Speech therapy tales

Some times I help my wife with MS Office controls and read her papers. Some are interesting.

All textbooks agree that it is harmful for speech development if parent speaks with a small child (0-3yo) using only a language parent is not proficient in. When you think about it it is pretty apparent, but I've seen some Russian immigrants who
1. Have very poor command of English/German.
2. Speak with a child only in this mediocre English/German, not in Russian.

There is a misconception here in German society that an immigrant should speak German with her children for better "integration". The curious thing is not that some of the parents make this mistake succumbing to social pressure. It is curious that most of the future speech therapists here also sincerely believe more in this common wisdom rather than agree with what their textbooks are saying.

Spooks in top500, conspiracy theory?

While browsing latest top500 (actually top 100) I found two suspicious entries. Other 98 entries can be attributed to some real workloads, some times sensitive like code cracking and nuclear blasts modeling. But those two do not have any viable raison d'être, so I suspect they are the big brother that filters traffic. (Echelon and such)

This one in US
http://www.top500.org/system/10623
I mean in US there are of course several supercomputers marked as "NSA" but this one is just "government" without any details.

And second suspect is in Sweden:
http://www.top500.org/system/10859
It is also marked as "government" without any further details. Why Sweden??? They don't have the bomb and it is unlikely they are building one. May be they are examining a big portion of Russia's internet traffic which happen to pass by.

My second guess is that if my suspicion that those two systems do real time traffic monitoring is true, it is very unlikely that both were fully offlined to run a LINPAK. What they did is probably they chosen a time of the day when there is least amount of traffic and used the idle part of the cluster to run tests. This means that the real capacity is 3x-5x of reported to top500, which would shift them from the bottom of top500 to top20 :)

It is another question why submitting the LINPACK scores for secret systems in the first place? May be everyone who's day job is to know what they are know any way so why not?