Ppcc

DANG!
 
If that's what you get in "an area that's horribly overprovisioned." -- sign me up!
 
Sorry I'm slow to respond dude, been consumed with building a couple of ZFS servers on the backend. It's a process, that's just about complete.

Where are you seeing this?
 
I do need to redesign the ranking to be more user friendly.  Definitely can be done better.
 
Here's the ranking, after filtering out some of the less popular locales.  I manually do a minimum test count of 45100 because every day the host database tables that the host averages are calculated from are trimmed, to 45000.  This isn't the right way to do this because it's dependent on the time of day. 
 
I have some additional logic I'm adding to that.  Where that part of the program will be made aware of the actual recent popularity.  It will default to filtering out the less popular locations to build a better list.
 
Sorting by download speed
https://testmy.net/rank/countrycode.down/45100
 
 
Sorting by upload speed
https://testmy.net/rank/countrycode.up/45100

 
It will look different using the defaults.  It also may have been different when you looked, because of the time of day.  The tables might have just trimmed and then it happened that a bunch of tests were taking place in the locations you mentioned above.  That tool was designed before I started trimming those tables.  It was easier to sort them by popularity before they started being trimmed... after time the majority of the tables are over 45K rows making that variable unreliable.
 
Going way beyond the topic...
 
Keep in mind, I built TestMy.net with really no prior programming knowledge... starting in 1996 before it was even TestMy.net.  I then dropped out of high school in 1999, the useless information and repetition... I learn way more, much faster by DOING so I decided to focus on that.   Building this taught me and I'm always learning.  I've definitely made a lot of programming mistakes but they've all lead to a better understanding and better future designs. 
 
 
I'm a lone wolf programmer.  All of the frontend, all of the backend.  I even built the physical servers and infrastructure by hand.
 
BTW, I was like 1 or 2 credits from graduating.  Didn't really want their paper.  What does it prove?  Got hired at one of the largest telecommunication giants, quit school the next day.  Worked for them for 5 years.  Working my way up from the warehouse, within 1 year my pay increased over 6X.  That company was good to me but TestMy.net was calling, I've pursued it full time ever since.
 
Side note, if anyone actually reads this: That company had a strict policy and would have fired me if they knew I didn't have at least a GED.  No matter what my performance was in the company.  People high up (like extremely high up) told me it would be an open shut case and to keep quiet about it, "Just go get your GED dude!".  Some told me it wouldn't matter even if I got it because I should have never been hired in the first place. That attitude was actually the reason I left them.  With my own company nobody could hang anything over my head.  Yes, I probably could have just got a GED.  But why?  I don't want that.  They assumed I graduated because I had college credits, never lied to them.  Thankfully it never became an issue and I resigned on top.  Great company, they taught me a lot too.

According to hosts who've contacted me, there's an option for them to "adjust the drop off".  So they can cut off the top and bottom portions and shape the result to suit the narrative they want to paint.  I know this because they were asking me if I had those options!  Not just one or two by the way.  And they've stated that these options only exist on the higher level licenses.  So basically, if you pay them enough.  Exactly what you eluded to.
 
More reason why an unbiased third party opinion from TestMy.net is necessary.  All connections are tested under the same internal variables.  Any variable that can be controlled is controlled by YOU, the client... not your ISP.

they won't . not in SEA asia sure. ALL of their ISPs seem to use Speedtest dot net as a standard. and they're obviously bribing  Speedtest to get the results they want.
but AFAIK speedtests to USA servers are still kind of accurate... as in they're definitely much lower than the expected  100 mbps for me. i'm suppose to get 500 mbps but i am limited by hardware and USA bandwidth that is very obviously throttled.