McGill-CSB / PHYLO

a gaming framework to align genomic data
phylo.cs.mcgill.ca/edge
Other
11 stars 14 forks source link

scoring/highscore #83

Open phaethonas opened 11 years ago

phaethonas commented 11 years ago

Hi, I was wondering if there is a way to know the highscore of a puzzle during the time we try to solve the puzzle. All I see is the "par", my current score and my best score. If the highscore can't be seen during the puzzle, I think it would be very useful for the programmers to create such a feature. And allow me to explain to you why.

When I succeed on matching the "par"-score I keep going, trying to make better my solution, but I don't know when and if I have the highest score. If something like that existed, then I would try even harder in order to become the best. This way players (such as myself) will have a reason to create better solutions. All for the good of science

waldispuhl commented 11 years ago

We can indeed display the highscore. The reason why we didnt displayed it is that all solutions above the par are valuable for us. After re-insertion in the original genome alignment, the score of your solution may change a bit. Sometimes, the best solution at the end are not the highscore. If we provide the highscore the first time you are playing the game, you may be tempted to reach that score at all cost. While we would like to use your creativity and see if you can provide a solution different from other players. As you noticed, you have access to the highscore at the end of the puzzle anyway. That said, it is true that we could still try to display the highscore to see if it helps to achieve better results. Thanks for the suggestion.

phaethonas commented 11 years ago

OK thanks for answering. I didn't know that all solutions that are better than the par are valuable.

fuzzyBSc commented 11 years ago

Still, coming up with a solution and being told "two stars" at the end is a bit jarring. To put you in the mindset of the player, I'm looking to come up with a solution that I know is valuable. Currently a high score is rewarded with three stars while above par is only rewarded by two. In-game I don't know whether to keep fiddling because I have a valuable solution or to move onto the next puzzle. If three stars are given for a high score then it's frustrating not knowing in-game whether I've reached that high or not. It seems like the scoring metric is not quite right - maybe three stars should be offered for a unique above-par solution rather than a high score solution, or the three stars should be calculated in some other useful way. I think the way scoring is currently structured acts as a disincentive to future play. Something in-game should be telling the player "you've created a good solution here and will be rewarded for it on the score screen, you can move on now".

waldispuhl commented 11 years ago

So far, we tried to keep it as simple as simple as possible, but indeed the scoring metric could be improved. I like the idea to give bonuses to unique solution. We'll try to implement and test a new scoring system based on this principle in the upcoming weeks. And as you said we should make clear that unique solutions are helpful. Thanks!

rasmus-storjohann commented 10 years ago

Say I am in the last round of a puzzle. Say I've had some good early rounds well over par. In the final round of the game I have no idea how to reach par, maybe I'm only one point short. The star button is achingly out of reach. My only option is to kick the bucket, which I assume means that my nice early alignments will be thrown away. Not being able to beat the computer in the final round does not invalidate my good alignments early in the game, or do they? I'm thinking, well, computer, if you're so smart, why don't you show me how you got these extra two points, eh? But the computer keeps its cards too close to the chest. Frustrating.

rasmus-storjohann commented 10 years ago

As to the high score issue, if it is the case that any solution above par is potentially valuable, then showing the high score in the end is counter-productive. The goal of this puzzle is not actually to optimize to your scoring function, since it is imperfect, so don't encourage that mode of play. The simple solution to this problem seems to me to be to not show the high score at all. It changes the game incentive from "you're the best" to "you did well", which is accurate, since "best" is not something that the software can judge. It will also reduce the influence of your scoring algorithm on the results you get, which could be good or bad, but I'm thinking that if you're trying to leverage our brains, it could be mostly good.