A new type of puzzle?

  • 4
  • Idea
  • Updated 9 years ago
I have been told many times from advanced players that EteRNA puzzles are now too easy. Once you know the strategy you can clear any puzzle by just following the strategy.

This is becoming a major problem, since players are solving puzzles faster than we can create.

I was thinking of adding a new type of puzzle to give our players new challenge.

How about a puzzle that'll ask you to come up with a design with exactly specified # of GU,GC,UA pairs and free energy?

There'll be very few solutions for each puzzle, so you can't expect to stumble upon the solution space by just following strategy. You'll have to carefully calculate energy contributions of all loops and stacks to successfully arrive at solution.

It's just my personal idea at the moment. It would be quite easy to implement, but I'm wondering what our players think of this.

Jee
Photo of Jeehyung Lee

Jeehyung Lee, Alum

  • 708 Posts
  • 94 Reply Likes

Posted 9 years ago

  • 4
Photo of Chris Cunningham [ccccc]

Chris Cunningham [ccccc]

  • 97 Posts
  • 13 Reply Likes
You could try involving the dot plot in the challenge: if you can get a readable version of the dot plot, you could require something like "no more than 10% bond probability in each unbonded pair." Combined with a GU challenge this would remove the repetitive strategy of GU GU AU UG UG UA GU GU AU ....

Good to see some brainstorming!
Photo of Ding

Ding

  • 94 Posts
  • 20 Reply Likes
I like this idea a lot, and also ccccc's idea (if we get more readable dotplots, and if it's possible to set reasonable percentage goals - I have no idea what the bond probabilities look like in "real" RNA).
Photo of dimension9

dimension9

  • 186 Posts
  • 45 Reply Likes
I say we give it a"trial" implementation - see how it works in actual practice
Photo of JRStern

JRStern

  • 42 Posts
  • 2 Reply Likes
I recommend strongly against that, I can't see it justified in either game terms or science terms.

I could probably dream stuff up, but even if everyone liked it, it might involve a lot of implementation work.

Here's what I'm thinking, is that it would be both scientifically, combinatorically, and gamefully interesting if we could somehow investigate the design region around a solution. This is similar to restricting bonding energies at every pair - and I'd like to *see* that in any case, and it might even help in solving harder puzzles. But for a given solution, I wonder - how many pairs can be swapped in ones and twos etc, without losing stability from this point? That might be an interesting thing to pursue!

I'll leave the design details to the team, see what they might dream up along those lines.

Anyway, the idea is to *expand* the puzzles, do not contract them by making a single (and arbitrary) target configuration the goal.

My two cents, anyway. Thanks for the game as it is!
Photo of JRStern

JRStern

  • 42 Posts
  • 2 Reply Likes
I just thought to add, that the GU challenges *are* difficult, at least for me, even on a small puzzle, a small error can block you out of the last couple of counts, and no easy way to bridge the gap. I'm not totally sure what the point is, actually, as I've said and asked before, but as puzzles go, that aspect is already challenging.
Photo of mat747

mat747

  • 130 Posts
  • 38 Reply Likes
I agree with D9 we need a "trial" run first.

Also the "new type of puzzle" would be best implemented after with get #431 "new energy display", if we are going to "carefully calculate energy contributions of all loops and stacks to successfully arrive at solution" a better/easier way for us to view the energy is needed. To view the stacks energy now is not an enjoyable process.

"major problem, since players are solving puzzles faster than we can create."
I believe most of the top player are waiting for the "Player Puzzles" to be implemented and yes i`m aware of that it will not be implemented in the near future.

Question
How and what app is used to make the Challenges puzzles now?

Mathew
Photo of Berex NZ

Berex NZ

  • 116 Posts
  • 20 Reply Likes
What is the purpose of giving us more complicated puzzles to solve?

Isn't eterna a radical experiment to have citizen's involved in cutting edge science?

The puzzles are merely there to make sure, we know the fundamentals. I think its been proven that some of us have grasped it really well. We've proven we can ace the puzzles, which is merely a practice ground for the lab.

You wanted aces, you have them. The last shape we did, we had 8 designs at 94% or higher.

If the goal is to satisfy the advanced players, shouldn't energies be focused more on ramping up, in stages, to the elusive 100 synthesized samples a week, and running a multitude of labs? At what stage does it become critical mass for that to occur?

If thats not possible in the forseeable future, I guess we'll have to settle for Player puzzles. But again, to what end? So then we can create our own puzzles to create bigger time sinks for ourselves? I don't know about others, but the best premise for me, with eterna, was the chance and ability to contribute to science. And puzzles are a dead-end.

Sorry if this sounds rant-ish.
Just don't want valuable resources diverted, if it doesnt need to be.

Sincerely
Berex
Photo of Chris Cunningham [ccccc]

Chris Cunningham [ccccc]

  • 97 Posts
  • 13 Reply Likes
I agree with the idea here.
Photo of dimension9

dimension9

  • 186 Posts
  • 45 Reply Likes
hmmm, I've had this post on my mind for a couple days now... at first, it kind of bugged me - I felt it was "overly critical, and "rant-ish" as Berex himself said, ...but then, I began to think about it more, ...and now, this actually has me asking myself, "What IS the point?" ...and "Why AM I here?" ...and "What am I actually DOing here?" ...and I'm having some real trouble coming up with satisfactory answers... hmmm... a bit disturbing...
Photo of dimension9

dimension9

  • 186 Posts
  • 45 Reply Likes
IDEA: "QTTN" Puzzles (Quest to the Native)

I must preface this post by saying that I am not a big fan of the G-U competitions, I think that the G-U competitions are too much like "keep busy" work - merely something to keep us players occupied during the long weeks between labs (the main attraction of EteRNA); and that there is little point to them beyond achieving the purley artificial goal of creating the most G-U's in a puzzle. One would never design a Lab Submission using the maximum number of G-U's ; so, for this reason, it seems to me to not be a practical, valuable skill to develop in terms of the ultimate future goals of EteRNA.

Ok, that being said, back to the point of this post on New Types of Puzzles.

I would like to see a new type of puzzle which has practical, real-world training value for the kinds of Labs that EteRNA is aiming for in the long term (the kinds of things written about in the "Where are we Going?" section of the "About EteRNA" page).

http://eterna.cmu.edu/htmls/abouteter...

And in pursuit of that goal, I think it is time to finally seriously implement (as soon as possible), Bug Tracker Item #269 (Create a Gallery of Native RNA molecules in EteRNA Game Format for players to study).

Get-Sat Post:
http://getsatisfaction.com/eternagame...

Bug Tracker:
https://spreadsheets.google.com/pub?k...

I believe that the implementation of this idea would segue nicely into this whole discussion of "New Types of Puzzles", as I am about to put forth and explain...

The MAIN idea is this:

Create a new type of puzzle called (as in Fold-it), a “QTTN” Puzzle (Quest to the Native).

I envision this as a puzzle where the correct solution will be either the exact native configuration, or within a pre-determined range (in number of correct bases) - of that perfect native - (or, in order to facilitate "levels of Play", within successively closer ranges to that native configuration).

Perhaps as an adjunct to this idea, there could also even be two or more "alternate target folds", that the player could choose from in order to more closely approach the specifications given for the native? (I think this part of the idea may only be possible with some significant interface updates, but it just occurred to me as an interesting addition)

This QTTN idea would pretty much necessitate the immediate co-implementation of the above-mentioned #269 Native RNA Gallery Item in the Tracker, so that players could begin to familiarize themselves with how things are similar or different in native configurations - from the puzzle configurations everyone is used to seeing, and begin to internalize whatever kinds of "do's" or "don'ts" they are able to extrapolate from studying these native formations.

It might also mean an “unlearning” of habits that were valid and useful (or at least accepted and not penalized) in our artificially devised puzzles, but which are perhaps not appropriate - or even workable - for “Native RNA Configuration Quests." (such as leaving long stretches of consecutive Adenines in loops... Does this ever really happen in nature? I honestly do not know but I seriously suspect not).

Once the Native Gallery is up, and players have had a chance to study it, the Devs could begin to offer some small, easy QTTN Puzzle Challenges, and progressively get more and more complex in order to break into it gently, and get players up to speed on it, incorporating whatever lessons we were able to derive from the "Eterna Game Version Gallery of Native Folds."

This idea is actually very much like the idea jee already proposed in the opening of this Get-Sat post; just a much more specialized case of it with more far-reaching goals and implications, and a much more definite description and composition. It would also represent a quantum leap (in my estimation) of the relevance of EteRNA Game Puzzles to the eventual stated goals of the EteRNA Project, by advancing Player's RNA folding education and experience away from the current somewhat artificially constructed puzzles, and more toward "real-world" RNA molecule emulation and construction within the EteRNA Game Framework.

Having this New Puzzle Idea tied to “Native” RNA’s is, I believe a necessary key factor to really stimulate player interest. I think Players are here at EteRNA largely out of a desire to not only have their gaming efforts benefit Science, but to be able to clearly see HOW, and I think that changing the goal and focus of a Puzzle Type Series to sucessfully figuring out Real Native RNA Folds would go a long way toward capturing Players' imaginations and re-kindling their interest and excitement about being a part of the EteRNA Experiment.

I believe there is a real hunger for this more advanced type of gameplay amongst the player community, and that implementing this idea wil reveal it.

Here are some of the advantages of this Idea that I percieve at this early stage: (I'm sure others will think of more as we get further into the discussion)

1) It will begin to educate players in a "real" practical skill - that of learning how to recognize and assemble RNA structures according to Real RNA found in Nature. (I believe this element of "realism" , or at least a more visible connection to "real" Rna Folds will prove to provide a significant boost to player interest and involvement)

2) It will pique player interest in a way that an artificial challenge (like G-U competitions) never can or will, and avoid the unintended "negative" effect that the G-U puzzles have begun to create: The feeling that the activity has no “Point” beyond itself (see Berex's recent post regarding this feeling of "pointlessness").

3) It will better prepare the players for the eventual more advanced project goals mentioned in the "Where are we Going?" section of the "About EteRNA" page.

4) It will offer Players who excel in QTTN challenges a new path to recognition in the greater EteRNA community at large, and perhaps, even if they are new players, significantly increase their chances of getting synthesized in Lab, where currently, many extremely talented new unknowns languish in relative obscurity and growing frustration.

5) It will make players feel more like they are really learning something of value beyond just becoming good at a stimulating mental exercise; it will connect PURPOSE much more solidly with game techniques, and with game achievements.

In fact, I can even envision this kind of "QTTN" puzzle possibly becoming a sort of "Master's Level Program" of EteRNA Game Play Education and Lab Preparation, with all the current puzzles assuming a more basic preparatory role.

The Lab, of course will remain the Centerpiece of EteRNA Goals and Efforts, for which all the other activities are preparatory, skill-honing devices, but this new kind of challenge could even conceivably come to occupy a kind of pinnacle position among EteRNA preparatory puzzles – perhaps even elevating lab skills of acknowledged top players.

I hope the Devs and Players alike can find value in this idea, and that ideas for specifics of actual implementation can begin to be discussed as well as ideas for improvements, and solutions for any possible implementation difficulties which may arise.

Thanks, and Best Regards,

-d9
Photo of alan.robot

alan.robot

  • 91 Posts
  • 36 Reply Likes
I agree with Berex and D9, and would actually combine their points:

As an arbitrary goal to make more puzzles, I think it's not particularly useful because it doesn't teach a skill that translates to better synthesis (in the absence of an a-priori way to know what free energy target a successful design needs). As it stands I would bet the money in my pocket an all GU-design will be as successful as an all GC or all AU design in real life, so the most difficult puzzle constraints are already out-of-line with learning design skills.

Given limited manpower, making more ambitious design targets would be my vote for a priority over creating more puzzles.

And regarding the design targets, I understand the desire to systematically create a shape library in order of increasing complexity, but it may pique interest of the players more if you intersperse basic shapes with more ambitious, complex targets where no-one may score well to start. Berex's point that there were 8 scores above 94 percent for the last competition tells me the challenges aren't hard enough to really challenge the top designers, and therefore perhaps the most potentially interesting outcomes of this game are not being harvested yet.

That said, shifting to D9's point, I DO think making more extensive use of partially "locked" nucleotide puzzles that come from naturally occurring sequences could be useful in multiple ways. Say, the locked region is required for RNA function and can't be altered, and you have a CG/AU/GU ratio requirement for the answer. This would indeed force players to know the energy functions really well, as Jeehyung imagines, and it would also be a useful design skill.

I really liked the "strategy guides" that were being created for that reason, perhaps if similar, small-ish examples are mined from actual shapes of interest in anticipation of future design challenges. . .
Photo of JRStern

JRStern

  • 42 Posts
  • 2 Reply Likes
I like the idea of a gallery, I like the idea of partially locked real-world puzzles.

I don't like the idea of making the real-world config the target. Nobody knows the optimality of real-world configs, and I don't recommend the idea of narrow, arbitrary targets.

Anyway, is real-world sequencing the target of the lab, or lab-engineered molecule folding? Those might be rather different.