This is an archived snapshot of W3C's public bugzilla bug tracker, decommissioned in April 2019. Please see the home page for more details.
From Rotan Hanrahan <Rotan.Hanrahan@MobileAware.com>: If validator gives boolean result, then "false" is so negative that you are not encouraged to try again. But if validator gives a ranking (e.g. percentage) then you are encouraged to improve your result. Eventually you get 100% and are conformant. [...] At a minimum the errors should be classified, and a weighting assigned to each category according to perceived severity and difficulty of correction. Add all of the (weighted) errors in the page. Divide the sum by a factor proportional to the log of the size of the document (so that larger documents do not have excessively large error scores, since the correction of one error in a document can easily be replicated for all such errors that occur in the document). When your score is Zero, you are conformant. Of course, this formula is quite simplistic, but would be easy to implement and, regardless of its actual interpretation, if the scoring system is consistent and has the property of always decreasing whenever a correction is made, then it should be useful. Anyway, the marketing people won't care as long as their scores beat the competition :-)
Target 0.7 for review in that timeframe.
Pushing to "1.0". This won't make it for 0.7.0.