Timo and linear coding

The issue of linear and nonlinear coding has received a fair bit of discussion among the Usenet newsgroups lately, as a result of a dust-up between me and Timo Autiokari. Timo contradicts the most basic principle of the Gamma FAQ. He claims linear intensity coding must be used, even with 8 bits per RGB component (or perhaps especially in 8 bits).

The skirmish started on 1998-02-25, when Timo published an open letter in a Usenet posting, some inflammatory remarks concerning my Gamma FAQ. Some of those postings contain serious misunderstandings about image coding, and even a few personal attacks. These remarks are sufficiently grave that I felt obliged to respond. You can read about the Timo incident, and you can read about Timo's technical argument.

I have prepared a series of notes that comprise an introduction to Linear and nonlinear coding. At the bottom of several of these notes, you will find a link to the version of that issue as described by Timo. If you're a stoic, read just the Timo thread among those notes:

Timo believes it desirable to have a perfect mapping from intensity in to intensity out.
Timo considers nonlinearity to be unnatural.
Timo mistakenly considers Calibrate to be synonymous with Linearize.
There are technical errors in Timo's argument concerning 12 bits.
Timo fails to consider Weber's Law.
Timo falls victim to Proof by example

Nonlinear coding is accommodated - or required - in PhotoCD, PNG, JPEG, JFIF, sRGB, PhotoFlash, and all of the major high-quality file formats of all the major digital photography vendors. Timo claims that I have mislead the leading scientists at the "major players." By this, I think he refers to the proponents of the file formats that I mention: Kodak, Hewlett-Packard, Microsoft, Adobe, Silicon Graphics, C-Cube, and others.