Date: Sat, 13 Apr 2002 23:45:20 -0400 (EDT)
From: "Keith F. Lynch" <kfl at keithlynch.net>
To: WSFAlist at keithlynch.net
Subject: [WSFA] Re: Netiquette
Reply-To: WSFA members <WSFAlist at keithlynch.net>

Steve Smith <sgs at aginc.net> wrote:

> One of the curiosities of the computer biz is that the old machines
> still do exactly the same thing they always did.

Until they break.  I am a little impressed that that 1991 keyboard is
still working.  I have a 1986 keyboard which I was using until last
year, but I had to dismantle, clean, and refurbish it a couple times.

Until recently, hard disks were very prone to head crashes.  Cable
connections are another trouble spot.  Chips are likely to walk right
out of their sockets after many cycles of warm-up and cool-down.
Electrolytic power supply capacitors sometimes short out, often
taking other components with them, and sometimes even causing a fire.
Screens fade and get images burned into them, and their filaments
burn out.  Dust gets everywhere, causing overheating.

> Only thing I miss is having it flag spelling errors as I type.

I don't like a program to take action until I'm done.  It's very
annoying to be warned while I'm in the middle of typing a line.

> For a bit of nostalgia, see
> http://www.bricklin.com/history/vcexecutable.htm
> This lets you download a copy of the first IBM-PC version of Visicalc.

I'm not a fan of Visicalc.  The APPLE ][ version of that program took
our data hostage.  There wasn't any easy way to get it out again in a
form where we could write our own programs to process it, or to use
other company's programs on it.  That's one sin that even Microsoft
is innocent of, as far as I know.

> It's under 28K.

28K is plenty for any reasonable application.  If you look inside
most gigantic executables, you'll find that they're mostly NULs,
i.e. totally wasted space.

> Actually, I've herd that one of the problems that the library types
> have is with old digital formats and media.  How long will 8" floppy
> disks *really* last?

There are really three issues here:

* The physical medium itself.  Old tapes, disks, cards, and even
  papers eventually become unreadable.  Especially if stored in poor
  conditions, e.g. where water and vermin can get to them.

* The machines that read the medium.  This is less of an issue, since
  new machines can be built.  (Or in some cases, emulated in software.
  For instance a program could easily be written to read data on
  punched cards or punched paper tape from scanned-in images.)

* Knowledge of the formats.  Some JPL space probe data tapes from
  the 70s are perfectly readable, except that nobody knows the
  interpretation of the numbers on them.

As a notorious archivist, I would really enjoy working on all three
problems.  I was enthralled to read about recovering data from PCs
crushed in the WTC collapse.

For Y2K work, I had to locate all dates in the almost completely
undocumented internal layout of my company's hospital database system.
Fortunately, the vast amount of data we had made it possible to do a
statistical analysis, and distinguish fields which actually contained
dates from fields which contained data which sometimes looked like
dates.  I found just about 2000 (appropriately enough) different
contexts in which dates appeared in our system.  (Once I had found
them, it was a simple matter to increment them all on a test database,
and test the resulting system as if it were in a hospital in the
year 2000.)

> Magnetic tape is an even bigger problem,

True, but it's not all that bad.  Google had no trouble reading Henry
Spencer's twenty year old Usenet archive tapes.

One of the biggest problems with tape is "bleed-through," in which
magnetization from one turn leaks into adjacent turns.  With a little
effort, this can be corrected for using a deconvolution algorithm.

> and I've heard that the "expected" lifetime of a normal (aluminum
> coated) CD is only about 10 years.

Definitely not.  Like Ted, I have lots of 18 year old CDs, and they're
all still good.  Even though I keep my apartment much warmer than
most, which is likely to accelerate any deterioration.  Except for CDs
with a blatant manufacturing defect (e.g. "bronzing"), I expect them
to last for centuries, since they're basically just bent metal.

> (CD-R recordable CDs have a much longer expected life -- 100-150
> years.)

Are you sure you haven't got those reversed?  CD-Rs use a chemical
process, which might fade like old photographs.
--
Keith F. Lynch - kfl at keithlynch.net - http://keithlynch.net/
I always welcome replies to my e-mail, postings, and web pages, but
unsolicited bulk e-mail (spam) is not acceptable.  Please do not send me
HTML, "rich text," or attachments, as all such email is discarded unread.