Beating the native score
| Case number: | 845799-986509 |
| Topic: | Biochem |
| Opened by: | Madde |
| Status: | Resolved |
| Assigned: | admin |
| Priority: | 3 |
| Type: | Question |
| Opened on: | Thursday, July 16, 2009 - 17:04 |
| Last modified: | Saturday, August 1, 2009 - 01:15 |
We've had eight "Quest to the Native" puzzles so far.
In four of these we (the Foldit players) have beaten the native score:
| Puzzle | Native Score | Highest Score | Link to Puzzle Page |
| 149: QttN 2 | 10730.100 | 10829 | http://fold.it/portal/node/986163 |
| 152: QttN 3 | 9181.154 | 9353 | http://fold.it/portal/node/986215 |
| 164: QttN 7 | 9825.048 | 9855 | http://fold.it/portal/node/986445 |
| 166: QttN 8 | 10571.069 | 10602 | http://fold.it/portal/node/986458 |
Could someone from the staff explain what that means, please?
- Is the scoring function of Foldit inaccurate?
- Didn't you find the real native shapes of the proteins?
- Isn't the native shape the one with the lowest energy level?
If the computer could calculate it perfectly what is the point of foldit?
| Status: Open » Resolved |
| Assigned: Anonymous » admin |
If the computer could calculate the score perfectly, but not the fold, that would still be useful. It's a bit like factoring: we can really easily tell if a factoring is correct, but finding one is still very difficult.
The score function is not perfect, nor are the tools, partly because the score function is a matter of current research, and partly because it was originally designed to be used by algorithms, not humans. But on none of the puzzles did players get significantly below the native score, and those that did were close to the native. It would be a matter for worry if players could get a significantly better score than the native while relatively far from it, but so far we have not seen such behaviour.
![](http://128.95.1.228:8000/tracking_pixel.gif?events=scAdd&f=1111&wdata={"x":12, "y":11}){kind=tracking}
Madde:
Our scoring function is just not that perfect. We do not actually solve the puzzles, an algorithm calculates the optimal state and produces the score. As this algorithm is not perfect our native scores not perfect. We're working on optimizing this algorithm.