Reprise : Reprise

Case number:845799-2009813
Opened by:Hanto
Type:DevPrev Issue
Opened on:Tuesday, June 9, 2020 - 10:03
Last modified:Tuesday, June 9, 2020 - 21:50

I, for one, don't have a cheating (0527). Why you might ask? Because I update ALL my working Clients and only leave a bunch in reserve that CANNOT play modern puzzles.

Now about (0527), if you read back through the feedbacks, you'll notice I had serious problems trying to increase my (0527) LOW POWER score once the (0603) came out. IN FACT I NEVER SUCCEEDED transitioning from LOW to MED or LOW to HIGH score. I strongly believe that if (0527) was a cheating devprev, then my end solution scores on that puzzle should have increased tremendously. In fact, they didn't move one thousandth of a pt.

So we're gonna re-invent the wheel again?

(Tue, 06/09/2020 - 10:03  |  1 comment)

Joined: 05/09/2008
Groups: None

Thank you for your feedbacks, Hanto.
The artifacts that you saw between the 0527 & 0603 clients are precisely why the 0527 client had to be permanently banished, blacklisted, and let-us-never-speak-of-it-again :-)

The client was allowing the protein chain to be folded in illegal ways. The structures thus produced are therefore completely invalid according to how we intended to model things with the puzzle. We cannot bring these solutions back, as there's not really an easy way to "fix" them to make them valid.

Again, we really apologize for this mistake and we really thank all of you devprev users for pinpointing this error so quickly... we can't imagine how devastating this would have been had this made it into the main client!


Developed by: UW Center for Game Science, UW Institute for Protein Design, Northeastern University, Vanderbilt University Meiler Lab, UC Davis
Supported by: DARPA, NSF, NIH, HHMI, Amazon, Microsoft, Adobe, RosettaCommons