Needless to say, I have no sympathy for some of the arguments that have been presented here. Hardware (including memory) prices are decreasing. Most software for Linux these days is being written for (1) the command line, or (2) the X Window System. Console-based full-screen applications are constituting a diminishing proportion of total software development effort, as a review of http://freshmeat.net/ will show. In terms of screen reading functionality, the Emacs terminal emulator is more advanced than any of the primitive screen readers for the Linux console which have so far been developed, largely due to the significant design and development effort that has been introduced into Emacspeak and the fragmentation of the Linux screen reader work into at least four separate programmes: SVPro, SpeakUp, SCReader and UltraSonix, with only the latter being reasonably well advanced in terms of configurability and features. The situation is as follows: 1. Today it is possible to access the shell and text-based applications by any of the following means: (a) a separate computer acting as a terminal; (b) the Emacspeak terminal emulator; (c) under development, UltraSonix with a terminal emulator; (d) under development: the Gnome terminal emulator; (e) BrlTTY with a braille display (and there is interest in adding speech output to this software as well). By contrast, if one wishes to access an X application, the choices are (1) the Gnome speech server, which is still under development, and (2) UltraSonix, which requires significant work in order to be ready for every-day use. It is clear where the access barriers are and where the priorities should lie, namely with (1) the development of open-source user software and user interfaces in the manner described in my previous message, and (2) access to the X environment and its applications. This becomes all the more important once it is realised that most interactive software these days is being written for the X Window System. The objections which have been raised against this approach are: (1) memory and disk space limitations, which are ultimately a legacy problem that won't affect most Linux users, especially in the future as X becomes (indeed it has already become) a standard basis for the user interface; and (2) dissatisfaction with Emacspeak for spurious reasons such as the key bindings, which of course can be readily remapped as needed.