FidoNews 5-03                Page 4                   18 Jan 1988


     Gateways to the future:
     Concerning FidoNet, Usenet, and the Future of Public Access
     Wide Area Networking              -- Doug Thompson, 221/162


          It's New Years (actually a little past) and the time for a
     review of the past 365 days and some forecasts for the next.

          I initially sat down to write this the week *before* the
     Alternet announcement in Fidonews. That event, it seems, has
     suddenly eclipsed most other discussions and brought a number of
     major issues into sharp relief. I suppose thanks is due the
     Alternet folks -- and I can't help wonder if this might not be
     part of their intent?

          The very fact of a having a choice of a technically similar
     alternate network provides an impetus for FidoNet to come to
     grips with what it is, and what it is not. In an area of such
     phenomenal growth as wide area computer networking, we must
     wrestle just as seriously with what we are to become, and what we
     wish not to become.

          There are three things I want to talk about:

          1) Growth -- the Future
          2) Usenet Gateways
          3) Expanded public Services
          4) Our Financial Base

          1) Growth rates

          First some numbers.  The oldest nodelist I have is from May
     2, 1986. That's 19 months ago. (from time of writing, Dec. 86)

     May 2        1986      789
     June 19      1987     1523 + 13 mos
     Dec 18       1987     2275 + 6 mos

     Avg monthly growth rate May '86 June 87=   56.46 nodes per month
     Avg monthly growth rate June - Dec   86=  125.33 nodes per month

          Over the past 6 mos FidoNet has grown at the rate of 125.33
     nodes per month. If this rate continues for the next year, we
     will see nearly 4,000 nodes in FidoNet by next Christmas. The
     limited stats here also suggest that the rate of growth more than
     doubled between May of 86 and December of 87. If that increase in
     the rate of growth continues, we could conceivably find ourselves
     with five or six thousand nodes by next Christmas.

          FidoNet seems to be doubling in size roughly every year. If
     anyone has more thorough stats drop me a line!

          Tom Jennings' hobby in 1984 has grown beyond what I imagine
     were his wildest expectations at the time. It seems to be an idea
     whose time has come, and is being more widely recognized as a
     good idea all the time.
     FidoNews 5-03                Page 5                   18 Jan 1988


         -- Rapid growth means most of us are relative newcomers --

          Growth of this pace inevitably causes strains, and we see
     lots of those in FidoNet. Having had some academic training in
     history, I grew accustomed to looking at the "flow" of events
     over the years rather than simply isolated snapshots. Among other
     things shown by these stats is that at least half the sysops in
     FidoNet have been participating less than 18 months.  Given that
     there is a drop-out rate as well as a growth rate, it may be that
     about half have been participating less than one year and about
     75% less than two.  That makes the vast majority of us relative
     newcomers, and our view of FidoNet is short enough to be called a
     snapshot.  But let's try to develop some snapshots into a movie.
     The figures in use here are rough approximations.  We do seem to
     have an exponential growth curve though. At current growth rates,
     by 1990, we could well have 30,000 nodes!

          I am not saying this *will* happen for certain, but I am
     persuaded at the moment that it is likely to happen. Forces which
     will encourage continued growth include:

          - declining costs and increasing capabilities of hardware
          - better and more user friendly software and interfaces
            which help non-experts jump in.
          - recent expansion into new countries and new continents.
          - self-propulsion: e-mail's value increases with the number
            of people you can reach. The bigger the net becomes the
            more the incentive to join.
          - Gateways to other networks making FidoNet an access point
            to many other and larger networks.
          - faster modems which reduce phone bills, making
            communication cheaper.

          I want to think about the implications of some of these
     things.  Rapid growth means that we will continue to have a
     majority of sysops with relatively brief experience in the net.
     We will remain "amateur" in more ways than one.

          I want to think about what it means to move from a hobbyist
     playing with an idea that society at large doesn't understand to
     the role of an operator in a world-wide computer network which
     more and more people will be depending on as an important channel
     of communication.

          I want to think about the economics of this: who is going to
     be carrying the cost of this international communication, and who
     might become interested to try to build a private business around
     the idea?


          2) Usenet Gateways

          During 1987 a number of FidoNet systems installed software
     which enables the exchange of mail and echoes with another and
     much larger computer network, Usenet.

     FidoNews 5-03                Page 6                   18 Jan 1988


          Usenet consists of about 10,000 machines around the world.
     Technically, Usenet is very similar to Fidonet and therefore
     gateways are not all that difficult.  There are vast differences
     in the character of the two nets however.

          Hardware in Usenet tends to be large computers owned by
     universities or businesses. While there are a small number of AT
     class IBM micros involved, for the most part the machines are DEC
     Vax minis.  Owners are commercial or educational institutions and
     the operators are employees of those institutions. The users are
     usually members of those institutions.

          Unlike FidoNet, the communication work of these computers is
     often a very small part of the computer's purpose.  The computers
     usually have other reasons to exist, and the communication work
     is incidental. To the extent that the owners support Usenet
     participation, they do so because of a perceived benefit for
     their employees and/or students.

          There is a lot of overlap in the type of people attracted to
     the two nets, and a great commonality of purpose. The major
     difference is that while the typical FidoNet sysop is an
     owner-operator who is really accountable to no one (except to be
     able to receive mail), the typical Usenet system administrator is
     an employee of an institution responsible to the accountants and
     managers for the system use.  He is also very much more
     responsible to his users, they pay him to keep the news and mail
     flowing.

          In these respects there is quite a difference, FidoNet
     consisting of "free-agents", paying their own way, and Usenet
     consisting of "corporation men" who do it as a job (although they
     may well greatly enjoy it). Where a FidoNet sysop pays for his
     mail (usually) a Usenet user or operator is rarely personally
     charged.  The institution absorbs the costs as part of the
     overhead.

          Another major difference is that Usenent is big. Some 10,000
     machines, the smallest of which are on par with the largest
     FidoNet systems. Usenet also has gateways to other networks,
     including ArpaNet, Bitnet, CSNet, and other national and regional
     networks around the world. The total number of users who have
     access to e-mail on all these networks is very hard to estimate,
     but I am pretty sure it is in the millions when you consider all
     the nets to which Usenet gateways.

          Accustomed to gatewaying to other networks, Usenet sites
     generally seem reasonably receptive to installing FidoNet
     gateways.  The perception generally seems to be that the value of
     Usenet increases in proportion to the number of e-mail addresses
     that can be reached. Fidonet represents some tens of thousands of
     addressable users, and is therefore worthwhile.

          My experience in "grovelling" for a news and mail feed from
     a Usenet site at a local university illustrates some other issues
     which are probably of general relevance. I was aided in my effort
     FidoNews 5-03                Page 7                   18 Jan 1988


     by the fact that I personally knew some of the Usenet operators
     and had other campus affiliations. The major concerns of the
     Usenet people were financial and technical. "How much is this
     going to cost us?" They wanted to know what kind of controls, if
     any, there would be on e-mail sent to them for forwarding, on
     their dime.  The second concern was technical reliability; "will
     your gateway cause us any hassles?"

          After providing much in the way of personal guarantees, a
     gateway was approved, although it was clearly pointed out that
     the approval could be suspended at any time. This raised a number
     of concerns for me. The first had to do with accounting. If mail
     was going to be moving through my system in both directions, I
     had to keep track of how much it was costing me, and how much it
     was costing the Usenet host. The second concern was my users. If
     I gave them access to Usenet newsgroups, what kind of problems
     might arise?

          There were two concerns about the users. I suppose we've all
     had problems of naive users not understanding that an echo area
     is not for private messages to local users. And most of us have
     experienced the twit problem, a user who is needlessly abusive
     and enters inappropriate messages. Should such material get into
     Usenet, I would receive the flak. I found myself in the
     uncomfortable situation of being personally responsible to tens
     of thousands of Usenet readers for whatever anyone might do on my
     system, as well as being responsible for any errors I might make
     :-).

          Dealing with Usenet differs from dealing with FidoNet both
     in scale, and in the level of professionalism expected. Usenet
     operators are professionals. In dealing with Fidonet they expect
     a similar professional attitude. Expensive errors are costing not
     just a few dollars of *personal* money, but corporate or
     institutional money.  An inconvenience can effect hundreds or
     thousands of people on whose approval the Usenet administrator's
     *job* depends.

          Running a gateway stretches the definition of hobby to the
     limit.

          Another very critical difference between FidoNet and Usenet
     is revealed here, the users. FidoNet is accessible to virtually
     anyone with a terminal and modem. Usenet is only accessible to
     the public at a very few "public access Usenet" sites. My first
     networking experience was in Usenet. I was so keen on it, I
     wanted to extend the capability to everyone, and became
     interested in creating a public access system. FidoNet nodes
     running gateway software appeared to be the cheapest way to
     establish public access to Usenet.  A relatively friendly user
     interface existed in Fido and Opus BBS systems, PD software was
     available, and the hardware needed was affordable.

          3) Financing

          Some serious difficulties are revealed in opening access to
     FidoNews 5-03                Page 8                   18 Jan 1988


     the public. Usenet is free. However, "free", in this context
     simply means that someone else pays for it.  Presently no device
     exists to charge back to the user the cost of services rendered.
     This is not because systems cannot charge users on a per message
     or per hour basis, but because the costs of any given message may
     be borne by hundreds, even thousands of different systems.  A
     large circulation newsgroup may end up occupying clusters on
     10,000 or more hard-disks, and be telephoned to sites all over
     the world.  As in FidoNet, Usenet sites do not generally charge
     each other for service.  At the moment this is quite workable.
     However, each Usenet site has a ceiling, a ceiling on funds and
     cpu cycles and disk space for the network.  General public access
     could, if it did become popular, come to swamp the network.  This
     will not happen tomorrow.  But, if network growth continues, it
     is a real possibility.

          Long term growth and general public access can only be
     accommodated through a system of financing that allows for some
     cost recovery when providing telecommunication services to
     others.  At the moment, extending services costs money. There is
     thus a financial *disincentive* to expanding services. If there
     were even the slightest financial *incentive*, and the money to
     buy new boxes, service could be expected to expand more rapidly.

          I would like to be in a position to "buy into" usenet,
     rather than beg into it. I'd like to advertise my tiny "public
     access" system and let it reach capacity. I'd like to be able to
     earn enough money from that user public to buy more machines and
     install more phone lines and bigger hard-disks. I'd like to know
     that in using the services of other systems, in Usenent and in
     FidoNet I was paying my way *and* making a financial contribution
     to those other services, and not stuck in a dependency
     relationship where every message was transferred as a favour.

          In short I'd like to see the system opened up, and service
     expanded. This cannot happen under current financing arrangements,
     where every expansion is a financial burden which must be
     limited.

          Reciprocity is basically the name of the network game. Each
     independent node or site in both networks provides value to
     others.  Some pay more than others, but it is in all our best
     interests to keep the connections open as best we can. The
     network's value *is* a function of how many people you can reach
     .  .  .  up to the saturation point.

          That saturation point worries me. With continued growth,
     traffic volume in echo mail (newsgroups in Usenent) will come to
     exceed the storage and throughput capacity of all but the largest
     systems.  Newsgroups also seem to have a maximum participation
     rate.  When the traffic volume in a newsgroup reaches hundreds of
     messages per day, it is impossible to keep up without some new
     kind of sophisticated "screening" software. The best one to date
     is the moderated "Digest". In this model, a moderator receives
     all submissions, and compiles them into a digest which is very
     similar in appearance and size to FidoNews. Unlike FidoNews,
     FidoNews 5-03                Page 9                   18 Jan 1988


     Usenet moderators usually cull inappropriate material (and often
     announce that they have done so).

          Optimizing the use of resources, and sharing the cost burden
     of expanded resources, can do much to increase the network
     capacity with minimum resource allocation. The price of that
     cooperation is a certain amount of autonomy.  A site cannot
     participate in a wide area cost-sharing plan to operate and
     manage collective resources without sacrificing a little
     autonomy.  Without that cooperation, however, many economies
     cannot be realized.  With it, the cost of inter-continental
     e-mail should drop to a few pennies per message.  The fee is
     nearly insignificant to the individual user.  The cost to
     individual gateway and zonegate systems, though, is crushing.

          I am not the first one to point out that continued growth
     will require centralized "network" services, rather than simply
     "site" services. Large machines with large disks and fast modems
     could be subsidized by the respective networks as store and
     forward facilities for mail. The use of leased lines and batched
     processing could bring the cost of reliable net-mail down to the
     vanishing point. But achieving those savings requires consider-
     able capital and a lot of labour.

          Individual sites could then subscribe directly, or groups of
     sites in a geographic area could pool their resources for a
     routed link to a central machine.

          In this manner, a skeletal device would be created whereby
     the actual costs of providing efficient services could be
     recovered.  It seems obvious that the funding and labour
     necessary to expand facilities to meet ever-increasing demand
     will exceed the capacity of individual site and node operators.
     There would also be a great deal more incentive for operators at
     every level to provide better service if the bulk of the costs
     were being borne by a large group of users, rather than the
     individual operator himself.

          One other fringe benefit -- based on the experience of
     Usenet -- an employee responsible to the net for his job is
     likely to provide a consistently high level of professionalism in
     network services.

          It would appear that the next few years will require us to
     solve these problems one way or another. The only alternative
     would seem to be to attempt to create a device to carefully limit
     the size of the network, and restrict access to a select few.
     While any particular network *could* do this, the public demand
     for electronic mail is unlikely to abate until virtually every
     phone line in the world has a telecommunication computer attached
     to it.

          Read that last sentence again.

          A question which concerns me greatly is that of addressing
     the issue of *organizing* and *financing* a public access e-mail
     FidoNews 5-03                Page 10                  18 Jan 1988


     network.  There appear to be two possible routes.

          One is commercial. Where there is a public demand, there
     will be businesses which attempt to meet that demand. The other,
     already foreshadowed to some degree in both Usenent and Fidonet,
     is the idea of a publicly owned, cooperatively managed,
     self-financing network.  The latter differs from the former in a
     number of ways.  While the end-product of the two might be quite
     similar, the public system is owned by the public, and its owners
     control it. Instead of subscribers, or customers, the user is a
     participant -- a citizen of the net, if you will -- rather than a
     customer of a service industry.

          A public network could provide a huge scope for volunteer
     participation. In doing so it would encourage innovation and
     reduce overall network expenses. Both these attributes would not
     only preserve some of the flavour of the amateur e-mail network
     we have grown to know and love, but would enable a public network
     to provide service at a cost well below that which a commercial
     enterprise would have to charge.

          The days of FidoNet being a small, exclusive club of
     dedicated hobbyists are numbered. The network is too good, too
     popular, too successful. Many who are not computer buffs in any
     real way now want in in order to *communicate*.

          It seems to me rather clear that some years down the line
     there will be an international wide-area networking system which
     is accessible to anyone for a fee. Either the present telecom
     using public will create it and keep it under public ownership
     (while keeping fees to a minimum) or private enterprise will
     provide it as a consumer service (maximizing profit, of course).


          -- Public vs. Private is not the same as Fee vs. Free --


          Why would a public, co-operatively owned system be better
     than a commercial one? This slips over into political philosophy,
     but there are a few things worth considering.

          1.) Because it doesn't have to earn a profit, it should be
     able to provide cheaper service.

          2.) Being already owned by the general public, it should be
     freer of government regulation than a strictly commercial
     enterprise.

          3.) Adverse government regulations will be less likely when
     the owners of the telecom utility and the voting public are
     precisely the same people.

          4.) Public control of network policy is much more likely to
     be meaningful under public ownership.

          5.) The user, rather than being a consumer whose only power
     FidoNews 5-03                Page 11                  18 Jan 1988


     is to not subscribe, becomes a full and equal participant, as
     active as s/he cares to be.

          6.) Continued dependence on a large amount of voluntarism
     could well not only encourage innovation and development but keep
     user fees absurdly low.

          7.) As the network grows in size, its influence in the
     computer industry and over government regulation nationally and
     internationally will also grow. In an age when high technology
     and regulatory decision-making is more and more removed from the
     ordinary life of the ordinary citizen, this would counter-balance
     present tendancies toward technological elitism and dependence on
     government "experts" to tell us what's good for us. The end-user
     would have a much larger influence.

          I guess my bias is out of the bag now :-)

          I am *very* much aware that many different perspectives
     exist on the large number of specific matters this paper touches
     upon.  My view is only one, and I'm not sure it is even the best.
     Too many of the discussions on these matters I've read and
     participated in seem to occur without an historical overview. I
     hope this paper can contribute to our thinking about where we are
     going in terms beyond simply "more of the same". The character of
     the network will inevitably change with growth, and with changing
     technology.  We do not have the luxury of choosing to keep things
     just as they are.  Events will overtake us and change our network
     whatever we choose to do, or not do.

          If a commercial service comes along, for instance, that
     offers access to news and e-mail cheaper than FidoNet itself,
     (distinctly possible) what point will there be to an amateur
     e-mail network?

          There is no such thing as "free" e-mail. There is only
     e-mail which you get someone else to pay for, or e-mail which you
     pay for yourself. Currently, the costs in FidoNet are very un-
     evenly distributed, and we depend not only on large volunteer
     efforts, but on large volunteer financial contributions. A big
     network, paid for only by a few of its most important links, does
     not strike me as having much potential

          I cannot begin to take the space to address each of the many
     reservations I can already hear being expressed. I can offer some
     points of departure for discussion, though.

          An International FidoNet co-op, funded by its members, could
     do two things almost immediately: it could provide cheap central-
     ized services such as echo-mail and software backbones and zone-
     gates.  It could begin to build an organizational infrastructure
     to reflect the will of the membership and influence the future of
     wide-area telecommunication.

          The latter involves many, many aspects, and could include
     such things as negotiations with other networks and standard-
     FidoNews 5-03                Page 12                  18 Jan 1988


     ization of gateway structures to assure universal access,
     negotiation with hardware manufacturers concerning standards and
     bulk discounts, establishing policies to prevent discrimination,
     injustice, and abuse, provide legal advocacy and defence in the
     murky waters of BBS liability, promote the expansion of network
     links to new parts of the world, sponsor public education efforts
     to promote intelligent use of e-mail and tele-conferencing,
     sponsor promising research in new areas, coordinate research
     internationally, liase with commerical and educational institu-
     tions where mutual benefit might accrue, study ways in which
     network services might be "sold" to subsidize cheap private
     correspondence, work toward international telecommunication
     standards and freedom of communication in other parts of the
     world .  .  .  etc.  etc.  etc.  And all in the name of the
     public, the public good, and the public's access to information,
     rather than solely in the name of profit, control, and restric-
     tion.

          There are some who would say that IFNA, with its elected
     Board, and wide-ranging, though largely undefined responsibility
     for the net could be that International Network Co-Op. I guess
     I'm one.

          We're riding a fast-rolling snowball here in FidoNet today.
     It's been five years since Tom Jennings' first preposterous
     experiments, and nine years since Usenent's first two sites made
     that first phone call. The snowflake has become a large snowball
     accelerating down the hill. What may be less apparent is that
     this snowball is on its way to becoming an avalanche.

          As we debate these matters I would really like people to
     devote 1% of their thought to where we might be in ten, twenty or
     thirty years.

          Forgive me for taking so long, if you have grown bored, and
     forgive me for leaving so much out, if you are still interested.
     In future weeks I hope to take time to probe some of the issues
     touched on here in greater detail (depending on the response to
     this piece).

          I'd be really happy to hear thoughts and opinions from
     readers. You can send me mail at any of the following
     addresses:

     -----------------------------------------------------------------------
     Fido      1:221/162 -- 1:221/0                         280 Phillip St.,
     UUCP:     !watmath!fido!221!162!Doug_Thompson          Unit B-3-11
     	  !watmath!orchid!imprint                      Waterloo, Ontario
     Bitnet:   fido@water                                   Canada  N2L 3X1
     Internet: dt@221.162.fido.waterloo.edu                 (519) 746-5022
     ------------------------------------------------------------------------
     watmath can be reached through utzoo and most backbone sites. My BBS
     number is 519-747-1332.

     -----------------------------------------------------------------