5 replies [Last post]
Tlaloc's picture
User offline. Last seen 10 weeks 2 days ago. Offline
Joined: 08/04/2008
Groups: Mojo Risin'

As you may know, the recipe rating system is not reliable. For one thing, as changes are made to a recipe, old votes may no longer be relevant. There may be bugs that were fixed, or optimizations to speed up the recipe. It doesn't take too many votes to move a 4.85 score recipe to a 4.35 score.

I know with my own recipes, that some are vastly underrated as they consistently produce points. My "tlaloc hydrophobe" recipe has a 4.7 rating, but is, I think, overrated. My "tlaloc gyre" has a 4.25 rating, yet is a consistent point gainer. "tlaloc rebuilder" also has a 4.25 rating, but is also underrated.

So what recipes (with more than 10 votes) do you think are underrated?

Joined: 10/23/2014
Groups: Contenders
Underrated Recipes

Can't think of any recipes I would rate much higher. Some recipes take a very long time to complete and may be downgraded as a result.

This does bring up what I consider more important recipe issues, namely lack of standardization. This should also result in more accurate ratings as well. Is the recipe for mutation, compression, stabilizing, etc.? In other words categorize the recipe. When's the best time to use it? About how long does it normally take to complete (using set standardized puzzles)? Not all players are knowledgeable of the terminology used. A simple explanation along with are more complex explanation should be included. This and much more information should be noted.

There should also be a standard on how recipes are started, ended, or cancelled. Each recipe should display the score it's using at the beginning. At the end it should clean up and reset to the best score or let you know it's not. If user cancelled it should act as if it ended normally. These LUA functions should be standardized and shared. Authors should strive to keep the recipe output the length of the display window or at least show the most important information first.

I can come up with many other issues but I'd like to see what other think.

Joined: 04/20/2012
Groups: Go Science
Recipe Output Length

On a PC, the Recipe Output gets printed to the scriptlog.default.xml file. If you want to keep the Recipe Output for later, you can rename this file, otherwise it gets overwritten when the next recipe is run. The Recipe Output Window is only about 11 lines long.

Joined: 10/23/2014
Groups: Contenders
Recipe Output Width

Thanks Jeff. I was aware of the file and have on occasion looked at the data.

The comment I made about the recipe output related to the width of the viewable area, ~57 characters. Of course you can use the horizontal scrollbar to view any text not in view. However some recipe writers waste much of the viewable area with such things as displaying numbers to 9 decimal places (3 is sufficient), spelling out long words instead of using standard abbreviations or shorter words, using words to identify obvious items when none are necessary.

I may use the recipe output screen somewhat differently than others. I have a fast desktop and can easily run 3 or 4 other programs while running foldit. Occasionally I may look at the progress of a running recipe and if it's not preforming as expected I cancel it and switch to another. That always includes looking at the recipe output. I know it's probably a trivial matter for some but for someone who writes computer programs it's unnecessary and a little annoying.

Joined: 04/20/2012
Groups: Go Science
You can always make & share your own versions of shared programs

Being a coder, you could always change shared programs so their output matches your own tastes. If you share the results and others like the changes you made, you might get more votes for your own version than the original version got.

When I have suggestions for improving certain programs, I usually e-mail the author to let them know or make the change myself. We all have different ideas of perfection.

I'm generally grateful that others take the time to write Recipes and share them, and if I like a Recipe made by a certain author, I am more likely to use other Recipes made by that same author. Also, if teammates like a new Recipe and use it a lot, I am more likely to try it myself.

Joined: 10/23/2014
Groups: Contenders

I have made some minor changes to some recipes and use them for personal use. To me it seems deceptive to take someones recipe, make a few trivial changes and then post it as my own. Some of the authors don't even play anymore so no way to contact them to ask for changes. Some recipes have so many versions with minor changes to the original it makes it hard to determine which one to use. You either stick with an author who's past recipes seem to work well or examine the code. A better version may be overlooked just because the author is not as well-known.

No one is asking for perfection. I have a number of books on "best practices", recommendation on what practices work well with code, interfaces, etc. None write about perfection. But they all write about generally agreed upon standards that can help make better software.

For the first 6 months I didn't join any group and worked on puzzles on my own. Which recipe to use was a crapshoot. After much trial and error I eventually reduced the number of recipes to around 40. Now after joining a group and seeing what recipes they use and more importantly when to use them I use generally use less than 10.

Don't get me wrong, I am also grateful for all the recipes people have spent time creating. There are some amazing recipes. I've done a few LUA recipes from scratch but don't have anything I'd like to share because there are better recipe solutions available. And I'm aware not all recipe authors are programmers or know what constitutes "best practices" however a half a dozen "best practices" pages, some simple "best practices" functions, and a few simple "best practices" recipe examples would be a good first step.


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, RosettaCommons