True Random Number Service

Statistics for 2009-07-04

This page shows statistics about the performance of RANDOM.ORG on 2009-07-04. The statistics measure various aspects of the generator's operation, including how heavily it's loaded and what the quality of the numbers is.

View statistics for

For general considerations about the testing of the random numbers, please see the Statistical Analysis page. You can also browse the graphs via the Statistics Browser.

Buffer Level

The buffer level graphs give an indication of how busy the generator is. The generator contains one buffer for each radio, which is constantly filled with randomness generated from that radio. If the buffer level for a given radio is low, it means that radio is busy. [more]

Traffic Level

The traffic level graphs give an indication of how busy the web server is. The amount of requests are shown per hour over a full day. The more requests it has to deal with, the busier the server is. Currently, only web traffic is shown. [more]

Source Purity

Any source of true randomness may contain skew towards 0 or 1 in the data. The skew is removed by a skew correction algorithm. The source purity graphs show how close to unskewed the random data generated is before skew correction is performed. [more]

Information Entropy

The information entropy graphs show the entropy level in the data as a percentage of the theoretical maximum. Information entropy gives an indication of how difficult it is to predict what the next bit will be based on the bits you have already seen. [more]

Frequency (Monobit) Test

This test is part of the NIST recommendations. The graphs show whether the number of 0s and 1s produced by the generator is as close to 50-50 as you would expect for a truly random sequence. [more]

Frequency (Block) Test

This test is part of the NIST recommendations. Like the Frequency (Monobit) Test, the graphs for this test show whether the number of 0s and 1s produced by the generator is as close to 50-50 as you would expect for a truly random sequence. [more]

Runs Test

This test is part of the NIST recommendations. The graphs show whether the number uninterrupted series of ones and zeroes of various lengths are as would be expected for a truly random sequence. [more]

Please note that the generator is not expected to pass all the tests all the time. In fact, if this was the case, the generator probably wouldn't be random. This may sound non-intuitive, so it's explained in detail on the Statistical Analysis page.

The tests were chosen on the basis of several analyses conducted by students here at the university (also available from the Statistical Analysis page) and are also the ones recommended by the US National Institute of Standards and Technology (NIST) in NIST Special Publication 800-22, which deals with evaluation of random number generators.

Note that while the RANDOM.ORG numbers passed a spreadsheet implementation of the entire NIST suite, I am still in the process of implementing the NIST tests such that they are run on all numbers generated. While the Statistical Analysis page gives all the tests, only the ones you see above have been implemented as real-time statistics so far.