I engaged in a small project about three weeks ago. The aim was to take star charts of the Second Life sky, in particular the stars, and draw constellations from them. A success would have been to find visible and repeatable constellations in the sky over the course of a few weeks.
As mentioned, a chart of the stars and constellations is worthless if it is only valid on one day and so I took to taking full snapshots of the midnight sky every night. The reason was, at the time, to establish the precise movement of the stars and extrapolate that data into charts that could be used to establish fun constellations such as "The Femur" or "Philip" or "The Gorean".
What I discovered was that each night the star positions changed completely. I took photographs over the course of a week and a half, and no two nights showed any correlation with any previous night. Attempts to discern movements of identified clusters of stars failed, as such defined clusters disappeared the next night and could not be identified.
The stars do not move in the sky. Taking photos at various times of the day yielded the same star positions, indicating that the astronomical sphere in Second Life does not rotate as the sun and moon do.
It was then suggested that the coordinates might be linked to a series lasting the course of a week. Again, photographs taken during the week of observation yielded no patterns. It was concluded that the stars' coordinates are not rotated on a weekly basis.
The mystery does not end there, however. On occasion, the stars will completely disappear from one's Second Life client, causing the night sky to appear completely dark with only the moon set in the sky. The first such event occurred at approximately 11 p.m. EST, and it was assumed to be the period at which Linden Lab changed the star map. However, a similar incident occurred two days later at 6 p.m. EST, leading to the conclusion that such occurrences are due to failures in the Second Life client or servers themselves than action on the part of Linden Lab.
The last test carried out was to determine if the star positions were similar across different viewers of different users. It was carried out by myself and Madison Rutledge. We stood in the same spot, focusing upon one direction (NW). The results were that to each of us a different star map was displayed.
In light of this evidence, the case can be made that it is implausible for one to create accurate and reproducible star maps of the Second Life night sky. This is due to failures of the client and viewer, randomness of star coordinates daily, and differentiation between different clients and IPs. The project to establish recognizable Second Life constellations was dropped.
Some further areas for study include logging in multiple times to determine if patterns change over each log-in, as well as photographing the night sky over a period of months if not years to determine if a series can be drawn on a monthly rather than a daily or weekly scale.
It is interesting to note that the Second Life sky itself is rather bland and sparse, consisting solely of white stars of variable intensities. No galaxies, nebula, novas, comets, meteoroids, or planets have been observed. Consideration should be given that no functional telescopes can be built in Second Life which severely limits the viewing possibilities, although the more likely theory is that Linden Lab simply was lazy and did not include such things for simplicity.
Finally, it should be noted that all of this probably could have been solved quickly and efficiently by asking a Linden, but that would have ruined all the fun.