Finding the Fun in Folding

Started by axcho

jeff101 Lv 1

I often ask questions like in enszyme's 1/20/2012 post above.

The Tutorial Puzzles were nice in that they went fast and I could see definite endpoints.
They also gave fireworks and rewards on completion of each puzzle.

The other puzzles have been incredibly time-consuming,
and it's hard to feel like I'm ever finished with one.
Sometimes I feel relieved when a puzzle ends.

I do get excited when my score is steadily rising.
I also like trying to think of better ways to fold things,
and seeing that others are getting higher scores shows me
that it can definitely be done somehow.

I have enjoyed writing scripts and trying to figure out how to implement new methods.

It is also nice to think that playing Foldit can be contributing to real advancement in science.
Many other similarly-addictive online massive multiplayer games cannot say that.
I also like how Foldit motivates people to learn more about science and computer programming.

It would be nice to hear more from the scientists about
how even the lower-scoring solutions are important.
If your results don't matter and you can't be one of the best, why bother?
Is it really that much fun to manipulate a protein and watch it move around on the screen?
How long can that be entertaining?

I think there need to be more smaller, attainable goals that folks can strive for.
Striving to be the #1 soloist or evolver in the world isn't very motivating to me.
What other incentives can be built into the game?

infjamc Lv 1

"It would be nice to hear more from the scientists about how even the lower-scoring solutions are important."

  1. Regarding this issue, the best example would probably be the monkey virus case from last September: the Foldit solution that helped the researchers solve for the structure of the protein came from the third-placed group. How can this happen? Because the score function isn't perfect. This is because the best one can do is to model the chemical interactions and determine the relative weights of the numerous terms by fitting them to real-world data. The result is a one-size-fits-all model, which may be more accurate for some proteins than for others.

  2. Another issue of note is that sometimes a "good" structure could score badly because it hasn't gone through enough "fine tuning." For example, suppose that you have a protein with 100 amino acids, and your structure has 98 of them scoring an average of 20 points each while 2 of them score -500 points each. Since you get 8000 points for free, the result is a grand total of 2098-1000+8000=8960 points. Suppose that another player has a structure that scores an average of 18 points each, which gives a total of 18100+8000=9800 points. In this situation, your structure is not necessarily worse as a whole despite the lower score, because all you have to fix are the two segments that score very badly. If you can even get them to zero points, you would score 9960; even if you can't, it might still be possible for the researchers to take your structure and further refine it on their own.

  3. The theory behind the idea of getting as high as score as you can is that the actual structure in nature is often one of the most energetically stable ones. However, this assumption is not always true, because there's also a possibility that a structure might be too stable to achieve the needed functions. (This can be an issue in design puzzles.) In such a scenario, a solution that scores "high enough" but isn't quite the unique highest-scoring structure might actually be preferable.

jeff101 Lv 1

Thanks, infjamc. It is good to know that lower-scoring structures do matter. I think proteins are not rigid rocks either. They must sample a range of conformations at finite temperatures, and some of these motions are probably vital to a protein's function. Maybe some of the lower-scoring (higher-energy?) conformations can be important as intermediates during protein motions.