Reward matching guide in QTTN, more like ED

Case number:671071-994413
Topic:Game: Other
Opened by:Susume
Status:Open
Type:Suggestion
Opened on:Friday, January 25, 2013 - 18:47
Last modified:Saturday, January 26, 2013 - 07:55

A newer player on our team has pointed out that it's discouraging to new players to learn that the best scores in QTTNs are for proteins that diverge from the guide. The QTTN puzzles would be more rewarding as a teaching tool if they actually gave points for matching the guide. This would also make them good training for ED puzzles. Since ED puzzles are an area where foldit players have a real chance of beating computers, it would be good to improve people's ED skills.

(Fri, 01/25/2013 - 18:47  |  6 comments)


brow42's picture
User offline. Last seen 42 weeks 2 days ago. Offline
Joined: 09/19/2011
Groups: None

I play a lot of QTTN and feel the same way, but I've thought about it a lot and there are reasons I haven't suggested it.

The biggest reason is that if you add points for matching, then wiggle and DRW-like scripts will pull the protein onto the guide like magnets. I think this is okay for ED since visualizing the ED is so difficult, and the ED problems are generally unsolved.

The proteins have flexible parts and are moving about. The so-called 'native' is probably just 1 conformation or some average or consensus conformation. They're all essentially equivalent to within the thermal energy of the protein, but Rosetta prefers one over the others.

The scientists are probably using the differences between our optimal shape and the published shape to identify tweaks to Rosetta.

gitwut's picture
User offline. Last seen 2 years 8 weeks ago. Offline
Joined: 05/18/2012
Groups: Contenders

I don't see a problem with the QTTNs, unless I'm missing something (if so, I want to know). I don't think there is any need for a bonus. Nothing is going to be as simple or straight forward as the tutorials.

The process of aligning to the guide and relaxing the protein is performed over and over again. Each of the align/relax iterations should produce a gain in score. When the score stops increasing or you find that align/relax fails to bring your relaxed protein any closer to the guide, you diverge from the guide because there is no other way (short of skilled hand folding) to improve your score.

marie_s's picture
User offline. Last seen 46 weeks 4 days ago. Offline
Joined: 05/18/2008
Groups: None

I dont think it is a good idea.

From the start of the game,I find QTTN boring and I dont see how somebody can learn to play in the task of putting trillions of bands between the protein and the guide or just watching recipes working.

If players dont learn the real score, it makes it worse.

Learning to fold is learning by experience what works and what doesnt work, alteration of the score will complicate the learning process.

Players have to learn that score is not perfect,it explains why the best score are not always the one who guess the right shape. This is a hard lesson some learn with the monkey virus, some in casp9 and some in casp10.

ED is only one type of puzzle, it is not proven at this stage that ED puzzles are the only future of Foldit and that the actual ED score is the best tool to that.

marie_s's picture
User offline. Last seen 46 weeks 4 days ago. Offline
Joined: 05/18/2008
Groups: None

I dont think it is a good idea.

From the start of the game,I find QTTN boring and I dont see how somebody can learn to play in the task of putting trillions of bands between the protein and the guide or just watching recipes working.

If players dont learn the real score, it makes it worse.

Learning to fold is learning by experience what works and what doesnt work, alteration of the score will complicate the learning process.

Players have to learn that score is not perfect,it explains why the best score are not always the one who guess the right shape. This is a hard lesson some learn with the monkey virus, some in casp9 and some in casp10.

ED is only one type of puzzle, it is not proven at this stage that ED puzzles are the only future of Foldit and that the actual ED score is the best tool to that.

Joined: 08/24/2011

"I dont see how somebody can learn to play in the task of putting trillions of bands between the protein and the guide or just watching recipes working."

Neither do I, because that's not how QTTNs are meant to be played. I still learn a lot by playing those. I use 10 bands at most on guide at the beginning of the puzzle, just to get the right shape and that's it. But that's where I'm learning the best way to rebuild manually. I got 9900 on the potato puzzle with no hard-working recipe (I mean I could have done everything either manually or using GUI scripts, but I'm a bit lazy), that's a pretty decent score. My point is that QTTNs are fun in spite of their easiness. But maybe setting those to "recipes locked" mode could be a good idea and a better school for newcomers.

Joined: 08/24/2011

Haha, even funnier. Don't know if it's possible but I think it's cool: merge all three levels (normal, <150, <15) into one and lock recipes for veterans.

Sitemap

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