Richard Stallman, an MIT AI Lab programmer turned campaigner, launched the GNU Project so that people using a program could change it and share it with one another. Sam Williams's 2002 biography follows that campaign from the lab's editor hacks to the arrival of open source. Williams built it from interviews with Stallman and with colleagues and critics whose memories often conflict. In one case, the likely other party to Stallman's printer story told Williams he recalled nothing of it. The book is worth reading because it shows the work a moral commitment needed before it could hold: engineering that let people use each other's changes, and licence terms that kept freedoms attached as copies moved.
Emacs supplies the first mechanism. Stallman extended an existing keystroke-by-keystroke TECO mode so users could save command macros and call them from key combinations. Private collections multiplied until a hacker at a colleague's terminal first had to decode that colleague's macros. Guy Steele began charting the common commands. Stallman, the lab's TECO maintainer, took over the work, and with Steele and others he standardized the commands controlling the screen while leaving extension open. Customization and collaboration needed separate answers, and Stallman's answer for collaboration was social. The Emacs Commune let anyone modify and redistribute the editor on condition of returning every extension to him.
Williams then traces that rule into copyright. UniPress threatened to enforce its copyright on James Gosling's Emacs, and the episode showed how little the informal commune bound anyone outside Stallman's circle. The licences for GNU Emacs, and then the General Public License of 1989, kept copyright with authors and used it to give recipients freedom to modify and redistribute. Under the GPL, anyone distributing a covered version had to grant recipients the same permissions and ensure they received or could obtain the corresponding source. Private modifications no longer had to be published or sent to Stallman, who told Williams that funnelling changes to one privileged developer had been wrong. Berkeley's licence from the same year allowed proprietary derivatives, which shows the GPL was one route among several.
Since the GPL separates freedom from price, commerce appears in the book as concrete exchange. Rich Morin charged a fee for distribution tapes he compiled for a Sun user group. Once strong programs carried the GPL, he no longer had to telephone authors about permissions, so he assembled tapes in half the time and turned the sideline into Prime Time Freeware. Michael Tiemann and John Gilmore launched Cygnus Support to sell contracts customizing and supporting GNU programs, while other firms could build services on the same GNU base. Williams reports both ventures through their participants' recollections, as particular arrangements for charging around freely copyable code.
Permission to fork settles what anyone may do with the code. It leaves open who decides what goes into the shared version. Stallman rejected patches, including one from Jeremy Allison, and Lucid's Emacs developers split away after he declined to fold their changes in. Linux developers impatient with the glibc maintainers considered a separate library, a fork the book says would have cost compatibility and contributors. Bruce Perens distanced Debian from the Free Software Foundation to escape what he called micromanagement, while keeping the GNU/Linux name. A licence giving everyone the right to leave did not remove authority or coordination costs.
Those tensions sharpen in the chapter on 1998. After Netscape announced plans to publish its browser source, Christine Peterson suggested the term open source. Tim O'Reilly's summit then adopted it for the press because executives kept hearing free as zero cost. Tiemann recalls preferring a pitch that sounded sensible to business over one that sounded morally righteous. Stallman judged that the term conveyed technical advantages but invited speakers to play down freedom, so he kept saying free software. His ethic concerns what users owe one another. He came to regard refusing a fellow programmer's request for source as a breach of the Golden Rule, so in his view the reason for sharing has to hold even when convenience pulls the other way. Williams lets Eric Raymond call Stallman territorial and Ian Murdock credit his consistency, which leaves readers to weigh the ethic against its costs.
The epilogue adds one last case. Williams leaned toward the Open Publication License to keep competitors from printing the same book. Stallman argued for the GNU Free Documentation License, and Williams accepted it in exchange for more interviews, though Stallman made his help with publicity depend on the finished content. The text remains Williams's copyright under that documentation licence and can be read at Project Gutenberg. Read it to see why free software and open source could share most of their code while their advocates fought over its purpose, as told by participants who still disagreed when Williams questioned them.