Before the Internet, there was the ARPANET, with bridges to BITNET and UUNET and other networks of computers. For email to crawl from site to site routes had to be explicit, so addresses looked like decvax!savax!mf@ucbvax.berkeley.edu and davis%wanginst.edu@relay.cs.net and bptpcapb%uiamvs.bitnet@wiscvm.wisc.edu and vinge%sdsu.UUCP@sdcsvax.ucsd.edu. Recently in an archive of DECUS (Digital Equipment Corporation User Society) traffic I found a message that I sent a quarter of a century ago to a flock of acquaintances. It was the source code for a program to compute statistical correlations among words in big files, part of an information retrieval project I was working on at that time. I don't recognize most of the names I sent the message to, but "vinge" is probably Vernor Vinge and "despain" is Al Despain. At the time I was called "science".

13-Jul-87 06:57:57-MDT,20882;000000000000
Return-Path: <science@nems.ARPA>
Received: from nems.ARPA by SIMTEL20.ARPA with TCP; Mon, 13 Jul 87 06:57:03 MDT
Received: by nems.ARPA id AA02207; Mon, 13 Jul 87 08:33:15 edt
Message-Id: <8707131233.AA02207@nems.ARPA>
Date: 13 Jul 87 08:32 EDT
From: science@nems.ARPA (Mark Zimmermann)
Subject: correl.13t3.c program listing
To: kotelly@mitre.ARPA, research@nems.ARPA, cperry@mitre.ARPA,
        mike@media-lab.media.mit.edu, gergely@drea-xx.ARPA,
        decvax!savax!mf@ucbvax.berkeley.edu, pbr%pco@bco-multics.ARPA,
        microsof!brianc@beaver.cs.washington.edu, dlcdev!glenn@eddie.mit.edu,
        mbj@hq.lcs.mit.edu, ed298-ca@violet.berkeley.edu,
        despain@ucbvax.berkeley.edu, science@nems.ARPA, master@nems.ARPA,
        programs@nems.ARPA, rthum@simtel20.ARPA,
        bptpcapb%uiamvs.bitnet@wiscvm.wisc.edu,
        davis%wanginst.edu@relay.cs.net, kersch%gmuvax.bitnet@wiscvm.wisc.edu,
        rezende%corwin.ccs.northeastern.edu@RELAY.CS.NET,
        vinge%sdsu.UUCP@sdcsvax.ucsd.edu, rada@mcs.nlm.nih.gov
Cc: science@nems.ARPA

Appended below, gods of UNIX willing, is a C program to take an inverted index
in my simpleminded format and perform statistical correlations on chosen
words in that index ... as with the indexer.c program, it seems to run fine
on a VAX, a Sun workstation, and on the Macintosh ... any and all bug
reports and critiques of my C coding style would be appreciated, as would
comments "Don't send me any more of this stuff!" .... ^z

The rest is C ...

(cf. [1], ...) - ^z - 2012-03-03

1 person liked this