So you got 100ms ping out of a geostationary satellites? Having light travel nearly 80,000km which will take almost a third of a second? This is assuming you’re only transmitting 1000km away.
Direct PC didn’t have any way to uplink to satellites that can fit in a home. They relied on dial up modems to do so. It’s probably why it failed. But it was also late 90s so tech wasn’t so advanced.
You have to admit starlink is a real game changer.
Hinet have their own cables going to data centers in the USA. I am using them to connect to my US Clients in the data centers they use. Only 12 hops from my exchange to theirs and the first ten hops are Hinet pa.us.hinet.net
Europe is a nice place too. Other than pings, they also have good beer there!
Just came across this thread and wanted to add my experience in case someone is also wondering this:
If using AWS’s backbone, TPE->FRA has a relative stable ping of around 201-205ms.
Add another 3-5ms from a home HiNet connection or around 30ms from a cable internet connection.
And then, of course, the ping from AWS to the target server in Europe (could be another 5-25ms or higher).
In total, 210-230ms is an optimal range, I’d say. Although there are a couple of peaks throughout the day and around 0.1% packet loss (which can occur in bursts and might be noticeable in Zoom/Teams/etc.!).
Direct routing through HiNet will sometimes give similar values - but can also give peaks with up to 100ms more (if the data travels through the US!). So use AWS’s backbone, if possible (running a small VPN should be around USD 15-25 per month depending on how much data you put through).
Anything sub-200ms will probably get really expensive…
* Plus that whole speed-of-light and laws-of-physics thing… 
However, depending on the use case (not gaming) if the presentation layer is less sensitive and you can abstract it from the operation, then you can remote into a system that’s closer. I.e. the latency-sensitive app runs somewhere close to the host so it can respond quickly and handle more data volume, so it’s just the screen output that’s slow coming to your eyeballs.
* Understanding that the primary component of latency is the switching, (and inefficient routing due to unbalanced peering agreements), this sort of thing will be much better in a few years when the hardware for quantum communications gets out of the lab and enables direct point-to-point communications without physical media in between.
This is my main concern with the possibility of moving to Taiwan…I’ll lose touch with my US league of legends friends if there’s no viable way to play NA server 
Pretty sure there’s folks in Taiwan that play it too. 
Haha you’re not wrong. Guess I just gotta make some new friends.