Many global tracks (approximately half) are given a negative flag. As I had not really understood this before, I made some brief investigations...

1) For what reason are they being given a negative flag? They almost all have a flag value of -102 (for TPC-only tracks, at least). This means that they have too few hits. (Tracks with abs(flag) values greater than ~100 are those with silicon hits in 2007, and FTPC in 2007-2008.)

2) Is this a new phenomenon? I looked at QA histograms from the past 4 years and found the answer to be that it is nothing new, although it appears to be a bit worse in 2010:

3) Is this some abnormal group of short tracks? One idea I had for looking into this was to see if there's a smooth distribution for the number of points on the tracks. In this plot (using file st_physics_11035026_raw_2010001), I've taken the QA histogram of that distribution for tracks with positive flag (black, note the special short tracks with 5<=nhits<=10 which point to the EEMC), and negative flag (red, altered the QA code to plot only these), and the sum of those two (blue). The distribution appears smoothly rising into this region of very short tracks.

I also tried making various plots of where in the TPC these negative flag tracks are located and found nothing particularly of note (they're everywhere, although there is a depletion of them in the region near the EEMC as those tracks are tagged as being the special short EEMC-pointing tracks).

CONCLUSION: although there seem to be a lot of these tracks (more than would make me comfortable), their presence seems "normal".

-Gene