Remove all-hands scores and Revise tutorial

Case number:699969-991847
Opened by:jakbinimbol
Opened on:Tuesday, February 14, 2012 - 17:18
Last modified:Monday, August 13, 2012 - 07:09

Can we remove All_Hands Scores?
It is a relic. Thing of the past.
The year 2012 is for computer power, as you can see with the new recruits.
Four Core, Six core, 8 core computers on line.
Teaching new comers molecular biology is futile.
They want to play games and level up.
How many hundreds give up each day?
Because we force them to move side chains, tweak coil and sheets...
We have LUA, and programmers make it easy for new comers.
And programmers make more good programs each day.
Why pull side chains? why manually do re-structure? Why do mutate manually?
Why shake? Why wiggle?.....all is automated now.....
We have to revise the tutorial it discourages everyone.
Teach them Recipes.
Combining recipes, when to stop a recipe and start another one,
Use recipes that are useful..midgame, end game...
These are still human intuition that computers can not do.
Rav3n and Tlaloc you are heroes!
But to force players to be like us! Hand making? Learn Molecular biology?
Just to enter the game? ....
Then we are not helping foldit.
We just become a bureaucratic force, to protect our ranks.

The ideal situation is to make new comers to play a game like pacman and
at the same time curing cancer.
Let us use the resource they are giving.
Human intuition and computer power.


(Tue, 02/14/2012 - 17:18  |  14 comments)

beta_helix's picture
User offline. Last seen 2 weeks 5 days ago. Offline
Joined: 05/09/2008
Groups: None

kxu, can you remove the All-Hands scoreboard from the website:

Joined: 09/19/2011
Groups: Beta Folders

I'm relatively new and have had success using the recipes. However, there have been times when my highest scores resulted from hand work done somewhere in the start, middle or end of the process. There seems to be a bit of luck involved. I'd hate to see the all hands option disappear.

As a matter of fact, I'd like to try some all hands puzzles now that I've played around here for about 5 months. I'm convinced that those understanding the manual manipulation process will generally outdue those using recipes exclusively. The exceptions being the programmers themselves. I think they have a better handle on which recipe to use where.

Both manual and recipes are needed here. Everyone just has their own style of playing and some simply don't have the patience.

Where does one even find the all hands puzzles to play? I've been wondering where those scores are coming from when I see them in someones profile. I don't think the option is obvious enough.

beta_helix's picture
User offline. Last seen 2 weeks 5 days ago. Offline
Joined: 05/09/2008
Groups: None

sorry for the confusion smilingone...

jakbinimbol basically put 2 unrelated feedbacks into one.
All Hands was a puzzle type where we would pick top Foldit solutions from previous puzzles and post it as a "All Hands on Deck" puzzle where everyone had access to them and as players improved on them, the new top scoring solutions would replace the old ones. So it was basically continually sharing top solutions (essentially everyone was in the same Foldit Team).

Unfortunately, what would happen is that the top solutions would all converge to the top-scoring ones and it would end up with less diversification than we started with.

This is why with the recent "Top CASP ROLL R000X Predictions" puzzles we've tried to give you previous states of top-scoring solutions (I believe this was suggested by Pletsch as a new way to bring back All Hands puzzles: starting from middle-scoring solutions instead of top-scoring ones).

Anyway, as you can see we haven't posted an All Hands puzzle in a long time (the Exploration Puzzles were an attempt to get around this convergence problem) and infjamc already mentioned getting rid of those obsolete scores:

As for the tutorial levels, there is no plan to dramatically change these and they were recently restructured last fall (although we could add some recipe/lua tutorials).

brow42's picture
User offline. Last seen 3 weeks 3 days ago. Offline
Joined: 09/19/2011
Groups: None

Blue fuze should be a treated as a primitive function on the same scale as tweak and rebuild and should be a default cookbook recipe. Perhaps there should even be a tutorial level introducing it.

Joined: 04/15/2012
Groups: Beta Folders

That is also a good point. There are many new tools and "classic" user tools that aren't currently included.

Joined: 10/11/2011
Groups: None

hand folders are still ranked higher than "script only" players.....think about it

infjamc's picture
User offline. Last seen 6 hours 48 min ago. Offline
Joined: 02/20/2009
Groups: Contenders

Just out of curiosity, I conducted a quick investigation of the top 25 players of the current soloist ranking. It turns out that all but two of them have the "Perpetual Moving Machine" (1 million moves) achievement. (The two that don't happen to be frood66, who joined in September 2011 and has 937k moves, and me, who only accumulated 865k moves despite having played for almost three years because I mostly fold manually and only use relatively simple scripts). So, you could legitimately argue that extensive usage of complex scripts does provide an advantage over most other players.

That being said, script usage alone isn't enough for continued success in protein folding for the following reasons:

1. All things being equal, those with better hardware would have an advantage because the same script would run faster.

2. Even having the best hardware one can afford isn't enough, because the player needs to learn when to use a certain script.

3. Because of the computational complexity involved in representing the biochemistry of proteins digitally, brute-force approaches become much less attractive for longer proteins.

4. Scripts are essentially a form of AI, which are only as good as their design. They are the most effective when used to conduct repeatable protocols, but less useful when it comes to judgment calls related to human intuition.

5. Even if a player can get points #1 through #4 perfectly, there is still a chance that you could run a script overnight and gain zero points. From the viewpoint of cost-effectiveness (relative to carbon output and the opportunity cost of your time), the best strategy would involve a mix of script usage and manual folding.

Joined: 04/19/2009

I agree with infjamc's analysis, and have two other points to offer...

First - all of the current scripts were adapted from manual moves.... In other words - someone, either the scriptwriter or the person who gave the scriptwriter the idea, had to find out from manual moves what might work.

Second - there are still many things that you can do to a protein that even in lua 2, cannot be scripted.

Joined: 06/17/2010

No hand work, scripts only is FAIL.
There HAVE to be at least some manual thing/s prior sets/s of script/s

spvincent's picture
User offline. Last seen 7 hours 16 min ago. Offline
Joined: 12/07/2007
Groups: Contenders

I have the general sense that scripts are often run too early in a puzzle. It's all too easy to fire off a script last thing at night or when leaving for the day with the feeling "What's the worst that can happen?" , but the end result is often premature optimization of a bad solution. It seems really important first to get a good structure which scripts can refine later: it's all about finding the right balance.

I'd be intrigued to see the results of a puzzle with 2 rounds: no scripts allowed on the first round and anything goes on the second.

Joined: 04/15/2012
Groups: Beta Folders

I don't know why this has -3 votes. Players shouldbe able to know how to use all recources, and properly, at least to an extent. I don't script myself because I haven't been taught, and the wiki tutorials don't give me what I need. There should be something on it in the game itself. Hand folding may work well, but everyone should have the same chance. If the tool is there, people shouldn't have to stumble through it. They should be taught all the different thing they can do. Strategy tutorials might be interesting, showing user info like that is on the wiki, and this would inprove their folding. This is the whole point of folding! Doing as good as we can and always inproving. So if there is something to help, it shouldn't be kept back for those who can find the right info. Everything should be readily available. We want to have everyone at the top of their game.

Joined: 12/06/2008
Groups: Contenders

There should definitely be a tutorial on the usage of scripts. They do come in handy, when appropriately applied.

But the suggestion to get rid of the manual work to make this just another video game? Absurd.

Allow me the liberty to use an analogy.

Before automatic dishwashers were invented, folks had to manually grab each dish, run the sponge over it, top and bottom, maybe even use the scraper on the tougher stuff. Then rinse the dish, and place it in the rack to dry, or dry it by hand. With a large number of dishes, this could take hours.

Then the miraculous automatic dishwasher was created, probably by someone frustrated with the lengthy manual approach. Load those dishes, add soap, turn it on, and --- VOILA! --- clean dishes, within minutes. Some of the time. Often, dishes still had food waste stuck on them when the cleaning cycle finished.

The problem was, the person loading the dishwasher placed the dishes willy-nilly in the machine, with no rhyme or reason to the arrangement. Cleaning results were erratic, and often worse than if the dishes were washed by hand.

Jumping to folding....

Folding a protein is like washing dishes.

Manually washing dishes is laborious and time consuming. Given enough time and effort, however, it usually produces good results. Manually working a protein is laborious, and time consuming. Given enough time and effort, it usually produces good results.

Using a dishwasher automates the process of washing dishes, is easier, quicker, and when done right, produces good results. Using a script automates the process of folding, is easier, quicker, and when done right, produces good results.

Running a dishwasher without first scraping off the dishes and then loading them correctly (highly manual processes) produces inconsistent results. Folding solely by scripts without first manually working the protein into a better structure produces inconsistent results.

Washing dishes by first manually scraping off the worst food residue, then properly stacking them in the dishwasher, and then running the automatic dishwashing routine on the machine nearly always produces excellent results. The best results are made by an appropriate marriage of manual and automatic methods.

Folding a protein by first manually working on worst red parts, then properly placing the secondary structures in appropriate places, and then running scripts on the machine nearly always produces excellent results. The best results are made by an appropriate marriage of manual and automatic methods.

Summary: Manual folding is essential. Scripting is essential. Doing both is most essential to getting the best results.

Joined: 12/07/2011
Groups: None

My general point was to encourage more people to play, and not discourage them early.
If they learn recipes earlier on, it would make them level up and be happy.
They will invest more brain time, hardware, and $.
However, as they continue to play, they will discover that this will not be enough,
this is the time that they will learn that manual folding is essential to go to the top.
So, they will read the wiki, read the links, study, pep-tide bonds, cysteine bonds,
glyceine hinge, C and N terminus, why banding is necessary, why restructure, why do acid tweak,
why walk, etc.

The point is not to discourage them early.

Then we can have more players.

In our publications, we boast of thousands playing the game.
But in reality we are not close to 300 active players.
That is not enough!


Joined: 08/29/2010

I think some basic recipes should be added through the tutorial, such as Blue Fuse and also a compressing recipe! Rebuilding is ok in the tutorial but compressing the whole protein wasn't mentioned there.

And yes the tutorials should also teach some scripts! A basic that someone can easily create new Recipes on top of it.


Developed by: UW Center for Game Science, UW Institute for Protein Design, Northeastern University, Vanderbilt University Meiler Lab, UC Davis
Supported by: DARPA, NSF, NIH, HHMI, Amazon, Microsoft, Adobe, Boehringer Ingelheim, RosettaCommons