Shareware and Public Domain: Software Distribution Before the Internet

Abstract

Before the web, getting software meant getting a disk, and a whole economy grew up around who copied it, who mailed it, and who (occasionally) paid for it. Between the early 1980s and the mid-1990s, shareware and public domain distribution turned copying from a threat into a marketing channel: authors gave their programs away and asked users to pay if they liked them. The model produced million-dollar one-person companies, the most lovingly curated software library of its era (Fred Fish’s 1,120 Amiga disks) and the business plan that launched Doom. Then the internet made distribution free for everyone, and the scarcity the model had monetized evaporated. Shareware’s DNA survives in every free demo, episodic release, and freemium app.

Before the Product: Software as Something You Shared

In the beginning, software wasn’t sold at all. SHARE, founded in 1955 by users of the IBM 704, and DECUS, the DEC user society founded in 1961, maintained libraries of contributed programs that members exchanged freely, software was something that came with the machine, written by the people who used it. Only after IBM unbundled software from hardware pricing in 1969 did programs become products, and with the microcomputer boom of the late 1970s the question of Bill Gates’ Open Letter to Hobbyists (is copying software sharing or stealing?) split the young industry.

The hobbyist answer was the public domain library: user groups and mail-order operations like PC-SIG in Sunnyvale collected programs whose authors had renounced payment entirely, catalogued them, and sold the disks (not the software) for a few dollars a piece. The software was free; the copying, cataloguing, and postage were the service. This distinction (charging for distribution, not for the program) would become the economic backbone of everything that followed.

The Three Fathers of Shareware

Shareware proper was invented twice in the same year, on opposite corners of the United States, by two people who then found each other.

Andrew Fluegelman, a San Francisco lawyer-turned-editor (he became founding editor of PC World and later Macworld), wrote PC-Talk, a communications program for the IBM PC, in 1982. Rather than seek a publisher, he distributed it freely and asked satisfied users to send $25, a model he called freeware, a term he trademarked. He described the scheme as “an experiment in economics more than altruism.”

Jim Knopf (who published as Jim Button), an IBM employee in Bellevue, Washington, released his flat-file database PC-File the same year under the label “user-supported software.” When Knopf and Fluegelman discovered each other’s experiments, they coordinated: matching prices, and mutual plugs in each program’s documentation. Knopf’s hobby soon out-earned his IBM salary; his company ButtonWare became one of the first proofs that voluntary payment could sustain a real business.

The name that stuck came from the third pioneer. Bob Wallace, one of Microsoft’s earliest employees, left in 1983 to write the word processor PC-Write and marketed it under a term he judged less proprietary than Fluegelman’s trademarked “freeware”: shareware. Wallace added a twist that anticipated affiliate marketing by a decade, registered users who passed on copies earned a commission when their copies were registered in turn.

The movement organized itself in 1987 with the founding of the Association of Shareware Professionals (ASP), which standardized what “shareware” meant (evaluation before payment, no crippled functionality) and ran author forums and disk-vendor certification.

The disappearance of Andrew Fluegelman

Fluegelman did not live to see the model he co-invented conquer the games industry. Suffering from cancer and severe colitis, he vanished in July 1985; his car was found near the Golden Gate Bridge. He was 41, and is presumed to have taken his own life.

Fred Fish and the Most Famous Disks in Computing

Nowhere did the culture of free distribution flower more completely than on the Commodore Amiga, and its gardener was a soft-spoken Unix programmer named Fred Fish (1952–2007).

Fish bought an Amiga 1000 shortly after the machine’s 1985 launch and noticed a problem: excellent freely distributable software existed, but scattered across BBSes and user group libraries with no curation and no quality control. Starting in 1986, he began assembling floppy disks of the best freely redistributable Amiga software he could find, tested, organized, and documented, with source code included whenever authors allowed it. He mailed master disks to user groups around the world, which duplicated them locally; anyone could copy them, and Fish charged only a nominal fee for disks ordered directly from him.

The community named them before he did: at a New Jersey Amiga user group meeting, developer Perry Kivolowitz dubbed them the Fish Disks, and the name (helped by the fish logo that soon adorned the labels) became a standard unit of the Amiga world. Software was cited by disk number (“it’s on Fish 342”) the way academics cite journals. By the time the series ended in 1994, it comprised 1,120 disks, a one-man, pre-internet package repository, complete with consistent metadata, that anticipated the curated software archive by a decade. As floppies gave way to CD-ROMs, Fish continued the collection on his FreshFish CD-ROM series.

The Fish Disks mattered because of what curation added. A BBS download was a gamble; a Fish Disk was a guarantee, the program ran, the documentation was there, and the source often came along. For tens of thousands of Amiga users, especially in Europe where telephone-based downloading was expensive, the disks were the software ecosystem, arriving through the mail and multiplying at user group meetings like their namesake’s loaves-and-fishes counterpart. See Dead End: The Amiga for the platform whose community made this possible.

Fish’s second career is less known but at least as consequential: he was a long-time developer of the GNU Debugger (GDB), authoring much of its Unix System V Release 4 support, a bridge between the free-software culture of GNU and the freely-redistributable culture of the home computer world, embodied in one person. He remained an active developer, including work on AmigaOS 4, until his death from a heart attack in April 2007.

Aminet: The Archive Goes Online

The Fish Disks’ successor was born in Zurich. In 1991, computer science students at the University of Zurich’s informatics club set up an FTP archive of Amiga software on an Amiga 3000UX; in January 1992, student Urban Müller took it over and built it into Aminet, a globally mirrored, consistently catalogued repository to which authors uploaded directly. From roughly 1992 until about 1996, Aminet was the largest public software archive for any platform, and from 1995 its contents were pressed onto quarterly CD-ROM sets so that the (many) Amiga users without internet access could still reach it. Its structure (moderated uploads, standardized per-package description files, worldwide mirrors) prefigured the package archives every modern platform now takes for granted, from CPAN to the Linux distribution mirrors.

The Apogee Model: Shareware Conquers Games

Utility authors asked politely for money; game publishers found a way to make paying irresistible.

In 1987, Scott Miller of Texas hit on the insight that defined 1990s PC gaming: don’t give away a crippled program and ask for donations, give away a complete, polished chunk of a game and sell the rest. He split his DOS game Kingdom of Kroz into three episodes, released episode one freely on BBSes, and sold episodes two and three by mail order through his company Apogee Software. The “Apogee model” made copying work for the publisher: every pirated copy of episode one was an advertisement for episodes two and three.

The model’s greatest catch was a team of young Softdisk programmers (John Carmack, John Romero, Tom Hall, and Adrian Carmack) whom Miller recruited to publish Commander Keen in December 1990. Its runaway shareware success let them found id Software. Apogee published id’s Wolfenstein 3D as shareware in 1992; in December 1993, id self-published Doom by uploading its first episode to a university FTP server and letting the world do the distribution. The most influential PC game of its decade was marketed by giving a third of it away; see The PC Gaming Revolution. Epic MegaGames, founded by Tim Sweeney on the same shareware model, became today’s Epic Games.

The Distribution Machine

What made all of this work was a physical and social infrastructure that is now almost invisible in retrospect:

BBSes were the primary upload channel, the file areas of the BBS era were substantially stocked with shareware and PD software. The files themselves usually traveled as .zip archives made with PKZip, itself shareware; its author’s story is told in Phil Katz and the Zip Format.
Disk vendors ran mail-order catalogues of shareware disks, legally selling the copying service. The ASP certified vendors who followed the rules (clearly labeling that registration fees were separate).
Magazine cover disks, especially in Europe, put shareware and PD software on newsstands; for Amiga and Atari owners, the cover disk was many users’ main software supply. The same disks, BBSes, and copy parties also carried the productions of the demoscene, free distribution as a culture, not just a business model.
CD-ROM compilations (“shovelware”) crammed thousands of shareware titles onto a single disc in the early 1990s, briefly a gold rush, quickly a race to the bottom in quality.
Retail shareware racks in software stores sold $3–5 disks of software that was free, the public domain library’s economics, scaled to retail.

Dead End: What Killed Classic Shareware

The model died of the thing it had been built to overcome: distribution cost.

Shareware’s genius was arbitrage. Physical software distribution (publishers, boxes, shelf space) was expensive and gatekept; copying disks was nearly free. Shareware routed around the gatekeepers by making users the distribution network, and monetized the result on the honor system, tolerable economics, because even a small registration rate was pure margin on zero distribution cost. Authors lived with the fact that only a small minority of users ever paid.

The web destroyed both halves of the bargain. Once anyone could download anything directly from the author’s site, the user-to-user copying network stopped being valuable; there was nothing left to “share.” And once distribution was free for paid software too, publishers adopted shareware’s best trick (the free taste) without the honor system: time-limited trialware, feature-limited demos, and licence keys replaced “please send $25 if you like it.” The ASP’s ideal of full-featured, trust-based evaluation gave way to crippleware; in the late 1990s and 2000s, download portals buried the remnants under ad-supported installers and bundled toolbars, and “shareware” decayed from a business model into a warning label.

But the model didn’t fail so much as dissolve into the mainstream. The free first episode became the game demo and the free-to-play tier; episodic pricing became DLC and season passes; “pay if you like it” became freemium and the in-app purchase; the curated, mirrored, metadata-rich archive that Fish assembled by hand and Aminet put online became the app store and the package repository. Every “try free, upgrade to pro” button is running Bob Wallace’s play from 1983.