Copy Link
Add to Bookmark
Report
AIList Digest Volume 5 Issue 178
AIList Digest Monday, 13 Jul 1987 Volume 5 : Issue 178
Today's Topics:
Theory - Symbol Grounding Poll: Nays,
Comment - Characteristics of Discussion Lists
----------------------------------------------------------------------
Date: 9 Jul 87 03:44:34 GMT
From: mind!harnad@princeton.edu (Stevan Harnad)
Subject: Re: Results of Symbol Grounding Poll: Nays (3rd of 3 parts)
[These are the 37 nays in response to the poll on whether to continue
the symbol grounding discussion in comp.ai/comp.cog-eng. I have removed
names and addresses because I had not asked for permission to repost them.
If you wish to communicate with anyone, specify by number (*and* whether
"yea" or "nay") and I will forward it to the author.]
-------
[The first three nays (from Harwood, Minsky and Booth) preceded this
poll; enumeration accordingly begins with 4.]
-------
4. Please do not take this personally. I have almost stopped reading
comp.ai because of the ridiculous quantity of material being posted by you
and Brilliant, and others. This discussion has been completely unuseful
to me and I would really like to see it stopped. It is much more like
philosophy than AI to me and I am sure there are others who feel the same way
but wont tell you. Please stop dominating this newsgroup.
-------
5. Either you start another newsgroup or I unsubscribe to this one.
I cannot take any more.
-------
6. Please start your own newsgroup!
-------
7. My vote: *do* start your own news group or private mailing list.
This discussion, however interesting it may be to the participants,
has gone on too long to continue in "comp.ai".
-------
8. I enjoy skiming your symbol grounding writing though for my research it
is totally irrelevant. However since there are relatively few people
who do AI who need to consider the TTT (most people in AI are just
trying to make machines more intelligent right now) I suspect that the
symbol grounding problem better belongs in sci.philosophy.tech. The
issues wont come up in the real world for at least 5 years because we
are not even close to human emulation at the moment. On the other
hand you may be working on psychological modelling. If so then there
must be a news group or mailing list more close to that topic than
comp.ai. All together I suspect that sci.philosophy.tech is the best
place along with periodic notes to comp.ai notifying people that there
is a discussion of importance to those who model human beings there.
This would get your messages to the relevant people. Also if
sci.philosophy.tech doesn't exist for some reason then talk.philosophy
would be the next best thing. If the problem is that you can not
reach the arpa world that way, I think there is a psychology mailing list.
BTW about graded vs ungraded concepts, point taken.
On the other hand most of the verbs in any language are regular but
most of the verbs used by the speakers of a language are irregular.
The dictionary is not meant to be and is not a fair sample of usage.
Nor does the set of nouns in the language necessarily correspond to
the set of concepts employed by its speakers, (it corresponds to the
set of concepts that the speakers find convinient to convey rapidly).
However you have presented inconclusive evidence that most concepts
are not graded. If you had a dictionary that was sorted by usage and
gave the usage of words rather than their definitions you would
have better evidence that most concepts are not graded.
-------
9. As to your polling request regarding the symbol grounding issue:
I am quite tired of all the traffic it has generated. Considering that no
real information has been revealed, I feel it is time to drop it. In the
recent time that these postings have filled the newsgroup, most all other
worthy postings have vanished.
The newsgroup should address a range of pertinent issues that will
enlighten subscribers. I feel that the symbol grounding issue has only
enlightened me in the use of the 'n' key!
While I am on the subject, the cross-posting to 'comp.cog-eng' are
atrocious. Either post to one or the other. Most every symbol grounding
article has appeared in both. This generated to much traffic on the net
and defeats the purpose of making special purpose groups.
I thank you for your ability to notice fellow subscribers views.
-------
10. Can't we bag this damn symbol grounding discussion already?
If it *must* continue, how about instituting a symbol grounding news
group, and freeing the majority of us poor AILIST readers from the
burden of flipping past the symbol grounding stuff every morning.
-------
11. I generally do not read the SGP articles simply because I do not
understand them (and they are so looong!). If there are a few people interested
in reading and discussing SGP, there is no reason to prevent such postings. But
if there are also many people who do not want to read that sort of things in
comp.ai, then it would be wise to consider the possibility of creating a
news-subgroup `comp.ai.sgp'.
-------
12. The ramblings on this topic passed my threshhold of boredom long ago.
I'm not proposing censorship, but if you choose to continue the discussion
with a smaller group of people who find this topic of interest, I will
applaud your good manners.
-------
13. I vote you start your own newsgroup--I was bored with "Symbol Grounding"
about 500 kilo-bytes ago. Ditto "The Total Touring Test" or whatever
your last filibuster was called. . . .
-------
14. My vote is for ending the discussion on the symbol grounding problem.
Thanks. p.s. If you are interested in finding out why I voted against
continuing the discussion, please let me know -- I will be glad to oblige.
-------
15. Thank you for taking a poll on whether the symbol grounding problem
discussion should or should not continue in comp.ai. My vote is to remove the
discussion from this newsgroup. Maybe it could be moved to a new newsgroup
talk.symbolgroundingproblem ???
-------
16. I think that the discussion has been out of hand for a long time now.
It doesn't seem to contain any useful insights, and is taking up inordinate
resources. Not the least of which is the time spent by the authors
expounding their viewpoints. I think that this sort of disagreement is
better done in position papers in and letters to journals.
The odd use of terms hasn't helped keep the discussion on a high level.
Not to point fingers, but your nonstandard use of "analog" made a large number
of your posts completely incomprehensible to me until you said that you meant
something other than the usual meaning of the term.
So, I vote to flush this discussion.
-------
17. Personally, I have been skiping most of the articles in this discussion.
I was referred to this newsgroup as a forum for other discussion but have
seen little other than what appears to be a war of words from two opposing
camps. By now the sides must be set--perhaps it is time to move the
discussion from "news" to an e-mail mailing-list.
-------
18. Definitely neither useful nor worth continuing.
-------
19. The manner in which the issue was raised *was* rather rude, but I regret
to say that I find much of what was stated about your extended discussions
very much to the point. I tried to keep up with discussion; I found it
rather interesting at first. But it rapidly became clear that you were
all talking at cross purposes, refusing to accept conventional usage or
even common-usage-for-the-purpose-of-debate of the key words in question.
The appalling level of quotation made things much, much worse and it became
well-nigh impossible to ferret out the pearls of insight in the flood of
verbiage. I do not wish your discussion to completely vanish from the
airwaves, as it were, but without a bit of self-restraint all round,
together with some sincere efforts to try to answer one another's
objections, I don't think the discussion is particularly useful. (e.g.
wrt all-or-none categories: pointing to concrete nouns in the dictionary
or to the very special categories that have "hardware support" is not,
in my opinion, a sincere effort to meet the objections to the contention
that categories are all (or mostly) all-or-none, a rather contrary-to-
common-observation position.)
Perhaps the new policy on quotation will help: there has been a modest
improvement in a couple of the recent postings. I remain hopeful. All
I can say is, until things improve quite a bit, I will probably be
flushing all the digests with "Symbol Grounding" in the topics list. Sorry.
-------
20. I do not find the symbol grounding problem discussion worthwhile.
Thank you for (politely) asking.
-------
21. I vote for discontinuing the discussion. It would be interesting except
that there is far too much confusion over who's using what terminology.
Probably dozens of articles have been wasted over "well, I don't know
what *you* mean by 'analog', but when *I* say 'analog' I mean etc etc etc".
-------
22. You have made an unseemly attempt to bias this vote. The question is
not whether your discussion is ``useful and worth continuing,'' but whether
we *ALL* need to read or even be sent the truly amazing volume that you
seem able to generate on this one topic !?!
** Please remove your discussion from the AI-list (to a new bboard?). **
{And if you find it absolutely necessary to be mad at how stupid and
unjust the rest of the world is, go ahead and tally this as a vote for
your discussion being useless and not worth continuing}
-------
23.
1. I find it neither interesting nor useful.
2. The arguments, until I stopped following it, somtimeseveral weeks ago,
are circular if not repetitive.
3. I've speculated privately that the argruments were cranked out by a
machine in someone basement as a Turing Test on the rest of the net.
Either that or ...
4. But none of this justifies setting up another news group. comp.ai
isn't being used for anything else. For a heavily used group, see
comp.sys.ibm.pc.
5. Personally, I'd suggest that you take all of the correspondence. Put
it in a folder, and open it again at New Years. Reread it, and write
a real paper.
-------
24. Please stop!
-------
25. NO! Please take this discussion to e-mail. It's gone far
beyond the point where it's interesting to anyone other than
you and the few people still arguing.
-------
26. Stop it!
-------
27. The symbol grounding problem - please start your own newsgroup.
DEFINITELY!
-------
28. Although I don't think that AI-list should be strictly limited to
discussions of algorithms and similarly down-to-earth items, I do think
that the symbol grounding discussion has gotten a bit out of hand and
should be conducted privately among the three or four major participants,
with perhaps a summary to appear at some future date.
-------
29. In article <977@mind.UUCP> you write:
>David Harwood has made two very rude requests
(Yes, he was way out of line.)
As a former philosophy undergrad and current A.I. grad student,
I've found the topic in general to be interesting.
BUT . . . I think it should in fact be moved to its own newgroup.
Comp.ai is now completely dominated by exchanges between you and
Marty Brilliant, Anders Weinstein, etc. After a while, "listening"
to a few other people argue gets tedious, no matter how interesting
the topic. Frankly, I think people have been frightened away from
the newsgroup in the past few months, with the result that there have
been no discussions other than this one, unless you count a few
requests for info on some language.
P.S. I enjoyed your "uncomplemented categories" talk at the Phil/Psych meetings.
-------
30. I vote to cease the endless symbol grounding discussion!
-------
31. I find the discussion neither useful or worth continuing.
------------
32. Please stop it. I agree with Law that most of the discussion can be
carried thru private mail. I can see that R is easier to type than
mail ...%....@...... etc but, then use the facilities provided by
Unix like aliases etc. I am looking forward to your results.
-------
33. You asked for votes. Mine is... no more on the symbol grounding
problem. Thanks for asking.
-------
34. I for one would greatly appreciate having the discussion removed from
subsequent AIlists. As in a conference presentation, if a heated topic goes
on for too long, the people involved should agree to meet later and discuss
the issue amongst themselves without burdening the whole group. You must know
by now who the interested parties are; can't you just send mail to each other?
-------
35. It not the discussion per se that I think people object to as
much as it is the size of the discussion. The replys are very
large, each addressing 15 points of reply to the previous reply.
It takes a while to read through the text, and extract
some salient points of interest. Having real work to do, I sometimes
just file the message, thinking I'll get to it later.
I save ALL my mod.ai mail for a time in the near future when I attempt
to complete my MS and want to scan back over the current "hot" topics.
Unfortunatly I've had to start a special archive just for this discussion,
and it's chewing my disk drive all to bits with saved mail.
I find the disscussion interesting, and informative but...
(Now for the poll): If discussion continues to involve ginormous reply's: END IT
If discussion stops taking over whole digests: KEEP IT.
-------
36. I'm sorry but for me the discussion is no longer interesting.
-------
37. I think that this discussion belongs to philosophy, not to AI. I hope
that it will relocate itself accordingly.
--
Stevan Harnad (609) - 921 7771
{bellcore, psuvax1, seismo, rutgers, packard} !princeton!mind!harnad
harnad%mind@princeton.csnet harnad@mind.Princeton.EDU
------------------------------
Date: Thu 9 Jul 87 10:41:46-PDT
From: Ken Laws <LAWS@IU.AI.SRI.COM>
Subject: Characteristics of Discussion Lists
[Excerpt from a message to Steven Harnad.]
A problem with large, permanent lists is that they are primarily for
those on the fringes of a field who want to monitor or join what is
happening further in -- but not so far in that it becomes a full-time
occupation or involves incomprehensible jargon. The professionals
already have channels of communication among themselves (including
personal visits, seminars, conferences, publications, and even e-mail
or phone calls) and have little time for list discussions that are
outside their own exceedingly narrow specialties.
As to the suggestion of continuing via e-mail, it's not really so bad.
Two options exist. One is to cc everyone on each message, letting the
mailer propagate the cc list from one message to another. It is usually
easy to add new members to such a discussion, but impossible to drop old
ones without retyping the whole list. There is also a problem that BITNET
gateways don't add necessary routing information to message that are
handed over to the Arpanet. The other option is for one person to
maintain a file with all the addresses, headed by a "label:" to suppress
the information in the cc field of each message. All traffic is sent
to this one individual, who then remails it to the distribution. That's
a moderated list. (Anyone can get in this business!)
One of the charges in your Nay summary was that discussion of other
topics has been down since the fundamentals discussion took over.
I believe that's true, although there seems no rational reason for
it. Even queries and replies have been reduced, although that could
be a coincidence due to the end of the school year and of the proposal
year. A few people have dropped off the list because of the volume,
many more have added themselves because AIList was beginning to
border on their interests. The effects are complex, and certainly
not just a linear addition of your text to whatever would have been
present anyway.
I believe that the proper model of a discussion list is the town
meeting. AIList began with my own announcement of myself as
moderator, or chairman/speaker of the house. A group of interested
individuals formed, and through custom and convention we have worked
out an informal social contract that governs the proceedings. Part
of the contract is that participants pay reasonable attention to
the proceedings, if only to avoid redundant or naive remarks. This,
together with the serial nature of current message streams, implies
that only one person (more or less ...) has the floor. Part of
my job as moderator is to insure a balanced discussion, soliciting
(or forwarding) new topics and viewpoints. Not every list is run
as a town meeting, but that's my view of AIList.
The symbol grounding discussion was carried out with great respect
for the participants and with incredible attention to detail. AI
needs to grapple with the problems you raised. (Whether AIList
needs to is debated in your vote summaries.) The difficulty is simply
that people can't pay attention to everything, and your discussion
was demanding more attention than they could spare. The other rings
of the circus require equal time.
Incidentally, much of the personal criticism has been sparked by the
one-against-all nature of your discussion.
If the level of discussion had been more approachable, we might have
had more people joining your cause and providing examples for your
position. That would have been more interesting, and might have
reached an obvious conclusion or stalemate sooner. It is a common
characteristic of net debates, however, that nothing is ever settled.
Points that are agreed to are simply dropped, with little or no mention
that agreement has been reached, and may even be picked up by some
other participant. Net discussions generate a continuous stream of
ideas, but conclusions are lacking. I thank you for repeatedly
reminding us that conclusions have not been reached in this particular
topic area, and hope you will continue to contribute to AIList.
-- Ken
------------------------------
End of AIList Digest
********************
