Bob Wyman in particular and others in general have argued eloquently for the utility of atom:modified in managing feeds and distinguishing between temporally-varying instances of the same thing.
I agree with the usefulness of doing what Bob and others want to do, and that they need to disambiguate successive versions of entries based on date. I don't think this is controversial or surprising.
I do not agree in the slightest that atom:modified is any more useful than atom:updated for these purposes. The only distinction between modified and updated is that there might be changes, not considered significant by the publisher, which occur between atom:updated- timestamped versions.
I'll repeat that. The only difference is that there might be changes not considered significant by the publisher.
So the only reason why atom:modified can ever be more useful is if you disagree with the publisher's opinion as to what is significant. If you disagree to this extent, what reason is there to believe that you will agree with the publisher as to what constitutes a change? Or that you will trust the publisher to actually insert a changed version in the feed upon a change that is not considered significant? Or that you will trust the publisher not to just lie?
And if you don't trust the publisher on these things, then why would you believe atom:modified anyhow?
I just don't believe that the difference between updated and published is operationally useful in the slightest.
In future argument, please don't argue any more that it's important to maintain successive versions or to track temporal differences. I believe all that. Instead, please explain why you expect to disagree with publishers' opinions as to what is significant and, in the case where you disagree, you feel you would trust atom:modified.
Absent such explanation, I am totally -1 on atom:modified. -Tim
