEERS
support center, just down the hall from NDC's AQCESS support.  Our
support center has an 800 number, and is open 24 hours a day, even on
Christmas.  Any problem our customer support representatives can't
handle gets referred, by beeper if urgent enough, to a programmer,
and if they can't handle it, to Mark or I.  (He has a beeper on odd
numbered months, I have it on even numbered months.)  They received
over 30,000 calls since I started work at NDC two years ago.

EDS's DEERS support couldn't be more different.  They only thing can
do is bring a DEERS line back up.  This is necessary because these
bisync lines keep dropping, whenever there's a burst of noise on the
line, or the protocol converter loses power or needs to be reset.
So there are several people there all day, doing nothing but answering
calls on their non-800 number from all over the country, bringing
DEERS lines back up for people.  Except at night, or on weekends and
holidays, when if your DEERS line drops, it stays down!

EDS was founded by Ross Perot.  It's no longer owned by him, but it
continues to have the same corporate culture.  A very strict dress
code, mandatory random urine tests, etc.  I would be very surprised
if any NDC people at our location voted for Perot.  I actually believe
he might have been an even worse choice than Bush or Clinton.  It's
common knowledge, at least at NDC, that EDS got its first real break,
back in the 60s, when it won a lucrative contract for Medicare in
Pennsylvania.  Perot himself was on the Pennsylvania board that
awarded the contract!  I don't understand how he got away with it,
how EDS won NDC's contract away from it, or how Perot managed to be
taken so seriously by reporters during the recent election, while the
Libertarian candidate was totally ignored, as usual, even to the extent
of fudging the numbers during reporting of the returns so that votes
for the "big three" falsely add up to 100%.  Sigh.

Getting back to my story, I attempted to rewrite the code for the new
board, only to discover that the EDS bean-counters had been fooled into
thinking this was a lower bid.  It wouldn't work without three pieces
of software, each sold seperately.  EDS had bought one of these three.
And had bought another, useless, piece.  After we got all the pieces,
(licensed for N sites, where N is very large), I was able to rewrite
the code and get it working.

It's still been a lot of trouble.  It failed once a minute.  After a lot
of study, I get it to the point where it failed once an hour.  Then to
the point where it still failed once an hour, but at least recovered by
itself.  Now, I have it to the point where it fails only about once a
day, and recovers by itself.  A bigger problem has been why it was so
much slower than the protocol converter, according to timing in AQCESS.
The DEERS people were totally unable to give me timings from their end.
It was fast in Alexandria, but slow onsite.  This was not due to the
heavier load onsite, since it was equally slow onsite in the middle of
the night with very light usage, and very fast in Alexandria with a
heavy simulated load.

A bisync consultant was hired, and worked with me for most of a day,
studying manuals and my C code.  Finally, he suggested a four character
(!) change in my program.  This speeded it up slightly, but not enough.

Finally, while trying to track down a completely unrelated problem with
an ethernet card, I discovered that some jumpers were set differently on
some of the bisync cards.  Oddly, instead of making the card not work,
an incorrect jumper merely slows it down.  This will soon be corrected
onsite.  (Addendum:  This was corrected at one site today.  Seems to
be working fine so far.)

There are currently two 486 sites, each with a suite of four 486
Everex machines networked together, along with four terminal servers.
One is in Tampa, Florida.  I, Mark, and two other NDC people went there
in October to install it, just a month after my attending the World
SF convention in nearby Orlando!  I knew four years ago that I'd be
attending that convention, but hadn't even heard of NDC, and 40486
chips hadn't been invented yet.  Curiously, the second site, which we
same four people installed a month later, was near Chicago, which was
where the previous year's World SF con was held!  Thus the two out-of-
town places that NDC sent me this year were close to the two most
recent out-of-town places that I had gone to myself.

Speaking of cons, I didn't see you at the con in Orlando.  Will you be
at next year's Worldcon, in San Francisco?

Eric Tiedemann wasn't there either, though he had signed up to attend.
I haven't seen him since Chicon.  I haven't seen Peter Glaskowsky in
even longer.

BTW, I looked at that _Scientific American_ ad you mentioned to me.
I can't make any sense of it either.

> ... will you be restarting your remail list?

Not at this time.

> My main mailing-list-reading activity this semester has been Libernet
> and Extropians.  I noticed that Eric S. Raymond has recently shown up
> on the latter.  If you have the free time, I suggest checking it out;
> I find it quite rewarding.

I am on Libernet, but most of it I file away for future perusal.

I was on Extropians briefly.  But the volume was very high, and they had
a redistribution policy that I didn't like, since I normally share all
my mail with my brother, and wish to reserve the right to show it to
others later.  (I'll seperately store any messages to me flagged as
private, but doing so for all of a high-volume mailing list gets
ridiculous.)  Also, it struck me as somewhat cliquish, for several
reasons, including the dates on the messages.  Why 372?  What
happened in 1620 that was so important?  Anyhow, aren't extropians
all anarchists, by definition?  I'm not convinced that anarchy is
viable, so I guess I'm not really an extropian.

I do read all messages on sci.cryonics and sci.nanotech.  Those are
low volume, high quality newsgroups.

I should have a little more time over the coming weeks.  I keep track
of the hours I work.  I don't get paid overtime, instead I'm supposed
to be able to take off "comp time," to compensate.  Also, I get a day
of comp time for every month I have the beeper.  Plus, my accumulated
vacation is almost to the point where it stops accruing, i.e. "use
it or lose it".  I'm asking for a day off every week until spring.
(I don't like winter commuting.)  I'm not sure if I'll get it,
however.  You see, EDS did it again.  They screwed up the MUMPS bid.
Intersystems succesfully challenged the bid award to Micronetics, and
we're currently involved in redoing everything for Intersystems MUMPS!
Sigh.  Maybe after we finish that, they'll decide not to use SCO Unix,
or not to use 486 machines.

It's frustrating at times, but it's worth it, having a boss who is on
my side, rather than, as at my old job, one who always sided with the
other guy.  It's still a weird feeling, though.  I say "maybe we should
try setting foo to 3," and suddenly everyone's running around saying
"Mr. Lynch says foo must be set to 3".  It's just the opposite from
SAIC, where if I said it was dark and clear out, everyone put on their
sunglasses and got their umbrellas out.

I do often feel like I'm thrashing, completely at a loss, mystified,
and baffled.  But I try one thing after another, try silly things
when sensible things don't pan out, obsessively experiment and collect
data, and almost never give up.  Sometimes I even read manuals.  As
painfully slow and inefficient as this is, it does seem to be more
constructive than what most people do.  Most people seem to follow set
algorithms, either memorized, or written in front of them.  And when
something doesn't go completely according to plan, they're as helpless
as ants placed around the rim of a bowl, which will follow each other
round and round until they starve.  As Nancy Lebovitz put it, it
sometimes seems like one is nearly alone in the world, that most of
what look like people are actually non-player-characters (a gaming
term for what look and act something like people but are only
simulated, to provide background for the real players).

I'm writing this on one of my two terminals on my desk.  I volunteered
to stay here in Alexandria while everyone else was at a Christmas party.
I've pretty much got the place to myself.  Earlier, I was writing this
on my PC at home.  I can quickly move it back and forth.  Ironically,
it's gotten to the point where I can compress, uncompress, ship thrice
around the world, and store in a dozen places lots of long interesting
files, but finding time to actually *read* these files is a little
harder.

Even when I'm not alone, I have the place largely to myself, at least
in the online sense.  I have hundreds of meg of personal files on the
company VAXes, and nobody's even noticed.  There's just so much space
out there.  And the 486 systems are even larger.  Each suite has over
7000 meg of disk space!  This is because the AQCESS databases are
very large.  The whole computer room probably has about 40,000 meg.
Possibly twice that, as there's lots of stuff in there I'm not
familiar with.

Now I'm at home again.  A local radio station is playing all nine
Beethoven symphonies, in honor of his birthday.  He would have been
222 today.  (Actually, his birthday was two days ago, the 16th.)

I have been posting to various newsgroups.  However, it would be very
time consuming to go through all my stored stuff and fish out my
postings.  Someday, I'll index all that stuff.

I've heard that one can get all of usenet on CD-ROM, with new CDs every
month.  And, of course, nothing ever expires.  This is very attractive
to me.  Especially if they have some kind of decent indexing scheme, so
I can quickly locate all postings from a particular user.

Much better indexed is our call-in database at work.  If you're
sufficiently desperate for text from me, it wouldn't be difficult for
me to pull out all the notes I've ever made in call-ins and forward
them to you.

I recently bought Feynman's complete _Lectures on Physics_.  $60, even
in paperback, but what the heck.  Why should I keep saving more than
half my salary, when dollars will be pretty worthless in a few years.
These books are excellent, and have already taught me a few things I
thought I already knew, as well as showing me new ways to see why other
things work the way they do.

Coincidentally, I then happened to read a Stephen Jay Gould column in a
recent _Natural History_ magazine, in which he criticizes Feynman for
working everything out from scratch rather than for trusting others.
(For what it's worth, this was a small part of a column devoted mostly
to criticizing Columbus.  Sometimes I think SJG got up on the wrong
side of the planet this lifetime.)  In my opinion, that's a major part
of the attraction of Feynman's work.  It's all first hand.  Not half
understood, half right, half obsolete, endlessly recycled pablum, which
is what makes up most textbooks.

Even more recently I bought _A Season for Slaughter_, the fourth book
in David Gerrold's "War with the Chtorr" series.  This is without a
doubt the best alien invasion story ever.  Thousands of pages in,
it's still not clear to *anyone* whether any of the alien species are
intelligent, and only now is anyone coming up with plausible guesses
as to how the aliens may have arrived.

My only argument with the series is with the human psychology.  I
discussed this at some length with DG at Magicon in Orlando.  Neither
of us convinced the other.  In fact, he doesn't believe my self-reports
of my own psychology.  In his defense, I must confess I've never
been in nearly as stressful an environment as his protagonists, who
are fighting what appears to be a losing battle against the certain
extinction of all native Earth life, including human beings.  I can't
honestly say what I would do in that circumstance, though his books
have given me dreams in which I honestly believed I was in that world.

I also chatted with Spider Robinson, Kelly Freas, and James White.
I took the latter to task for his _Deadly Litter_, in which trash
discarded from interplanetary spaceships is a major hazard to
navigation.  He just has no sense of scale.  Such a collision would
indeed be deadly, but is vastly less likely than being hit by a
meteor in your backyard, even if millions of spacecraft were to
dump rubbish non-stop for millions of years.

> uc482529@mizzou1.missouri.edu, also ccgarcia@mizzou1.missouri.edu

Why ccgarcia, when your name is Tony?

I bought a new bike this summer.  21 speeds.  The old 10 speed (later
upgraded to 12) wore out after about 8000 miles.  Almost every week
I was getting flat tires, broken spokes, snapped brake cables, or a
broken derailleur.  Every five minutes the chain would come off either
the front or back -- occasionally both, and end up lying in the road.
Finally, the bearings went, and it felt like riding a garbage disposal
with a full set of silverware jammed in it.  So I bought a new one.
In the 3000 miles I've ridden it, I've had no problems at all.  No
flats, no busted spokes, nothing.  The chain did come off once, during
a ride on a surface so rough that my glasses flew off and my keys
hurtled out of my pocket.

Sorry if this seemed somewhat incoherent.  Instead of starting at
the top and writing until I reached the bottom, I generally went
back and interjected stuff wherever it seemed to fit best.  And
I'm not going to take the time to proofread it, or pretty it up.
If it's like most of what I write, there are no typos in it *now*
anyway, but a few will materialize over the next few weeks or
months. :-)

As always, I eagerly await your reply.

Keith Lynch, kfl@access.digex.com


%%% overflow headers %%%
Cc: 76060.1103@compuserve.com, C451976@umcvmb.bitnet, C503024@umcvmb.bitnet,
        Raymie@ATHENA.MIT.EDU, UC445585@umcvmb.bitnet, amck@emx.cc.utexas.edu,
        bagwell@CAF.MIT.EDU, consp10@bingvaxu.cc.binghamton.edu,
        dennis@cs.washington.edu, eafu011@orion.oac.uci.edu, eli@ima.isc.com,
        eric@snark.thyrsus.com, est@cs.nyu.edu, jpbonsen@ATHENA.MIT.EDU,
        maler@decvax.dec.com, mbr@larch.lcs.mit.edu,
        pglaskow%telebit@uunet.uu.net, uvfowl@uncecs.edu,
        wb8y@vax5.cit.cornell.edu, wisner@hayes.fai.alaska.edu
%%% end overflow headers %%%

------- End of Forwarded Message