I had turned my attention to google for the last quarter of 2002 and have just come back to Ink having had a break - so I am a bit behind.
Here is what I have found over my last 4 refreshes on 3 different domains.
Adding a keywords tag did not change the ranking.
Adding a description tag did not change the ranking.
Adding a text box named aptly did not change the ranking.
Adding a paragraph of text with a heading - with keywords etc did not change the ranking.
Adding an image named by the keyword did not change the ranking.
Adding Alt tags made the page drop two positions.
Removing them again made the same page drop another two positions.
Adding keyword at top of page in large text did not change the ranking.
Adding a second sentence to the description and placing this on the page (of course with keywords in it) did not change the ranking.
These changes were meticulous and logged across three domains and two separate submission accounts under different names.
The only way I have changed the ranking of a site is by placing a completely different page on there with a different target phrase. I.E. by removing all references to the original keyword - it no longer ranks for that word and gets shoved in for the other word "somewhere".
In addition to my own testing, a friend of mine with a third account has been modifying pages in far more dramatic ways - I.E. retaining the target keywords, but completely re-writing the pages and letting them refresh. He hasnt moved much either - 3 spots up or down which is probably due to new additions/deletions anyway.
I therefore surmise that Inktomi uses some major off page ranking factor in addition to the on the page parameters which appear to get you stick in or around a given position - until that is - there is the periodic major algo change which incorporates further random factors. When this comes along its always back to the drawing board.
Anyone record any similar effects? And what am I missing........hmmmmmmmmmmm