perm filename RAND[RDG,DBL]3 blob
sn#558869 filedate 1981-01-26 generic text, type C, neo UTF8
COMMENT ⊗ VALID 00010 PAGES
C REC PAGE DESCRIPTION
C00001 00001
C00002 00002 Mailed to CSD.LENAT 14:47 18-Dec
C00003 00003 Date: 18 Dec 1980 1714-PST
C00006 00004 ∂Mailed to RICK@RAND-AI, CSD.LENAT 11:59 19-Dec
C00009 00005 ∂19-Dec-80 1237 RICK at RAND-AI Re: Meeting Dates
C00010 00006 ∂Mailed to RICK@RAND-AI, CSD.LENAT/CC 11:42 22-Dec
C00012 00007 ∂22-Dec-80 1516 RICK at RAND-AI Re: The best laid plans of ...
C00013 00008 1. Details of RLL
C00014 00009 ∂31-Dec-80 2354 RICK at RAND-AI Re: The best laid plans of ...
C00015 00010 Mailed to RICK@RAND-AI 16:12 26-Jan
C00024 ENDMK
C⊗;
Mailed to CSD.LENAT 14:47 18-Dec
LA Bound
Doug
This is just a reminder of my trip to LA next week. Current plans
are to leave sometime like Tuesday or Wednesday, and to return on the week-end -
driving both ways. (If Rand prefers, I could visit there on Monday, en route
back.)
Also, remember to ask Rick about the possibility of letting
us use his RAND-AI machine for our RLL work.
Russ
Date: 18 Dec 1980 1714-PST
From: CSD.LENAT
Subject: [RICK at RAND-AI: Re: RLL]
To: CSD.GREINER
You should send a note to Rick, proposing which days next
week are good for you.
I think he expects this to be planning meeting, rather than a real paid consulting
day. If a joint project does emerge, you should inquire
about arranging a consulting concract, getting a clearance, etc.
Let me know what's happening.
Regards,
Doug
---------------
Mail-from: ARPANET site RAND-AI rcvd at 18-Dec-80 1658-PST
Date: 18 Dec 1980 1655-PST
From: RICK at RAND-AI
Subject: Re: RLL
To: CSD.LENAT at SU-SCORE
In-Reply-To: Your message of 18-Dec-80 1609-PST
doug:
i'm interested enough in rll that you ought to have your emissary
drop by with some concrete proposals. i am interested in the language
issues following ROSIE and still wish we were working on competition,
etc. i think it may be possible to work something out if we can find
some target application. at this point, there isn't any specific application
underway, but i'd still be interested in the competitive theory problem
and i have a new colleague who might be good. have russ propose
some meeting times.
on the book, we now have hard deadlines from all the authors
ranging from jan15 to feb 15 for their drafts. it's going slower
than i hoped, but it looks like they'll all come in soon.
i hope everyone up there is doing well. our aaron has his first
illness, laryngitis perhaps virus/flu related. oh well.
cheers,
rick
-------
---------------
∂Mailed to RICK@RAND-AI, CSD.LENAT 11:59 19-Dec
Meeting Dates
Rick:
Doug forwarded your message to me. Sorry to hear about Aaron; but
hope his reduced vocal abilities have, at least, lead to increased
sleep for you and Barbara...
I'm still formulating plans for this trip down south.
The current best candidate involves leaving here
this Tuesday, and returning, say, Sunday morning.
This means I could stop off at Rand en route on Wednesday, the 24th.
Another possibility would require postponing my return departure a day,
allowing us to meet on Monday the 29th.
(I would actually have no objection to meeting anytime in between,
if that would jive better with your schedules.)
As to the content of that meeting:
I wasn't able to infer from your response what sort of "concrete
proposals" you were after.
Did you mean finding a connection joining the reprsentation-related research
(of RLL/MRS) with the linguistic based approached examplified in
ROSIE, or some particular application which required the joint facilities
of these two systems? ... or something completely different.
(Perhaps you were comtemplating something based on the
"competitive theory problem" you mentioned.
I'm totally unfamiliar with that work (or even just what it means).
If it's relevant, could send some reference to this work?)
Anyway, I look forward to seeing you et al. soon. Let me know when.
Russ
∂19-Dec-80 1237 RICK at RAND-AI Re: Meeting Dates
Date: 19 Dec 1980 1234-PST
From: RICK at RAND-AI
Subject: Re: Meeting Dates
To: RDG at SU-AI
cc: csd.lenat at SU-SCORE
In-Reply-To: Your message of 19-Dec-80 1159-PST
russ:
let's meet on the morning of the 24th for our first
interaction. how's 9:30.
concrete proposals refer to sow's ears and other
such yuletide offerings.
see you then,
rick
-------
∂Mailed to RICK@RAND-AI, CSD.LENAT/CC 11:42 22-Dec
The best laid plans of ...
Rick -
The weather in LA & SF conspired to keep my brother (and
part-time auto mechanic) from arriving here last evening.
This made it very hard to begin tuning up my car this morning, as we'd planned.
This will, in turn, push back my entire schedule.
So...
Would it be possible to meet sometime AFTER this Wednesday? As I mentioned
earlier I would have no objection to convening on the 25th or 26th.
(These do seem appropriate times to discuss concrete sow's ear, purses,
and other things pertaining to this Solstice Season.)
The other obvious candidate is next Monday.
If these are all out-of-the-question, I do have plane reservations (for tomorrow
afternoon) which I could honor.
Russ
∂22-Dec-80 1516 RICK at RAND-AI Re: The best laid plans of ...
Date: 22 Dec 1980 1507-PST
From: RICK at RAND-AI
Subject: Re: The best laid plans of ...
To: RDG at SU-AI
cc: csd.lenat at SU-SCORE
In-Reply-To: Your message of 22-Dec-80 1142-PST
russ:
next monday is best by far. what time would you like?
how about 9:30 at rand (main "east" lobby is where we'll have to meet).
--rick
-------
1. Details of RLL
2. Paper on MRS
3. Discuss:
a) Plans to cooperate on project involving RLL (/ROSIE?)
b) Use of RAND-AI machine
(for RLL, perhaps other HPP can buy time as well)
c) My involvement:
Consulting with researcher on 3a
Suggesting/critiquing ROSIE?
4. DBL might drop by
Call Steve - deliver memos to him
∂31-Dec-80 2354 RICK at RAND-AI Re: The best laid plans of ...
Date: 28 Dec 1980 1930-PST
From: RICK at RAND-AI
Subject: Re: The best laid plans of ...
To: RDG at SU-AI
cc: csd.lenat at SU-SCORE
In-Reply-To: Your message of 22-Dec-80 1142-PST
russ-
you're right about the best laid plans...
i've got the flu tonite (sunday) so I'm almost positive
we can't meet monday morning as scheduled. we'll have to
do it by phone i guess. sorry, but i'd rather not
expose you. i'm trying various ways to reach you before
you come to rand, but if i fail, i'm sorry for any
inconvenience this causes.
rick
-------
Mailed to RICK@RAND-AI 16:12 26-Jan
RAND-AI & Consulting
Rick -
Greetings. Hope your flu has subsided, and didn't inhibit your New Year's
Day celebration.
I'll try to briefly paraphrase in this letter (my side of) the discussion
I'd hoped to have with you during my last visit there. There were two
basic questions: How available is the Rand-AI computer, for RLL-related
work; and are you as interested in a continuing dialogue between Rand and
Stanford as we are. Related to the second point was the possibility of
Rand hiring me as a consultant on some issues.
First, computer issues: Over the last few months I've only once seen
Rand-AI's load average top 1.0. Are these apparent cycles-to-spare for
real? Does ths reflect a permanent condition? As Doug might have
mentioned, the computers here are all overcrowded -- I spent hours
performing trivial tasks, even on week-ends, because of their consistently
high load average. Would it be possible/permissable for us RLLers here to
use your apparently abundant resources?
There seem many approaches:
1) I could simply use the (ROSIE-inspired) account you guys so magnimously
gave me. I've not done that so far, lacking your permission. This would
certainly be the easiest solution, and would be 90% of everything we would
want. (The only 10% corresponds to a bit more disk space.)
Of course you at Rand would gain by this deal as well. With these closer
ties it would be trivial for us to provide, and maintain, a variety of
software tools developed here. This would include
i) CORLL, a (by-now-well-debugged and tested) demand-paging facility to
provides virtualy infinite storage space for data, avoiding InterLisp's
256K limitation. [This includes a library of small, general (almost
documented) user functions written over these years, to sidestep some of
InterLisp's limitations/problems.]
ii) RLL itself (I'll very soon freeze a version, which would be
available to anyone who wishes to play with it.)
iii) MRS - the "other" representation language language here at Stanford.
In addition, there would be someone (ie me) to maintain (or at least serve
as liason to the real developers/maintainers of) these various systems.
2) We here at Stanford could work on some RLL-related project with youse
guys at Rand. The first example which comes to mind would be a
continuation of the quick&dirty comparison between
expert-system-building-systems begun in San Diego last summer -- using RLL
and ROSIE as the two examplars. Another proposal would involve
programming some actual application task (either something Rand is
currently working on, such as Early Warning simulations, or something
totally new) in RLL.
3) If Rand isn't feeling generous, and none of these proposals seems
feasible, there is a small amount of money which could be used to buy some
time on your machine, at least for a short while.
Of course, if that fast response time at Rand-AI reflects a poor sampling
on my part, or if it is only ephemeral, this whole point is moot. One
final comment: Doug assures me that by the year we'll have ready access
to Xerox's Diraldos - at this point we'll probably move RLL from the
slower machines onto such systems.
----------
On to the other point -- the possibilites of consulting.
While developing first RLL and more recently working on MRS, I was forced
to address, and (at least attempt to) resolve, a wide variety of
representation related problems. These issues seem fairly ubiquitous to
representation languages; it is unlikely the ROSIE developers will manage
to continue avoid these situations.
In fact many of the problems that Danny was just beginning to address
seemed closely ties with such matters - such as how to make ROSIE suitable
extensible, and how to handle variables/intentional-objects.
As Doug may have mentioned, I would be interested in sharing my both
experiences and suggestions with ROSIE's developers. A consultation job
would also serve to cement Stanford-Rand ties, and help avoid (or at least
minimize) the otherwise unavoidable duplication and reduplication of
effort.
If this is acceptable, there are several ways I might fit in. The most
obvious would be to visit Rand abot once a month, and discuss the recent
developments. Another tack would be to involve myself, with others there,
on some particular application task; one designed to "stretch" ROSIE's
boundaries. This more goal-directed approach would expose rough spots
more rapidly than the more hit and miss approach suggested above. The
third would follow the proposal #2 mentioned in the earlier part of this
letter -- performing some direct comparision of these various
representation languges, towards a more complete analysis of each of them.
----------
I am quite open to any different suggestions you might have, on either of
these questions. Let me know what you think.
Russ