The Cathedral and the Bazaar

"The Cathedral and the Bazaar" is a 1997 essay by Eric S. Raymond that contrasts two models of software development, drawn from Raymond’s experience building the fetchmail mail retrieval utility. It became one of the founding texts of the open-source movement and supplied much of the vocabulary still used to argue for open development today.

The essay names its two models after their contrasting styles of construction.

  • The cathedral model: software built like a cathedral, "carefully crafted by individual wizards or small bands of mages working in splendid isolation, with no beta to be released before its time." Raymond associates this style with traditional commercial development and with early free-software projects such as GNU Emacs and GCC, which released source code but developed it behind closed doors between infrequent, polished releases.
  • The bazaar model: development conducted in the open, in full view of a large, loosely coordinated crowd of contributors, with source released early and often, and outsiders drawn in to find and fix problems as they appear. Raymond took the Linux kernel, and Linus Torvalds’s habit of shipping new kernels sometimes more than once a day, as the model’s exemplar.

Linus’s Law

The essay’s central claim is a maxim Raymond names after Torvalds: "given enough eyeballs, all bugs are shallow." A bug that stumps its author may be obvious to some other contributor who reads the code with different assumptions or hits it under different conditions. The bazaar model recruits a large, diverse population of co-developers and users as unpaid debuggers, which Raymond argues finds and fixes defects faster than a small closed team could manage alone, however skilled. This is the essay’s most direct architectural consequence, and it is the reasoning behind release early, release often – frequent releases give the crowd more chances to exercise the code and report back.

Influence

The essay is widely credited with shaping the thinking behind the 1998 Netscape decision to release the source code of its browser, which led to the Mozilla project, and with popularizing the term "open source" itself over the more ideologically loaded "free software." It remains one of the most-cited justifications for open, distributed development as a working engineering strategy rather than a purely ethical stance.

The essay’s temperament sits alongside Richard Gabriel’s "worse is better", published a few years earlier. Both essays argue, from different angles, that a rougher process open to real use and outside contribution tends to beat a more controlled, closed process aimed at up-front perfection.

See also

References