Puzzle progression - Puzzle requests

  • 1
  • Question
  • Updated 5 years ago
Hi all,

We're working on building puzzles for the progression and we could use some help.  In some cases we have a puzzle specification, but no puzzle.  So I thought I'd make a new thread specifically for puzzle requests (main thread about the progression).

For example, here is a puzzle we would like to have:

A moderately sized puzzle (~30 nts) that requires the use of all bonds, and includes at least an end loop and an inner loop (or bulge). We can lock some of the bases, but the key will be that a GU bond will be necessary (i.e., have a locked U or a locked G where only a G or U can go respectively).  This would be the sixth puzzle in the progression as it currently stands (Google Doc).

If you 1) know of a puzzle like that or 2) want to build a puzzle like that, please let me know!

Thanks!
Ben
Photo of bekeep

bekeep, Learning Researcher

  • 98 Posts
  • 20 Reply Likes

Posted 5 years ago

  • 1
Photo of whbob

whbob

  • 218 Posts
  • 68 Reply Likes
I have not found an ideal puzzle as yet.  I did come across this link in the puzzle walkthrough wiki:
(Just Another Small Puzzle 12)  It's too long for your requirements, but a very informative walk through.
Photo of bekeep

bekeep, Learning Researcher

  • 98 Posts
  • 20 Reply Likes
Thanks so much for this!  Not only is it helpful for the task at hand, I also learned a lot.  JASP 12 might not fit the specifications exactly, but it might work (or I might be able to tweak it a little to work).
Photo of bekeep

bekeep, Learning Researcher

  • 98 Posts
  • 20 Reply Likes
Hi all,

I have another request.  We have a few Eternabot puzzles in the current progression (these are from the PNAS paper). Earlier, Nando previewed "The Key" and it was helpful to get your takes on what was working and not working.  The Water Strider and Making it up as I go are more difficult (for me, at least) and I'd love to see some other solutions.  Post your highest score and the sequence and we can talk a little bit about what works.

http://nova.eternadev.org/game/puzzle/3390166/


Highest I have: 88.69 with the following sequence: GGACAGUACAGUCUGCUCGAAAAAGAGUGGACAGAGGCUACAGUCGAAUAGACAGAGGAAUAUACUCGUAGACACACUCAGUACAAAGAAACAACAACAACAAC

http://nova.eternadev.org/game/puzzle/3390165/

Highest I have: 85.55 with the following sequence:
GGAAUGGACUUGGCGACACGCGAAAGUGCAGGAAACUGGUCAUAUACAGGACGAAAGUCGAGGAAACUCACUGGCUGAGACCUAAAGAAACAACAACAACAAC

I'm sure you can blow my numbers out of the water.  Thanks!

Best,
Ben
Photo of Eli Fisker

Eli Fisker

  • 2328 Posts
  • 541 Reply Likes
Hi Ben!

http://prntscr.com/8z1rec

I recall trying an earlier version of EternaBot with score. And I recall thinking that it was judging way too harshly on an relatively easy lab puzzle. There were multiple along the variations I made, which didn't made the score cut, that I judged to have a shot at winning in an actual lab. So I think it was judging too harsh.

Similar in the case of the Water strider, which was a real hard lab. Of the kind that I call pressured. This kind of puzzle require a higher than average GC ratio to have a shot at solving in lab. That could be useful for a teaching puzzle. However trying to solve it in EternaBot is going to be trouble for players. 

Eternabot doesn't even score the actual player made eterna classic lab highscorer 92% very positive.

http://www.eternagame.org/game/browse/479426/?filter1_arg2=479892&filter1=Id&filter1_arg1=47...
http://prntscr.com/8z1uqe

Which was later made into a full blown winner (96%) in the cloud lab.
http://www.eternagame.org/game/browse/5634134/?filter1_arg2=5706005&filter1=Id&filter1_arg1=...

So even an actual winning design won't make EternaBot happy.

What I'm saying is that this particular puzzle is better if made into a teaching puzzle on that designs with many short stems are better off with an unreasonable high GC ratio, something which I usually won't recommend.

Check this post on pressured lab:
https://getsatisfaction.com/eternagame/topics/_market_strategy_energetic_pressured_designs?topic-rep...

A good puzzle for EternaBot score testing is from a lab with a very big amount of winners (in particular if EternaBot has also been capable of making winners) as this will give players a better shot at getting in the range of a winning score.
Photo of bekeep

bekeep, Learning Researcher

  • 98 Posts
  • 20 Reply Likes
Hi Eli!

Thanks for this.  That makes a lot of sense.  Omei said much the same thing to me within minutes of your post.  I should have tested some of these designs myself before posting.  I'll rework things a bit, keeping this in mind.

Best,
Ben
Photo of Eli Fisker

Eli Fisker

  • 2328 Posts
  • 541 Reply Likes
Np, best wishes!