Originally Posted by pilotbob
If you just put them all in one field you have to scan through each record to pull out tags, etc. While indexed records can be retrieved so much faster.
Yes, of course. And the main system I administer that actually uses freetags (Drupal) does create a table, and CafePress does as well (I can tell, because sometimes my capitalization changes if someone has used the same tag before me with different capitalization). I suspect some of the older systems I've used really did just dump the tags into one field, but it would of course be much slower to search that way.
I remember trying to figure out how Otakuworld had done their "similar titles" search at one point and coming up with this convoluted solution involving bitwise comparisons (because they seemed to have a fixed number of tags). In retrospect, I'm sure they had something much simpler going on than that.