A colleague in Cuba ran traceroute from a computer on a dial up link in Havana to several destinations in Cuba, Venezuela and the US. The following table shows the round trip latency time for three tests to each destination:
Not only are the times slow, the variance to the same location is very high. Consider again the last line -- three tests resulted in three very different times. Even the variance within Cuba is high.
I cannot explain this -- some may be due to problems with my colleague's computer or Internet connection -- but hopefully we will see improvement soon.
If others run tests from Cuba and get different results, let us know (in confidence).