[1461] in Commercialization & Privatization of the Internet
No subject found in mail header
daemon@ATHENA.MIT.EDU (Dave Hughes)
Wed Oct 9 14:58:23 1991
Date: Wed, 9 Oct 1991 12:57:46 -0600
From: Dave Hughes <daveh@csn.org>
To: com-priv@psi.com
While many on this mail list are spending their time attacking
everything about Abernathy from the political-computer-correctness of
his knowledge to the way he blinks his cursor, I prefer to address the
issue he raises - that of how to cope with juvenile behavior on the
Internet. It is a wholly legitimate question and one which I have given
more than a few minutes thought to, for a number of years, and in
far broader terms than just user 'behavior.'
For I can't for the life of me figure out why there is this
unquestioned assumption that seems to permeate all the discussions about
"K-12 and the Internet" that the only way to serve the K-12 world is to
put all the kids directly 'on' the net, as distinct from getting them
most all the benefits of use 'through' the net. There is a huge
difference.
Thus I can partly agree with Tom Grunder's approach to the
problem in his suggesting that some classes of users of the Internet be
able to get to it only 'through' an interconnected system such as his
Cleveland Free Net, which itself can be set up to enable only that which
the local supervisors want enabled.
But I don't think Tom has gone nearly far enough. For he too is
talking about only the case where all users have to be able to rlogin
and telnet if the supervisors permit it - i.e. that they are still 'on'
the net in a TCP/IP sense.
I go much further. I think the only sensible answer to the
'behavior' *and* a host of other problems which 'access to the
Internet/NREN' will create is to develop - by progressive hooking-up -
a vast 'net of nets' using - or at the very least starting with - the
same distributed computing/conferencing model that Usenet has
represented - but extended right down through LANS, or Fido-type MSDOS
or Mac systems with one or more lines, to the last networked Apple II
'Fredmail' system - or even end-user Point software connections running
on a field HP laptop with packet radio.
While I am, and will remain the most vigerous, outspoken,
and unapologetically activist champion of the proposition that access
to any national, publically funded, regulated, or backed, data network
be made available at reasonable cost to all citizens of the US, from my
pre-school granddaughters to the President of the United States, I am
not so open minded or ignorant about the practical problems this will
pose that my brains have fallen out.
The other problems besides dealing with behavior while
anonomyous ftp'ing pornographic GIF files during study hall, using the
library computer are (1) the daunting logistical problem of extending
true TCP/IP nodes all the way down to every local-dial-code area with a
school in it in the US, and (2) the number of dial-in, as well as
directly-connected ports that would be required in every school
'community' to support direct student-user connection where the brain
I/O rate of the user will not be any 9,600 baud, and thus (3)the
practical economics involved in serving 16,000 school *districts* with
40 million students with access to NREN/Internet in a reasonable time
frame.
My answer has been for a long time 'distributed' conferencing,
e-mail, file transfer, which in the work I have done in Montana with Big
Sky Telegraph, is proving pretty satisfactory.
For of equal importance to solving these problems by technical
fixes or vague references to relying on 'personal responsibility' the
'distributed' model is not a static concept, but an ever-changing
progressive model - where power (and responsibility) is increased to the
users and small system operators progressively as they both learn how to
use more powerful features, how to behave while using them, and for
administrators, teachers, and network operators how to supervise online
social behavior, the same way they had to learn (and not overnight
either) how to handle grafitti in the lavatories and the use of school
xeroxes to copy porno comic books. And the systems themselves can and
will 'progress' from simple dial up systems to complex NSF net
powerhouses possible on even the smallest 486 today.
Thus when I set up Frank Odasz first Big Sky Telegraph Unix
system 5 years ago with both dial in and terminal access, it became
connected to 'the nets' first through my system (1,000 miles away, where
the cost of calls is cheaper than the 60 miles away to the university
city) and Usenet, then through my Old Colo City system to the Internet
(Colorado SuperNet) and finally - this year - directly to the Internet.
And it *may* go further to be SLIP connected to the Internet. But only
as everybody involved - supervisors, sysops, educators, and the 1,500
direct users of the system are ready for it - having had to learn a
whole range of technical skills, cultural norms, and behavior patterns
from terminal or modem control, CR/LF mapping, through e-mail, file
transfer computer-conferencing, mail addressing, *distributed* (new
group) conferencing, and mail-listing. And if they use SLIP there will
be plenty more to learn before serious use of ftp, rlogin, or telnet can
begin. Yet messages, conferences, mail, and files flow in and out of the
Internet from Big Sky and its users, and have been for 3 years.
That progressive network and 'online community' development
pattern took care of the 1,500 direct login users of BST. And beleive
you me they are still learning! But those not local to Western Montana
College (only maybe 25 users) were either direct long distance dial or
800 number. There was still a major economic problem, which Tom's
big-urban area Cleveland model does not address.
What about Butte, Montana, with its high school, 60 miles from
Dillon where Big Sky Telegraph is, and about the same from Helena, where
the closest direct connection to the Internet would be? A T-1 line to
Butte? Or even a 9.6 line? The $800 server Tom refers to is only a tiny
piece of the cost - the data link to the nearest gateway could cost
every two months as much as that server, not to speak of the labor costs
of administering it. Prohibitively expensive. (How many K-12 schools in
the US are a local-no-toll call to an Internet server?)
The answer was setting up a Fidonet, with V32 9,600 baud modem
in the Butte High School - which is local to both the school and town,
and linking it only once a night with compressed files by Fido protocol,
talking to 'Tiny Sky' Fido, across the room from 'Big Sky' and linked to
*it* by Ufgate <--> UUCP as often as needed to handle mail, files, and
conference/echo/newgroup comments.
This was used far less for 'behavior control' reasons that the
fundamental economic realities which will face every school not right
next to an Internet institution or with a server on their premises. The
cost of dial up access. When Dr. Johnston of MIT taught the first 'Chaos
Math' course through this net, the cost of the 5 Butte student access to
Big Sky Telegraph by direct dial was over $600 a month, but the minute
the Fidonet got going, it dropped to under $50. Instead of 5 students
dialing into Big Sky Telegraph staying online for 20-30 minutes apiece
several times weekly at $22 an hour long distance, they logged into
their local machine, which called up nightly at compressed 9,600 baud
and transferred their work and accepted the incoming messages in 5
minutes or less at $11 an hour.
Of equal importance, the school computer coordinator was able to
handle the technical challenge of running an MSDOS Fido system on a
school computer - itself hardly trivial for people just getting into
this telecom game - where I would have been reluctant to tell him he had
to run a TCP/IP server, or unix box for his first network outing.
Now that model has worked so well (6 distributed Fidos were
emplaced within 6 months) that (1) the State of Montana is equipping
over time, all 800 K-12 schools in the state with 9,600 baud V32 modems
and (2) the Office of Public Instruction level (dept of ed) is deploying
17 more 'Fidos' themselves across the state by their own staff AFTER we
installed a 4 line Remote Access Fido on one of their IBM Model 70s and
they learned step by step how to do it.
They knew their main frames and minis and I would not have
hesitated to put a TCP/IP server at that level. But for all their
professional computer background they were no more knowledgable about
how to set up and run (culturally as well as technically) a Fido net
than the elementary schools were able to set up and run a Unix system.
They could just learn faster (though I always have my doubts about such
professional-level DP shops being able to handle the BBS 'culture' any
better than BBS users use the Internet/Bitnet culture right off.)
We also introduced them to 'point' software which permits
end-user machines (PC, Mac or about any other flavor) to link to the net
(either to Fidos or Usenet/Internet servers) by fast modems without even
setting up a BBS.
All this has been done not only to tolerate the costs of
'network access' in rural, small town, or metered-call places where the
FIRST cost is reaching the server, whatever the NREN/Internet cost is
going to be for someone, but ALSO to deal with the unmentioned problem
of the number of ports, terminals, modems (and therefore phone lines)
required to be ON the server if everyone has direct access to it.
Example right here in Colorado Springs, where Colorado SuperNet
has a Internet server with both direct connections (such as Cray
computer company, University of Colorado, Colorado Springs) and dial up,
including SLIP, connections. With at least 75 K-12 schools, 40,000
students and probably 3,000 teacher/adminstrators (not to speak of the
10 or so higher-educational AND community/technical college level
institutions AND 'independent researchers, self-learning students NOT in
formal educational institutions - a genre which will grow in number
immensly in the future) in the local dial area, I have a hard time
thinking that the best solution is to have 500 ports on the local
server, or even a singular 'community' Cleveland FreeNet.
Scale. Scale. Scale.
But if you go 'distributed' - and certainly including MSDOS/Mac
TCP/IP software right on the end machine - ala that 'ftp' company's
products, OR, intermittent SLIP access - both to permit full use of
Internet features, and not just UUCP-Fido-Frednet-Point type exchanges,
I think you deal with the other end of the economic problem too. For as
I watch big institutions with big computers complain about their
operating costs per message, and compare it with the $$$ costs of
running a multi-port desk top Unix system, multi or single-port
networked BBS, I have a hunch there are NOT economies of scale at the
'number of direct ports' and 'number of login users on one machine'
level of computing.
There certainly are economies of scale in network traffic,
but at the point where numbers of 'people' log in to a system?
Finally there is the whole question of 'user interface' which
keeps being debated in terms that seem to me to imply that the only
solution is for some big grant to create some one centralized way of
users interfacing with the Internet/NREN, with some singular master
point-click, look and feel by all connected systems. I vehemently
disagree with that. I believe that decentralized distributed systems, so
long as the critical behind-the-crt technical interconnect standards are
paid attention to (and I know that the differences between Fidonet and
Usenet and Bitnet and Internet, not to speak of Frednet can give plenty
of gas pains) that the end user interfaces can and should be just as
decentralized. For that also permits, say a teacher in an overcrowded
school full of uncontrollable teen agers to enable, disable, functions
according to the social norms of that computer classroom, while the one
in a Jesuit private school with 6 students schools can be quite
different.
Not to speak of 'second language' and 'appropriate language'
interfaces. I know plenty of teachers of younger children who object to
the use of the term 'kill' on a menu, and the double-entendre of
'finger' on a unix machine. The LAST group of people I would turn over
the task of writing an interface to the Internet, would be system
programmers of TCP/IP. Technical 'functions' may have to be standarized,
but let the end users, teachers, sysops, and server-adminstrators
select the interface for their local users. One of the reasons Phil
Becker's TBBS multi-user MSDOS program is so highly regarded and
profitable for e-soft, is that the sysop can put any ascii language,
functions, permissions on the screen one wants -without have to code or
recompile the code. I was the first BBS outside Aurora to run that
software, circa 1982 - and my own system became rather famous partly
because I was able to carry out an 'Old Colorado City Main Street -
Rogers Bar' consistent metaphor, in ascii text only (before the Mac
point click desktop visual metaphor was around). Nobody ever got lost on
my system when they saw 'Go <B>ack to Main Street' four menus deep in
the system. It was my 'mental picture' of a small town approach to
user-unfriendliness of functional computer terms. It is also why we
selected the Remote Access version of Fido-capable software (one of 11
versions). Because WE, and then later those whose systems we set up, not
the original programmer in hard-code, could create within very wide
limits - the 'look and feel' of the interface. (Including dual NAPLPS
graphic-Ascii sub-interfaces). Linguists, poets, video-producers and
artists know more about human information interfaces than programmers)
Jack Rickard of Boardwatch Magazine has long contended that one
great future role of 'BBSs' will be to be the 'front ends' of the
Internet/NREN. I think he makes a very important point, even though if
you think about it, that is nothing more than saying that the individual
tailoring which the owner of a personal computer can do, ought to be
extended to networks. We better think much more clearly about what has
to be 'standardized' and what should be left to the user or lowest-level
system administrator. Give them the tools to fashion their own interface,
not the already-carved electronic Totem Pole.
So when, say 8 high school students simultaneosly use a school
BBS, even a 8 port TBBS or Fido, or SCO/Interactive $5,000 Unix system
with one Digiboard, or connected to a classroom LAN - and 10 such high
schools are doing the same thing across a city or county - and then the
sum of all their work, or the selected-newsgroup feed, passes through
the local Internet/NREN server at 19.2 baud at intermittent times
sequentially, not simultaneously (thank goodness for auto-redial)
through 1 or 2 ports on the central 'community' server, it has always
struck me as more sensible, economic, and socially controllable than
adding 80 ports to that server to handle the same level of traffic.
That does not replace Tom Grunders Cleveland Free Net 'front
end' model, but simply extends it by other means and protocols to the
ends of the earth. When in Cleveland (or other large cities where
a powerful full-time administered, or institutional-affiliated,
Internetted machine can be supported) do as the Clevelanders do;
when in Wisdom, Montana, use the local Apple II running Fredmail,
as very small town Montana'ns can and have to ($$$) do.
In Montana - the model I have described is going to (and already
has, through Big Sky Telegraph's system itself) get a far percentage of
schools, teachers, and students on the international Internet line
faster and cheaper than anything the University of Montana is going to
do starting at the other end with their campus mainframe. Its already
got them - while the state wide 'higher education' network that can
interconnect to the Internet and Bitnet is still in the talking and
under-utilized T-1 line stage.
My own 13 years experience online 5 hours a day on systems from
early Source, Compuserve, one line, and multi-line BBSs, as well as
designing, running, adminstering hobby, educational non-profit, as well
as commercial such networks has told me that it is far easier to deal
with actual social misbehavior on many small systems than a few big
ones. In one three year period I had 50,000 logins by 8,600 different
users who left 26,000 messages on my TBBS BBS and it only cost me 30
minutes a day, and 1-2 hours a weekend, less than $10 a month phone
cost, to 'administer' the system and be active myself on it - as a wide
open, self-registering, unlimited use, and serious-purpose system. I
experienced virtually NO serious juvenile-behavior problems I couldn't
handle by more than the same observation-of-behavior I have as a
parent, or would as a teacher, or school computer coordinator. The
smaller the 'community' online or off, the easier it is to deal with
a-social behavior.
I was able, and did, read all posted messages. (The Electronic
Privacy Act of 1986 insured that *my* behavior as a Sysop with such
access would be kept within social bounds). Not possible on large
systems. Not even my Unix AKCS conferencing/some-news-group system.
All of which factors considered - user behavior, costs at all
points, progressive learning curve by all which will be required
(technical, economic, and online-cultural) - tells me that rather than
this willy-nilly movement toward directly connecting all K-12 kids
to the Internet, that a far broader, decentralized, distributed,
protocol, not 'rule' connected model is indicated and ought to be
the focus of discussion and planning.
Is it because so many of those debating this issue know *only*
the Internet that they can think only in its terms? (I have observed
similar thought-encapsulation among those who 'only' know BBSs, or
MacIntoshes, or Bitnet, or Dec Terminals, or LANS)
So come January we expect to have another 16 week formal course
in Chaos Math and Physics taught/learned online where Dr. Johnston of
MIT, directly on the Internet through the Plasma Fusion Center Vax will
be 'communicating' with 20-30 high and junior high school students in
Colorado, Wyoming, and Montana, each of them on one of perhaps 8
different local systems, some Internet, some Usenet, some Fidonet, maybe
one Frednet, and Point net, where teachers will be able to
observe,participate, and where necessary intervene when social problems
involving their kids arise. And it will not be limited to ascii, but in
graphical (such as fractal images) and non-ascii symbol (calculus) form.
And during it the students may run, for one part of the course,
a supercomputer. (which, unless I don't know the limits as well as
capabilities of such utilities as remote executable functions of TCP/IP
and Unix - could also be done by sending command-messages and not just
by being logged directly into a TCP/IP machine - by long distance call
to Colorado from Wyoming/Montana).
So we certainly intend to have K-12 students 'use' the Internet,
and because of the decentralized and distributed method of access, don't
really worry excessively about rancher's kids running X-rated data
streams during parent's night at the Cody High School.
Anything really wrong with this model of ever-upgrading (people,
machines, and institutions) 'net of appropriate nets' approach to
massive future K-12 national networking?
----------------------------------------------------------------------
Dave Hughes Old Colorado City Communications
"Its better to light one pixel than cursor the darkness"
daveh@csn.org
dave%oldcolo@csn.org
Fidonet 1/128/67 or Point 67.1 719-636-2040 voice
----------------------------------------------------------------------
