Seeing some large numbers in Table B02018 (2012-2016 5-year estimates) at the tract level. I thought tracts had about 4-8,000 people, why are there some with 14,500+?

I went ahead and took a look at the highest Asian population tracts in the 2012-2016 ACS. The biggest one was 06059052420, with a population of 12,452, as you mention, in the Los Angeles area. I took a look at Google Earth to see if that might have any hints.

Here is a look at the area with the most recent imagery. The orange is the tract border (the boundary overlaying the Google Maps imagery comes from PolicyMap.com). The extreme northeast part of the tract which I've cropped out doesn't appear to be heavily populated.

Now, looking at Google Earth's imagery from 2010 (when the Census data comes from that would have drawn the boundary), that large development in the northeast wasn't built yet:

You can see plots being cleared, but no buildings yet.

I don't think this fully answers your question, but it's interesting to look at.

I went ahead and took a look at the highest Asian population tracts in the 2012-2016 ACS. The biggest one was 06059052420, with a population of 12,452, as you mention, in the Los Angeles area. I took a look at Google Earth to see if that might have any hints.

Here is a look at the area with the most recent imagery. The orange is the tract border (the boundary overlaying the Google Maps imagery comes from PolicyMap.com). The extreme northeast part of the tract which I've cropped out doesn't appear to be heavily populated.

Now, looking at Google Earth's imagery from 2010 (when the Census data comes from that would have drawn the boundary), that large development in the northeast wasn't built yet:

You can see plots being cleared, but no buildings yet.

I don't think this fully answers your question, but it's interesting to look at.

Wow, thank you so much Bernie! This is helpful.

It's still weird, though. The 2010 total population was 21,098-- way above the 8,000 target max (and it only grew to 22,727 in the latest ACS). It's possible the timing of the Census and aerial imagery accounts for that. In 2000, this tract had a population of just 7,438, and it's shape didn't change from 2000 to 2010.

All over the place, it's often frustrating how much tract shapes changed from 2000 to 2010, because it makes comparisons difficult. But usually it's necessary to keep tracts at that target size. So it's definitely odd that this tract didn't change shapes at all, despite the huge population growth from 2000 to 2010.

This could lead to a whole separate discussion on tract revision, commonly known as retracting. If a new development was built between 2000 and 2008 that added 14K population to that tract, the tract should have been split and the new development isolated into its own tract(s). The ideal tract size is 1500-2000 HUs, which translates to 4K-5K population.