Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!uunet!kddlab!icot32!nttlab!nttyrl!nttmhs!yam@nttmhs.ntt.jp
From: yam@nttmhs.ntt.jp (Toshihiko YAMAKAMI)
Newsgroups: comp.groupware
Subject: Re: Group Communication Protocols
Message-ID: <4329@nttmhs.ntt.JP>
Date: 7 Jan 90 10:43:51 GMT
References: <90005.104713UH2@PSUVM.BITNET>
Sender: news@nttmhs.ntt.JP
Lines: 72

From article <90005.104713UH2@PSUVM.BITNET>, by UH2@PSUVM.BITNET (Lee Sailer):
> In article <4318@nttmhs.ntt.JP>, yam@nttmhs.ntt.jp (Toshihiko YAMAKAMI) says:
>>
>> What types of high level group communication
>>protocols do we use in group activities?

> Just thinking off the top of my hat 8-)

> 1.  Some of my communications are 'informational'.  I don't expect any reply,

> 2.  Some are 'requests'.  "Can I have $xxx to buy a widget?"  I'd really

> 3.  Some are 'responses to requests'.

> 4.  Some are 'proposals', as in "I think we should do Z."  I want to sesome

> 5.  With 4 comes the ugly spectre of Amendments, Discussions, Points
> of Order, voting, and so on.

> 6.  Some are 'arguments'.  Joe says let's do Z, and I say "No, because"

I considered it last year.
I picked up 5 types of high level communication primitives.
They have three elements:(recipients, deadlimit, primitive type).
I considered 5 types of primitives.
(1) Information -- it requires no reply, the same as the first one of
Lee's. It requires only delivery notice.
(2) Simple communication -- it waits for replies.
(3) Scheduled active communication sequences -- it will invoke communication
    with specified schedule. (In my view, it is "don't forget" or "do-it"
    type communication)
(4) Summary communication -- it waits for summary of replies.
   (In my view, it is "voting", "paper work authentication by supervisors",
    users do not have interests in individual messages, but in the
    final result caused by the communication sequences)
(5) Scheduled passive communication sequences -- it will wait for
    message arrival. (In my view, it is "daily report posting" or something
    like that).

It is one level of abstraction. I think these primitives are not
sufficient.
For example:
(A) It is not clear for me they can be constructive primitives to
    make more concrete and complex primitives such as "argue", "propose",
    "make appointment", and so on.
(B) I think there should be 'negative' primitives. How can I construct
    "don't disturb", "bargaining", or "ignorance" actions?
    Silence or such kind of non-responses should be managed by
    groupware. Or we should make 'positive' first and it is not the
    time of negative groupware?

I do not know the details of Winograd's work. I read some in last year's
BYTE and "Computer Supported Cooperative Work: A Book of Readings(Ed
by I.Greif)".
I think negative action management has some possibility of extending
action theories.
All I can imagine is "to embed interruption mechanicms by timer 
and re-scheduling mechanisms of actions when action sequences 
meet someone's ignorance'.

I am sorry that my usage of term 'negative communication' is
not well defined.

Any comments will be welcome.

-- Toshihiko YAMAKAMI


Toshihiko YAMAKAMI	NTT Telecommunication Networks Laboratories
 Telephone:	+81-468-59-3781 	FAX:	+81-468-59-2546
 junet:	yam@nttmhs.ntt.jp		CSNET:	yam%nttmhs.ntt.jp@relay.cs.net
 snail-mail:	Take 1-2356-523A, Yokosuka, Kanagawa 238-03 JAPAN