Each file downloads at 1000kb/s. This appears to be a hard, arbitrary limit. If two files are downloaded concurrently, the max speed between the two increases to 2000kb/s total, indicating bandwidth is going to waste due to arbitrary limits. I'm no IT expert, so I have no idea how complicated it would be to make ensure bandwidth is distributed fairly and efficiently using the number of concurrent files are being downloaded from the server as the condition the software takes into account when distributing bandwidth.
This has been a problem since the launch of this service (see the first 2 pages of this thread where we discussed this). I get only around 200 to 300 kB/s per download thread (on my 100mbit downstream link) which I'm able to scale up linearly by using 10 or even more parallel threads (axel is my friend!). But I don't think these are mjcmurfy's throttling rules either. This probably comes from some ISP/router along the way when routing packets to certain hosts. mjcmurfy says that he's able to get pretty high, unthrottled speeds in his own testing:
As for your speed, 400kb is indeed VERY slow. I don't understand how it could have been that bad. My website runs on a 100mbit dedicated server with unmetered bandwidth, based in europe. I max out my 30mbit home broadband connection downloading from this server every time, without fail. It could be due to the fact that the server is located so far away in europe, or maybe you just got it on a particularly congested time. It's hard to say. But I know that I am downloading something off it right now at a speed of 2.8 mb/s.
It's a pity that this is a problem for a website dealing in downloads. It would be fantastic if this could be investigated (I offer to be a testing guinea pig if needed!)