How do twitter mashup websites stay under the api limit?


I’m trying to figure out how all of these sentiment analysis sites are able to search all of twitter thousands of times a day without exceeding the api limit…

For example,

lets you query as many times as you want. How? What kind of black magic are they using?