Linked by Barry Smith on Wed 26th Nov 2003 18:11 UTC
It seems to me that a lot of attention lately in the commercial Linux development area has concentrated on either large enterprise customers, or wooing the home user who can barely turn a computer on. Even distros claiming to offer the perfect solution for both ends of the spectrum don't quite seem to fit what I am looking for.
Permalink for comment
To read all comments associated with this story, please click here.
I realize that there is always some subjectivity to any review, but to me what stands out about this review is that the reviewers grading "system" is almost entirely arbitrary. For instance, this reviewer gave Lindows a +1 for installing correctly on one system, and a -1 for failing to detect that the onboard video was turned off in the BIOS on a second system. If you're going to compare installation on various different computers, and award points based on that, the results are statistically meaningless unless you have a large number of computers to do the trials on. Using just two is like evaluating students performance in an exam with just two true/false questions.
Another example: the reviewer gives Lindows a +1 for detecting and configuring his Winmodem, something which every other distro failed to do. Then the same reviewer did not take off a point when Lindows failed to detect or configure his US Robotics modem on the second system, because "I have only seen one Linux distro to date that was capable of finding the USR modem". In other words, Lindows gets a +1 in one case because of the failings of other distros, and avoids a -1 in another case because of, err, the failings of other distros.
A third example: Lindows got a -1 for failing to recognize a cheap IBM webcam, because it is "common as dirt". This time the points were awarded based on the authors level of expectation, not on the performance of other distros, since there was no mention of whether other distros succeeded in detecting and configuring this webcam or not.
Using other distros for comparison is a reasonable scheme. Using just user expectations is a more subjective scheme, but may have its place in a "Linux wish-list" sort of review. Mixing the two randomly, however, makes for a poor review.
I realize that there is always some subjectivity to any review, but to me what stands out about this review is that the reviewers grading "system" is almost entirely arbitrary. For instance, this reviewer gave Lindows a +1 for installing correctly on one system, and a -1 for failing to detect that the onboard video was turned off in the BIOS on a second system. If you're going to compare installation on various different computers, and award points based on that, the results are statistically meaningless unless you have a large number of computers to do the trials on. Using just two is like evaluating students performance in an exam with just two true/false questions.
Another example: the reviewer gives Lindows a +1 for detecting and configuring his Winmodem, something which every other distro failed to do. Then the same reviewer did not take off a point when Lindows failed to detect or configure his US Robotics modem on the second system, because "I have only seen one Linux distro to date that was capable of finding the USR modem". In other words, Lindows gets a +1 in one case because of the failings of other distros, and avoids a -1 in another case because of, err, the failings of other distros.
A third example: Lindows got a -1 for failing to recognize a cheap IBM webcam, because it is "common as dirt". This time the points were awarded based on the authors level of expectation, not on the performance of other distros, since there was no mention of whether other distros succeeded in detecting and configuring this webcam or not.
Using other distros for comparison is a reasonable scheme. Using just user expectations is a more subjective scheme, but may have its place in a "Linux wish-list" sort of review. Mixing the two randomly, however, makes for a poor review.
-nubuddy