Well, I wrote something up.  Its a crude first pass, so bear with me.  Here
were the test protocols.  The BBS is a 486-50DX running WWIV under DOS5.  My
terminal was an IBM L40SX (20) running Telix under Desqview.  Modem
connection was at 2400 baud; faster times would help immensely.  The phone
line has in the past been noisy, but I don't think that was an issue here.

The first experiment was to try a straight upload script: this turned out to
be not useful.  It turns out that Martin does not have the sector input
hotkeyed; you have to hit return after each sector value is entered.  So my
script had lines of the form
F1
2
F2
1
F1
3
F3
1
...

The problem was that since it wasn't hotkeyed, I couldn't set the ascii
prompt character to : -- so synch problems were severe.  Also, 2002 simply
ignores characters its not ready to accept, so I lost a *lot* of data to
"Illegal Interrogation Mode".  The alternative was to set line pacing (the
pause after each line) to something large, and it would have taken an hour
to complete one pass.  Bad.

The second alternative was to write a simple salt script.  I haven't used
salt in a long time, so it undoubtedly could be improved.  Basically, what
it did was spit out an F, then base sector, then return, then target sector,
then return, then wait for a ":", then wait a little for luck.  The "wait a
little" I first set to 0.1 sec, but that was much too short; 0.4 was better,
with only a little lost data, then I went to 0.3 with a little more lost
data.  Each pass consisted of asking for all routes of the form From: base
To: n where n varied from 1 to 1000 (except n <> base).  With a delay of 0.4
sec, a complete pass took 19'28", and with a delay of 0.3 sec a complete
pass took 17'30".  The first pass was based at 1, the second at a random
sector (500).

The output from each pass was logged to a file.  That was fed to a short
program that generated a "pseudo .SCT" file as Meyers suggested.  I then fed
that to CONVERT.  Here are the results:

source\smc.dat
Space Dock location unknown
Info for 0 notes
Info for 0 ports
Finished reading 999 sectors, 0 ports, 0 etc.

One sector, 168 I think, had one warp in, but both the path from 168 to 1
and from 500 to 1 was garbled.  This of course could be patched manually.
For comparison, here is the complete listing:

Name of data file?  stmarys
Space Dock location 598
Info for 5 notes
Info for 379 ports
Finished reading 1000 sectors, 382 ports, 387 etc.

As for completeness of data:

Distribution of tunnel lengths:
(pseudo version)
 1: 251.  2:  65.  3:  24.  4:  7.  5:  6.  6:  0.  7:  1.  8:  0.
(actual)
 1: 241.  2:  48.  3:  15.  4:  6.  5:  1.  6:  0.  7:  0.  8:  0.

The tunnel thought to be of distance 7 has at position 3 a "back door": in
actuality the tunnel is of distance 4 and the back door is a two way warp.

Stellar Dispersion from what Sector? [0 to abort]  1
(pseudo version)
Computing distances...
    1    6   14   27   51   97  176  231  198  113
   52   20    9    3    1    1    0    0    0    0
    0    0    0    0    0    0    0    0    0    0
There are 0 unreachable sectors, 1000 reachable sectors.
Average number of warps in reachable sectors: 2.224
Average distance to (known) sectors: 6.99
(actual version)
Computing distances...
    1    6   14   27   51   97  177  233  198  112
   50   20    9    3    1    1    0    0    0    0
    0    0    0    0    0    0    0    0    0    0
There are 0 unreachable sectors, 1000 reachable sectors.
Average number of warps in reachable sectors: 2.411
Average distance to (known) sectors: 6.99

Note that 187 warps were omitted (2224 detected, 2411 actual) for 92%
accuracy.  The distance information was preserved exactly.

One Way Warp information:
(pseudo version)
   18-> 77  27->505  27-> 15  27->234  33->778  35-> 65  49->424  50->916
   51->238  56->883  57->651  59->399  60->198  65->614  68->528  71->744
   73->465  76->816  80->105  94->730 105->534 105->459 109->240 118->747
  123->608 137->151 137->198 139->769 144->408 149->831 150->998 150->250
  151->484 151-> 77 156->504 156->356 159->513 173->758 176->985 179->894
  181->212 184->831 190->628 194->227 198->731 208->469 211->820 212->715
  228->792 237-> 76 240->548 243->766 250->391 256->574 257-> 56 270-> 74
  271->903 274->307 290->772 296->753 302->221 307-> 80 311->390 316->950
  325->463 328->437 335->246 343->211 343->743 348->975 358->963 358->835
  359->711 371->733 372->850 388->976 388->271 390->359 390->591 399->286
  428->196 437->388 465->574 485->156 487->771 488->668 502->923 515->517
  528-> 49 534->327 554->738 561->789 561-> 63 561->145 561->228 561->849
  574->614 587->869 587->725 587->993 598-> 57 598->913 598->687 598->849
  608->100 610->912 614->248 628->788 633->913 648->199 651->859 652->362
  660->987 663-> 51 665->633 668->966 679->851 679->942 680->358 687->779
  690->466 698->725 707->184 725->575 730->652 730->298 733->979 738->615
  743->175 746->799 747->601 778-> 20 780-> 89 788-> 24 789->979 789->429
  792->797 797->118 801->868 816-> 39 831-> 27 834->254 835->290 842->911
  846->929 851->130 852->190 869->194 889-> 59 913->648 917->657 922->330
  929->938 930->428 942->265 944->922 950->506 951->598 963->343 987->366
  989->957 991->257 992->371 993->789 994->570
total of 165 one way sectors discovered.
(actual)
   49->424  76->816  94->730 105->459 123->608 150->250 157->859 159->513
  208->469 212->715 243->766 270-> 74 274->307 296->753 325->463 328->437
  390->359 399->286 416->328 561->849 561-> 63 561->145 561->228 561->789
  569->630 575->889 587->725 587->993 587->869 598->747 598->849 598->687
  598-> 57 598->913 601->157 608->100 614->248 633->913 648->199 660->987
  663-> 51 680->358 690->466 705->575 707->184 730->652 743->175 746->799
  780-> 89 788-> 24 797->118 801->868 816-> 39 852->190 869->194 889-> 59
  913->648 930->428 944->922 987->366 989->957 991->257 992->371
total of 63 one way sectors discovered.

As to be expected, most of the missing warps are one direction out of a
two-way warp.  Naturally, this makes picking the class 0 ports out a little
harder... the stardock here is at 598, and has five one-way warps.  561 also
has five one-way warps.  Under the pseudo version, 598 is adjacent with four
one-way warps, although 561 still has five, and no other sector has as many
one-way warps.

Conclusions:
This kind of problem really calls for a term program, both to manage the
comm port better and to be smart enough not to make unneeded inquiries.  It
may be that a better telix script or not being under desqview would have
helped.

However, I think this experiment is a success.  Given fourty minutes (two
fast passes and a little hand work to catch the ones that got garbled) would
achieve extremely fine results; even what I did was only 37 minutes and
achieved 92% efficiency.  This can of course be repeated to produce any
desired level of accuracy.

Its all over but for the details.  The universe warps can be known on day
one.

I will put all this together into an archive, if anyone wants to repeat my
experiment (or is starting on a fresh game... :))  INSTMAP.ZIP or somesuch.
I'll place it here, so its on a fido board, and also at 510-376-1554.  The
latter is a USR Sportster 14.4, and you can log in as GUEST, password GUEST,
phone 9999 and not bother with applying for an account.
