FidoNet


 __
 / \
 /|oo \
 
 _`@/_ \ _
 | | \ \\
 | )
 ______ |__U__| / \//
 / FIDO \ _//|| _\ /
 John Madill

FidoNet logo by John Madill

FidoNet is a worldwide computer network that is used for communication between bulletin board systems. It uses a store-and-forward system to exchange private and public messages between the BBSes in the network, as well as other files and protocols in some cases.
The FidoNet system was based on several small interacting programs, only one of which needed to be ported to support other BBS software. FidoNet was one of the few networks that was supported by almost all BBS software, as well as a number of non-BBS online services. This modular construction also allowed FidoNet to easily upgrade to new data compression systems, which was important in an era using modem-based communications over telephone links with high long-distance calling charges.
The rapid improvement in modem speeds during the early 1990s, combined with the rapid decrease in price of computer systems and storage, made BBSes increasingly popular. By the mid-1990s there were almost 40,000 FidoNet systems in operation, and it was possible to communicate with millions of users around the world. Only UUCPNET came close in terms of breadth or numbers; FidoNet's user base far surpassed other networks like BITNET.
The broad availability of low-cost Internet connections starting in the mid-1990s lessened the need for FidoNet's store-and-forward system, as any system in the world could be reached for equal cost. Direct dialing into local BBS systems rapidly declined. Although FidoNet has shrunk considerably since the late 1990s, it has remained in use into 2016 despite internet connectivity becoming more widespread.

History

Origins

There are two major accounts of the development of the FidoNet, differing only in small details.

Tom Jennings' account

Around Christmas 1983, Tom Jennings started work on a new bulletin board system that would emerge as Fido BBS. It was called "Fido" because the assorted hardware together was "a real mongrel". Jennings set up the system in San Francisco sometime in early 1984. Another early user was John Madill, who was trying to set up a similar system in Baltimore on his Rainbow 100. Fido started spreading to new systems, and Jennings eventually started keeping an informal list of their phone numbers, with Jennings becoming #1 and Madill #2.
Jennings released the first version of the FidoNet software in June 1984. In early 1985 he wrote a document explaining the operations of the FidoNet, along with a short portion on the history of the system. In this version, FidoNet was developed as a way to exchange mail between the first two Fido BBS systems, Jennings' and Madill's, to "see if it could be done, merely for the fun of it". This was first supported in Fido V7, "sometime in June 84 or so".

Ben Baker's account

In early 1984, Ben Baker was planning on starting a BBS for the newly forming computer club at the McDonnell Douglas automotive division in St. Louis. Baker was part of the CP/M special interest group within the club. He intended to use the seminal, CP/M-hosted, CBBS system, and went looking for a machine to run it on. The club's president told Baker that DEC would be giving them a Rainbow 100 computer on indefinite loan, so he made plans to move the CBBS onto this machine. The Rainbow contained two processors, an Intel 8088 and a Zilog Z80, allowing it to run both MS-DOS and CP/M, with the BBS running on the latter. When the machine arrived, they learned that the Z80 side had no access to the I/O ports, so CBBS could not communicate with a modem. While searching for software that would run on the MS-DOS side of the system, Baker learned of Fido through Madill.
The Fido software required changes to the serial drivers to work properly on the Rainbow. A porting effort started, involving Jennings, Madill and Baker. This caused all involved to rack up considerable long-distance charges as they all called each other during development, or called into each other's BBSes to leave email. During one such call "in May or early June", Baker and Jennings discussed how great it would be if the BBS systems could call each other automatically, exchanging mail and files between them. This would allow them to compose mail on their local machines, and then deliver it quickly, as opposed to calling in and typing the message in while on a long-distance telephone connection.
Jennings responded by calling into Baker's system that night and uploading a new version of the software consisting of three files: FIDO_DECV6, a new version of the BBS program itself, FIDONET, a new program, and NODELIST.BBS, a text file. The new version of FIDO BBS had a timer that caused it to exit at a specified time, normally at night. As it exited it would run the separate FIDONET program. NODELIST was the list of Fido BBS systems, which Jennings had already been compiling.
The FIDONET program was what later became known as a mailer. The FIDO BBS software was modified to use a previously unused numeric field in the message headers to store a node number for the machine to which the message should be delivered to. When FIDONET ran, it would search through the email database for any messages with a number in this field. FIDONET collected all of the messages for a particular node number into a file known as a message packet. After all the packets were generated, one for each node, the FIDONET program would look up the destination node's phone number in NODELIST.BBS, and call the remote system. Provided that FIDONET was running on that system, the two systems would handshake and, if this succeeded, the calling system would upload its packet, download a return packet if there was one, and disconnect. FIDONET would then unpack the return packet, place the received messages into the local system's database, and move onto the next packet. When there were no remaining packets, FIDONET would exit, and run the FIDO BBS program.
In order to lower long-distance charges, the mail exchanges were timed to run late at night, normally 4 AM. This would later be known as national mail hour, and, later still, as Zone Mail Hour.

Up and running

By June 1984, Version 7 of the system was being run in production, and nodes were rapidly being added to the network. By August there were almost 30 systems in the nodelist, 50 by September, and over 160 by January 1985. As the network grew, the maintenance of the nodelist became prohibitive, and errors were common. In these cases, people would start receiving phone calls at 4 AM, from a caller that would say nothing and then hang up. In other cases the system would be listed before it was up and running, resulting in long-distance calls that accomplished nothing.
In August 1984, Jennings handed off control of the nodelist to the group in St. Louis, mostly Ken Kaplan and Ben Baker. Kaplan had come across Fido as part of finding a BBS solution for his company, which worked with DEC computers and had been given a Rainbow computer and a USRobotics 1200bit/s modem. From then on, joining FidoNet required one to set up their system and use it to deliver a netmail message to a special system, Node 51. The message contained various required contact information. If this message was transmitted successfully, it ensured that at least some of the system was working properly. The nodelist team would then reply with another netmail message back to the system in question, containing the assigned node number. If delivery succeeded, the system was considered to be working properly, and it was added to the nodelist. The first new nodelist was published on 21 September 1984.

Nets and nodes

Growth continued to accelerate, and by the spring of 1985, the system was already reaching its limit of 250 nodes. In addition to the limits on the growth of what was clearly a popular system, nodelist maintenance continued to grow more and more time-consuming.
It was also realized that Fido systems were generally clustered – of the fifteen systems running by the start of June 1984, five of them were in St. Louis. A user on Jennings's system in San Francisco that addressed emails to different systems in St. Louis would cause calls to be made to each of those BBSes in turn. In the United States, local calls were normally free, and in most other countries were charged at a lower rate. Additionally, the initial call setup, generally the first minute of the call, was normally billed at a higher rate than continuing an existing connection. Therefore, it would be less expensive to deliver all the messages from all the users in San Francisco to all of the users in St. Louis in a single call. Packets were generally small enough to be delivered within a minute or two, so delivering all the messages in a single call could greatly reduce costs by avoiding multiple first-minute charges. Once delivered, the packet would be broken out into separate packets for local systems, and delivered using multiple local free calls.
The team settled on the concept of adding a new network number patterned on the idea of area codes. A complete network address would now consist of the network and node number pair, which would be written with a slash between them. All mail travelling between networks would first be sent to their local network host, someone who volunteered to pay for any long-distance charges. That single site would collect up all the netmail from all of the systems in their network, then re-package it into single packets destined to each network. They would then call any required network admin sites and deliver the packet to them. That site would then process the mail as normal, although all of the messages in the packet would be guaranteed to be local calls.
The network address was placed in an unused field in the Fido message database, which formerly always held a zero. Systems running existing versions of the software already ignored the fields containing the new addressing, so they would continue to work as before; when noticing a message addressed to another node they would look it up and call that system. Newer systems would recognize the network number and instead deliver that message to the network host. To ensure backward compatibility, existing systems retained their original node numbers through this period.
A huge advantage of the new scheme was that node numbers were now unique only within their network, not globally. This meant the previous 250 node limit was gone, but for a variety of reasons this was initially limited to about 1,200. This change also devolved the maintenance of the nodelists down to the network hosts, who then sent updated lists back to Node 51 to be collected into the master list. The St. Louis group now had to only maintain their own local network, and do basic work to compile the global list.
At a meeting held in Kaplan's living room in St. Louis on 11 April 1985 the various parties hammered out all of the details of the new concept. As part of this meeting, they also added the concept of a region, a purely administrative level that was not part of the addressing scheme. Regional hosts would handle any stragglers in the network maps, remote systems that had no local network hosts. They then divided up the US into ten regions that they felt would have roughly equal populations.
By May, Jennings had early versions of the new software running. These early versions specified the routing manually through a new ROUTE.BBS file that listed network hosts for each node. For instance, an operator might want to forward all mail to St. Louis through a single node, node 10. ROUTE.BBS would then include a list of all the known systems in that area, with instructions to forward mail to each of those nodes through node 10. This process was later semi-automated by John Warren's NODELIST program. Over time, this information was folded into updated versions of the nodelist format, and the ROUTES file is no longer used.
A new version of FIDO and FIDONET, 10C, was released containing all of these features. On 12 June 1985 the core group brought up 10C, and most Fido systems had upgraded within a few months. The process went much smoother than anyone imagined, and very few nodes had any problems.