Redundant Results

From Unofficial BOINC Wiki

[edit] General

Redundant Results are used in scientific research to verify that the process of obtaining data and the data itself is not flawed. Experiments are often repeated several times to confirm that the outcome is valid.

BOINC Powered Projects approach this in different ways.

• The most common way is identical Work Units are sent out multiple times. The Results are then compared and Validated. The exact number of times each Work Unit is sent out and how many successful Results must be returned varies between Projects.
• Climateprediction.net handles redundancy in a different manner than most Projects. Since their Work Units are months long, running identical Work Units is impractical. Therefore they run similar Work Units and check that the Results fit a pattern.
• Folding@Home uses internal checks in the Science Application, random Redundant Results, and pattern fitting to ensure that their Results are accurate.