Not exactly correct Fenris.
The issue was we had a strict timeout but if your net connection was good all the time and our server was happy as well it can download well before the timeout phase is reached.
It was when the download time takes longer than the timeout that the issue appears. Hence why a lot of customers haven't noticed it being on very fast connections.
The speed of the connection is not the governing factor, it's the latency that can affect you. That latency can be due to any link in the chain, so it could be down to contention local to you, between two of the nodes in the link, or even at the server if it's handling many connections.
But I'm sure you know all that, the real point of this is that the fix is simply to not timeout the network traffic if a single refresh (or a small number) is missed, which is likely to be the fix that has been applied.