Wonderful ideas – nice article. I have always felt that a web site needs to decide – is it serving data, user functionality or both ? Often times, the designers and owners of web sites confuse the two.

For example, a site serving stock market data will have tools that will show trends – but only the trends they think are important. If a user wants to check his own custom trend, he/she is out of luck.

I always recommend biterscripting ( http://www.biterscripting.com ) for mining data from web sites. That way one gets his raw data, and can process it in the way he sees fit.

Libraries are servers of data (raw information). In terms of whether (or how) they should develop APIs, it needs to be done rather carefully – eles they will end up being like these stock market web sites – where to get complete information on a particular stock in the format one desires – one has to go to several web sites to accomplish that.

Sen