Auto-complete is a good idea. But I find that it has slowed down searching. I can see how it increases the specificity of what is being searched (John can be recognized as the person versus the book). But often I find that I have to wait staring at the little "computing, computing, computing" bar. I can contrast it to a drop-down menu (I know it's not the same thing) for illustrative purposes: you click on the drop-down, you get your list. Click, get, get on with it. Not quite the experience with Auto-complete.
However, this varies. It reminds me a bit of how the home page and other features have behaved: first time you load it in a session is slow. After that, it is instant. This morning, my first attempt in Factbook was the staring at the bar type. Next attempts were instant.
This behavior is not limited to Factbook. It shows up wherever Auto-complete may be called on. To be honest, this was the reason I had a negative reaction to the feature at first and I suppose there are some who don't like it for that reason.
Before datasets like places, persons, cultural concepts, etc were in place, searching for an entry would have been dealing with text rather than data types. It made for more false-positives but it also left search outcomes in the hands of the users. In other words, I could look for places where the Gospel of John was discussed and had to manage my search terms in order to find them out of other irrelevant search hits. Now if Gospel of John comes up as a concept, it distinguishes from other similar search terms, but it also limits me to Logos' tagged entries. In my rather large commentary collection, are there really only 7 overviews of 1 Chronicles? I have my doubts.
Still, I recognize that for all its limitations, the concept has merit (it is handy to have quick access to the 7 overview sections of 1 Chronicles). If however, Auto-complete is bogged down by having to make "too many phone calls" as it were to various datasets, what may be gained in specificity is lost in performance. Both are important to users. Conceivably, as more datasets are added, Auto-complete would be more slowed down yet.
This comes back to indexing (yet again). I don't understand all that goes on under the hood, but this I know: indexing is frequent and painfully slow and it's purpose is to optimize searches. Perhaps that optimization does not apply to Auto-complete? Whatever be the case, I think that the feature does need optimization.
