Google CSE Interview and Complaints

There's an interview today with the product leader for the Google Custom Search Engine (CSE). I always find numbers interesting:

We have more than 100K registered Custom Search Engines, and that's growing pretty rapidly.

(Although it will be far more interesting to see how many of these are getting any traffic several months from now.) Since I just put up my Customize Your Own Search Engine page and built a State Libraries: Custom Search page, I might as well mention a few of my complaints with Google's CSE.

I'm hoping that I'm just missing some variables on some of these, or that these capabilities will be added soon. Overall, I found the Google CSE easy to set up and use. I just wish it would fix a few issues that basically would make the results display more like a regular Google search. These criticisms mostly apply to CSEs that are not just limited to a single host.

• Supplemental results: These should be included in all CSEs, and not just those with fewer than 3 sites or with no site restrictions. See Google's explanation, and note that "We're working hard to bring supplemental results to more Custom Search Engines." I hope that does happen. Why does this matter? Because a CSE, like mine, that limits to more than 3 sites makes the most sense when it digs deeply into those sites. Some pages that Google has indexed may only be in the supplemental index, and if supplemental results are not included, searchers will not see those pages listed in the results. What can happen is that a regular Google search could find pages that the CSE will not.
• Cluster results: When you get a second hit from the same site at Google, the second one is indented under the first. And sometimes, if more results are available from the same site, there is the [ More results from . . .] link as well. The image to the left shows both of these on a regular Google search. Unfortunately, the CSE results neither indent nor add the [ More results from . . .] link. The addition of both would be helpful. If you are searching a group of several dozen sites, this would provide at least some easy grouping by site. Instead, searchers just have to check the URLs for each result and determine for themselves which are from the same site.
• Fix counts: OK, Google's estimated total number of results are way off anyway, but on the CSE, it just says "Results 1 - 10 for . . . " at the upper right without any estimate, however inaccurate, of the total number of results.
• Cache: Wouldn't it be nice to have links to Google's cache copy and for documents, the "View as HTML" versions? The odd thing with the CSE is that sometimes one or the other of these two links are there. For my state libraries search, at first, neither were there. Today, the "View as HTML" links are there. But I've not seen any cache links. Yet on other CSE, I've seen both, one, or none at all. I hope this is just a glitch and that both will appear on most CSE, or at least that Google gives the option to the creator as to whether those links appear or not.