According to: https://codeforces.net/blog/entry/68204

MikeMirzayanov announced a way to uphack recent problems for Div.1 participants. I'm curious to listen to the details on why this is limited only to last week problems. He clarified why it was limited to Div.1 participants but not why only recent problems.

I mean, for instance I believe that Educational Round 3B is hackable with this test case (as many participants don't use long long for the length of the frog's tongue):

1 200000
1 0
1 1000000000
1 1000000000
1 1000000000
...

So what is the harm in enabling uphacking for old problems as well?