Previously, dr. Fritte Cornelius and I made observations that confirmed that the wormholes of Sanctified Vidette spawn according to a uniform distribution drawn from all Jove observatory systems. The research was written up in a research report.
But that observation is perhaps no longer true after the changes, since capsuleers have to trigger the wormhole appearance. Hence the uniform distribution can be distorted by capsuleers' behavior, which may differ from region to region. I set out to do the analysis again - with a more limited amount of data.
Nevertheless, over my recent observations in Sanctified Vidette, I have catalogued 275 wormholes (mostly before 22:00 New Eden standard time, and during the second half April). The observed numbers of highsec, lowsec and nullsec holes, and the expected values based on the uniform distribution are listed below:
Nullsec: observed = 189, expected = 175
Lowsec: observed = 46, expected = 45
Highsec: observed = 40, expected = 55
The number of nullsec holes is somewhat high, but still quite within what you would expect to see by chance. The number of lowsec holes is perfectly ordinary and in line with expectations. But the number of highsec holes is indeed significantly lower! The overall observed distribution does seem to deviate from what existed before [1].
From this, I would tentatively conclude that capsuleers in nullsec regions are a little bit more likely than average to breach wormholes to Sanctified Vidette, perhaps to look for bridges to the Empire worlds. The denizens of highsec, however, are quite more risk-averse than average and tend to jump less into unidentified wormholes [2].
Perhaps this is not all to surprising, but it is still nice to see it reflected in the data. And with more data, this can be refined to statistics for smaller regions, and allow to detect movement of troops or material by groups using the Drifter wormholes. The Imperial Navy will be pleased to hear this.
Today's footnotes are for the statistics afficionados:
[1] Statistical note: using a multinomial model (with probabilities 0.63618, 0.16261 and 0.20058 for null, low, and highsec) the p-value for an outcome this far or farther from the model is still p=0.10 (chi-square of 4.6, with 2 degrees of freedom). While this indicates that the outcome is unusual, the deviation may not yet be confidently called significant.
[2] Also this can be quantified, using Bayesian inference. Let's say I am looking for the conditional probability P(breach|HS) that a Drifter wormhole is breached by capsuleers, given that it is in highsec. The number I observe in Vidette gives me the probability P(HS|breach) that a Drifter wormhole originated in highsec, given that it was already breached (otherwise it would not have appeared in Vidette). Bayesian inference then tells me:
P(breach|HS) = P(HS|breach) . P(breach)/P(HS)
Here, P(HS)=0.20, the probability of that a Jove observatory system is in highsec, and P(breach) is the probability that any Drifter hole is breached, estimated as the average number of wormholes in Vidette divided by the maximum - here I take it 40/60 based on observation (obviously the result just scales with this value). Following the Bayesian inference formula above, the outcome for highsec, lowsec and nullsec are:
P(breach|HS) = 0.48
P(breach|LS) = 0.69
P(breach|NS) = 0.72
These are the estimated probabilities that a capsuleer has breached a Drifter wormhole (by 22:00 NEST), per region.
