This is a thread for comments, suggestions, and questions on the new style of exploration puzzles.
I think the new set up is much much better! My only critique is that, in the situation, where someone has a slightly negative score, and do some action, that increases the exploration score, and normal, energy state score remains the same, the score gets more negative. This scenario was why I suggested the idea of an "absolute zero" analog to scoring, and then some other mathematical normalization of the score.
But overall, this is quite significantly more efficient at accomplishing the goal of increasing diversity without the other confounding effects that could lead to a massive exploration score by having everything compressed to a single point, and without the discouraging part of the old design in which someone could play for hours and not even get credit.
It's an innovative response and now allows players on the board, and to get involved, which can only be an inclusive and positive step.
I'll be curious however to see what you give us as 'best options' in round 2 and 3.
Since the focus of the New Exploration, is exploration, discouraging helix-chasing was probably a necessary step. And I hesitate to say it, but it seems that 'sheet-stacking' may have taken it's place as the new easy way to score well in the New Exploration framework.
But it's just an observation.
And there also needs to be an LUA coding change to bring Explore puzzles in line with normal puzzles.
get_ranked_score, is a great addition, but needs an aligned set_recent_rank_best and restore_recent_rank_best.
(Although, as a non-programmer, I'm sure someone will post a derisory fix to this problem, as soon as I post this)
WIP I guess...
The intuition with these puzzles is we have a rough idea of how far the native is from the starting structure. We've set the parameters of the puzzle such that the native will have a exploration multiplier of around 3. The native should also theoretically have the highest stability score. So after the multiplier, the ranked score of the native should be greater than any other structure.
In light of this, I'd be interested to see if any sheet-stack solutions end up in the top scores.
Such things will give you a high multiplier immediately, so your score will shoot up very high in the beginning, but you'll have trouble when it comes to stabilizing. If someone can come up with a more reasonable structure which also has the multiplier near the max, they should win by a significant margin (even if the structure isn't the native or even near it).
Sheet-stacking may end up being a strategy just to get up to somewhere reasonable on the scoreboard without too much effort, for people who don't intend to actually be competitive, but just want points. We will see whether this becomes a problem.
How do we find out about the existence of new functions such as get_ranked_score() and get_exploration_score()?
They should be documented in the release notes that are distributed with the updates.
I restarted my 414... Ripped it apart, and quaked it. Then rebuilt it. Maybe I just dont get what you're reaching for...?
Okay... After almost three years of awarding 100 points for the best scoring solution to every puzzle, someone please explain the logic of assigning only 25 points to the top score of #414.
Not very fair that the top scoring solutions are worth only 20 or so points more than someone ranked, say, in 180th place.
Should I be spending less time on these puzzles, since they are worth less points?
The intention was that you spend less time on these puzzles, yes. However we're not sure people are interpreting it in that way, so we may have to adjust.
There's a lot of users these days that are uncomfortable prowling around the file system looking for a release notes .txt file.
Why not provide an option either at the first login after an update, or within the main game display to bring up the latest release notes for review?
Another option would be to put a release history on this site, so people can go back through the releases to see what's been implemented that they don't know about.
Because I don't have much of a clue still about what will or won't work, I usually make a series of large and small moves with wiggles and fuzes between them until something starts making sense and scores reasonable well.
That method is a disaster on these puzzles because I get a high multiplier too early and then can't get my stability score up.
Also having to explicitly save every few minutes takes most of the fun out of it. (and can also lead to harm being caused to those in proximity when the save cannot be found.)
Are these puzzles designed only for people who can look at the puzzle and immediately know how to correct them? Thereby getting a nice stability score. Then they can go on and explore and build the multiplier.
I'm quite confused by it all.
(By the way how does sheet stacking work? :-))
These puzzles came about because players often tend to drill down on folds which are close to the starting structure. This can be helpful on some puzzles, but on others, we know that the native fold of a protein is actually further away, and so this tendency hurts the chances of finding the native.
In particular, often times we derive the starting structure of a puzzle from its sequence's closest known homolog. In layman's terms, we know how to fold something that is "close" to this protein, so we use an alignment from that fold as our starting structure.
This method also provides us with an estimate of how far away the native is from our starting structure. In the case of the most recent puzzle, the native is a distance away from the starting structure which provides a multiplier of around 2.95.
So basically, if you aren't close to this multiplier, you aren't close to the native. However, being close to this multiplier doesn't mean you are close to the native. The multiplier rewards all exploration equally, so you could have gone the wrong direction and ended up even further from the native than the starting structure. I am assuming this is why you are feeling like there's a lack of direction.
However in these puzzles, we're actually giving you MORE information than you're used to (basically, the extra information of how far off the native is). Your direction should be guided by your expertise in knowing which changes are obviously incorrect, and which changes have potential.
Exploring just once and finding something close to the native is near impossible, so we aren't designing these puzzles for players who, as you say "can look at the puzzle and immediately know how to correct them". If some attempt at exploration isn't developing as well as you would like, you should try something different. We are interested to see what strategies players develop to deal with exploration.
"Sheet-stacking may end up being a strategy just to get up to somewhere reasonable on the scoreboard without too much effort, for people who don't intend to actually be competitive, but just want points. We will see whether this becomes a problem. "
...or simply flipping some sheets from the starting position then shaking, since that requires very little effort...at least sheet-stacking takes more effort and skill than making a giant helix, so it'll be less of a problem, even if it tends to lead to structures with more sheets than there ought to be.