Bumping this one up so as to alert GT to what looks to us like a significant interface oversight (and hopefully get a workaround).
rayhne is correct. If a link appears in more than one category, it can only be modified by editors who have permissions in ALL those categories!
Here is our story ... Having completed the first stage of our project (compiling an initial set of links, and putting them into appropriate categories with provisional descriptions), we persuaded our reluctant (they unanimously dislike the browser.cgi interface) editors to give browser.cgi another chance.
Result: no editors. All our links are in multiple categories, so effectively the editors can do nothing. We don't have a single link that can be modified by editors through browser.cgi!. None of the links are editable, and none of the editors can do anything. Our only way out would be to give all editors full permissions in all categories, which they don't want and neither do we.
I'm sure that this is OK for many LSQL clients, eg: directories in which multiple categorization is rare, or where it is understood that links may not have equal significance in all the categories where they are placed. But the uses of LSQL are diverse ... GT makes a virtue of LSQL's flexibility in its marketing, and really should not have allowed errors of this magnitude to remain unnoticed. This is a failure of the Testing Dept.
GT: We would be happy with an assurance that browser.cgi will get some serious attention in the eagerly awaited next release. Meanwhile -- is there anything we can do now?