It is a common observation in many real-life events (such as examination results, corporate performance grades etc) that data tends to cluster around a mean (average). In probability theory and statistics, the normal distribution or Gaussian distribution is a continuous probability distribution that describes data that clusters around a mean or average, and this is how the graph looks.
Inspired by the bell curve model, Team GDC analyzed the voting patterns of participants. We filtered out the spikes and anomalies (at the fringes) caused by false/maliciously engineered votes or lazy applicants who dropped out mid-way. Thus we were able to determine the number of votes (a band) that would determine ‘average’ performance’. If we marked this band as the ‘mean’, we could now move up/below the mean, assigning voting bands (a range of votes) where the performance could be considered below (or above) average, and following the same rational identify those who did poorly or excellent (compared to the median score of votes).
This is a kind of relative grading principle, which helps us neglect the ‘errors’ that otherwise creep in due to a few extra votes (duplicates from known friends) or a few misses (emails not confirmed). We are not in a position to share the bands, cutoff scores etc, and needless to mention the decision of the Jury is going to be final.