Randomized Testing

Double Blind Procedure, Experimental Group, Control Group

Randomized testing means without any definite method.

Google also uses randomized testing to validate their Paxos implementation. Random testing in software will help avoid the problem of only testing what you know will work.

In some circumstances, random testing methods are more practical than any alternative, because information is lacking to make reasonable systematic test-point choices.

When doing randomized testing you must ensure that your tests are sufficiently random, and that they cover the specification.

Randomized Testing of Alternative Survey Formats Using Anonymous Volunteers on the World Wide Web. You can do randomized testing on your hunches.

You don't need to market online to test your hunches online. You can test online and deploy in any medium: magazines, TV, telemarketing, etc. Through randomized testing of your hunches, you build the confidence you need to spend the needed resources on your campaign.

Randomized testing allows you to focus on key aspects of your marketing message. You can create two or more variations of your message based on various hunches, and test which hunch delivers the best result.

Adaptive Random Testing
Chen, T.Y.
Fac. of Inf. & Commun. Technol., Swinburne Univ. of Technol., Hawthorn, VIC;
Abstract
Summary form only given. Random testing is a basic testing technique. Motivated by the observation that neighboring inputs normally exhibit similar failure behavior, the approach of adaptive random testing has recently been proposed to enhance the fault detection capability of random testing. The intuition of adaptive random testing is to evenly spread the randomly generated test cases. Experimental results have shown that adaptive random testing can use as fewer as 50% of test cases required by random testing with replacement to detect the first failure.

These results have very significant impact in software testing, because random testing is a basic and popular technique in software testing. In view of such a significant improvement of adaptive random testing over random testing, it is very natural to consider to replace random testing by adaptive random testing. Hence, many works involving random testing may be worthwhile to be reinvestigated using adaptive random testing instead. Obviously, there are different approaches of evenly spreading random test cases. In this tutorial, we are going to present several approaches, and discuss their advantages and disadvantages. Furthermore, the favorable and unfavorable conditions for adaptive random testing would also be discussed. Most existing research on adaptive random testing involves only numeric programs. The recent success of applying adaptive random testing for non-numeric programs would be discussed.