Regarding the API traffic, it looks like a variation on JSON-P. Basically, you do this:

—————————

/* write a callback function */
function myJSONCallback(aJSON) { alert(aJSON) }

/* append a script element to the document */
var myScript = document.createElement(“script”);
myScript.src = “http://api.example.com/makeArray?elements=3&callback=myJSONCallback”;

—————————

The browser sends a GET to request the script src, and this can be any host. The body of the request looks like this:

myJSONCallback(["","",""]);

The contents of the parens are a JSON object literal.

http://www.franklinmint.fm/blog/archives/000951.html

Also, the server-side processing argument seems to miss the point. The canonical “mashup” is a web page that pulls in JS data from Google Maps and some other source. If you want people to use Google Search results on their web pages, the wrong answer is “here is the SOAP/XML-REST/HTTP/Whatever API”. That is for the (vocal) priesthood, not web authors. Your main point seems to be that it doesn’t allow bots. *shrug

it’s possible to programatically access the AJAX Search API. Please see here for proof (calling from plain ol’ Java):
http://blogs.cocoondev.org/dims/archives/004722.html

thanks,
dims

Ever signed an employment agreement that makes anything you invent while in their employ the IP of the employer? No? Then you never worked for a professional company of any significance.

It is not surprising they took this long to get around to this. Dope dealers get you hooked first. It is only surprising that you are so angst-ridden about that.

“As the twig is bent…, “, David. The web changes nothing in the human equation except short cycle behaviors. The long cycle behaviors reemerge dominant as always. Over a longer time, you may see some subtle variations change features in certain environment but these will come as side effects, not direct effects. Beware pleioptropy.

len