Some puzzles (like single nucleotide puzzles or most line puzzles) are solved upon opening (“space bar”). Other puzzles can be fun but completely scientifically and algorithmically uninteresting (like a Xmas Tree solves).
New players enjoy the fun of creating their first (usually simple) puzzles. Some gravitate to puzzles like those in #1
Players should be free to ignore puzzles that a “weak”bot can easily solve; especially if ignoring them allows them to spend more time on labs or puzzles that may be interesting or notable.
Players interested in competition cannot currently ignore the space bar type puzzles because every point counts.
Possible solution:
Create a ‘weak”bot that attempts to solve puzzles as part of the puzzle publish process. Puzzles that can be solved by weakbot would still be permitted, but would be assigned a score of zero points.
Weakbot algorithm
pushes space bar: to test if already solved
completes any pairs with one locked base as follows:
o if locked A then fill other base of pair as U
o if locked C then fill other base of pair as G
o if locked G then fill other base of pair as C (intentionally ignoring GU match)
o if locked U then fill other base of pair as A (intentionally ignoring UG match)
fills in all other pairs as random mix of GC and CG (finishing as XMAS tree)
checks for successfully meeting the shape AND for meeting any constraints
Any puzzles that solve this way would be valued at zero points when they emerge from the publish process.
Discussed in chat with RedSpah, and hoglahoo and Eli and starryjess
Might run “weakbot” on some tiny real but interesting RNAs to make sure they won’t fail too often before implementing. I would worry if it found too much of natural RNA forms unworthy of further examination
eternacac’s reply makes sense to me. strands with no shape like the lines of a single nucleotide or even just one nucleotide aren’t (I think) going to be interesting. And we saw that XMAS trees do not work in the lab.
Very good idea. There needs to be some automatic (or at least semi-auto) way to separate trash puzzles no one wants to see nor play for any reason other than points, from real, challenging puzzles.
What about locked puzzles that are easy to solve but is meant to be a tutorial . If it is marked as a zero point puzzle , players new and perhaps some not so new players could skip it xnd not benefit from the tutorial .
Fair point. Like Hyphema’s point, the proposal would also mean that Eli’s Classic Lab Failures series would come in at zero points which I believe is ok.
In a perfect world, we should have a separate category for TUTORIAL puzzles. I think it will be sufficient to communicate that you can find these helpful puzzles by searching on “TUTORIAL”, even if there is zero value.
I like the idea of having a new category along with challenge and player puzzles . One for tutorials . A great idea lroppy. However, jnicol’s switch tutorials are increasingly difficult. And should be given point value.
I have a collection of tutorial puzzles. The idea has been to collect them for it to be easy to find them, and with the wish for getting them put somewhere collectively in the future. I would love a separate area for both the player made puzzle tutorials and lab tutorials.
Mat, I think that’s a great idea. The option to test the ‘strength’ of the puzzle is a great option to have. Especially If newbies want to test out their puzzle making prowess.
Thanks everyone for reviving this discussion. As some of you may know we are exploring the use of scripts as ‘boosters’ within the game, both to aid advanced players in tackling complex lab puzzles and also as a potentially very fun and transformative game mechanic for players at earlier stages.
Based on the above, weakbot (perhaps with a different name) may be one of the first ones that newbie players would take advantage of – they would then be able to focus on honing their skills on non-weak aspects of puzzles.
A thought came to me last week during/after a dev chat.
What if puzzle maker included a script which was similar to a puzzle solving bot, but instead of solving the puzzle, would predict how difficult the puzzle would be for a human player to solve? When I mentioned this in chat, some players interpreted this as being similar to jnicol’s script that predicts difficulty based on the history of past solves (how many solves and how quickly they were done.) But this script would be different in that it would have no solution history (in general) to work from. I would propose using John’s current script as the criterion for judging between candidates for this new script – the better the new script’s predictions correlated with John’s script’s evaluation, the higher it would score in a competition between scripts.
Assuming a fairly good prediction was possible, the result from the script could then be used to assign a point value to each new puzzle. A space bar script would have a solution reward of 0 points, and an intermediate difficulty puzzle might still be 100 points, but puzzles predicted to be harder would have higher rewards.
Additional thought. I think that high reward of really hard puzzles will still make many players go nuts on solving puzzles that only run in the simulation, instead of getting good for the labs.
As a way to alleviate this, I propose a favoring of the highest rewards, to go to the switch puzzles. Since we need a lot of switch kung fu for the labs.
