Collection policy was Re: No space for vinatge computers in

From: Tony Duell <ard_at_p850ug1.demon.co.uk>
Date: Sat May 24 18:09:06 2003

> I think the right way to display a machine is to get it working while
> simultaneously preserving its historical fabric. The Computer History

Exactly.

> Museum revived an IBM 1620 in a very intelligent way: they used whatever
> modern parts were necessary to get it running again (in this case
> semiconductor memory to replace the damaged core plane) but they made all
> the modifications in a non-destructive, easily reversible fashion, i.e. no
> permanent modifications to the original machine. More information on the

I have a rule for my own collection of never making permanent
modifications other than possibly mounting screw holes for replacement
components. And never changing the design or architecutre (at least not
significantly -- sometimes you have to use a differnet chip as a
replacement, but make it emulate the part you're replacing). And, of
course, document all replacements and changes. Yes, documentation can be
lost or become separated from the machine, but it's better to at least
start out with the modifications documented.

> Is it better to try and fail to restore a machine than to just leave it
> static fretting that it might be destroyed in the process? What is
> destroyed? Killing a rare (or one-of-a-kind) chip? If you put the best
> people on the job and something goes wrong then I think that's better than
> never having tried a restoration to begin with. I mean, the thing isn't
> going to explode or anything. Worst case a chip fries or a board burns
> up.

I have _never_ had a failure that's done visible damage to the machine
(not even burnt traces on the circuit board). If a chip fails, it looks
the same as the chip did before it failed -- it doesn't explode, or fly
off the circuit board. So what's to be lost by repairing and running the
machine? If some irreplaceable part does fail, you _still_ have the
machine as a static exhibit. And by not running it, you don't make
buckups of firmware, etc while it's still readable, and you don't record
important signals while the machine is still operational, so you _do_
lose something.

-tony
Received on Sat May 24 2003 - 18:09:06 BST