I'm familiar enough with the history of computer science to know that it uses three basic logical operations: and, or, and not. (discovered in fact long before computers even existed in the mid 19th century by an English mathematician and philosopher named George Boole). This is called Boolean logic. And operations basically say that if something AND something else are detected, a yes result is returned. OR operations say that if either something OR some other something is detected, a yes result is returned. And a Not operation says that if something is detected, its opposite will be returned. With your ordinary average keyword search some goof who knows nothing of computer science performs, it appears that its using an AND type logic. IE if you search on the terms "revolutionary" and "war" it says: return a yes result for documents that contain "revolutionary" AND "war." When you put the term in quotes, i.e. an optimized search, it seems to do a kinda "strict" and search. i.e. it finds the titles of articles that have those terms in sequence right next to eachother. It does not however care if there are other words in the title. Unless, it seems, those words are very unrelated. Then it will reject the document because it has words which don't normally go with that category. In a non optimized search it doesn't seem to necesarily need them all next to each other, that seems to be about the only difference. I guess also it doesn't do any "OR" stuff later on or anything too unrelated. instance: I searched "revolutionary war mouth water" and obscure narratives came up. These were not the same sites that, obviously, came up with just "revolutionary war." However there were many sites that had a lot of other RELATED extra words when I searched "revolutionary war," such as "American" and "independance." i.e. these words were even in the title in some of the first results. So it didn't care if there were closely related words. If it didn't somehow differentiate this, there would be no way for it to rank "revolutionary war Independance American" as any higher than "Revolutionary War mouth water." So it seems the search engine must have a whole catelogged set of words that it will accept as "extras" and then anything further afield it will reject as unrelated or put way further down the list. another instance: I added the unrelated word "dime" to "revolutionary mouth water" and the revolutioonary mouth water page Id gotten before didn't come up. I took the word dime and water off and the revolutionary war mouth water page also didn't come up. So clearly the "further afeild" extra terms are, the less they are considered as results, even if the AND conditions have already been satisfied. So the servers must have, in a sense, a kind of "subconscious" set of associated words to go by i.e to judge what is "far off" and what's "close and asscociated." Also just to note: it does use an "or" logic at times. If it finds really unrelated words it seems to resort to an "or" logic to give you a mix of pages that contain 2 of 3 words: this 2, then that 2, or something along those lines. So all this says to me that for best results, use the closest MATCH to the keywords you've used, both as regards the actual terms and their context. If you put MORE words in, say, your title than the search words you researched, then make sure they're as closely related as you can possibly make them. I spose the rest is all page ranking stuff i.e. number of backlinks, quality score, etc. so I'd say the hierarchy for a non-optimized search is like: Find the exact words along with related words if rank is high. After the "top sites" have been given priority, find the exact words in highest frequency with fewest related words, and of course no unrelated words. Next: allow for more related words along with the exact words Next: Decreasing order of relatedness of related words Next: go into some "OR" logic when you've run out of all the "AND" documents. Loose and rough I know, but I bet the algorithm goes something like that. any comments? Will[/B]