Recommend
6 
 Thumb up
 Hide
102 Posts
1 , 2 , 3 , 4 , 5  Next »   | 

BoardGameGeek» Forums » Gaming Related » Trades

Subject: suggestions for TradeMaximizer? rss

Your Tags: Add tags
Popular Tags: [View All]
Chris Okasaki
United States
White Plains
New York
flag msg tools
badge
Avatar
mbmbmbmbmb
TradeMaximizer 1.2c is close to six months old now. There haven't been any major problems requiring updates, and I've refrained from tinkering just for the sake of tinkering until we collectively had more experience with using it. But now is a good time for me to work on a new release, so I'm asking...

What are your suggestions, big or small? What annoys you about it? What could be improved? What fancy new ideas do you wish it would support? What interface tweaks would make your life easier?

I'm interested in hearing both from users (from the point of view of preparing your wishlists) and especially from moderators (from the point of view of running it).

Feel free to thumb suggestions by others that you really like. I don't promise to implement everything (in fact I'm sure there will be some suggestions that I have no idea *how* to implement), but I'll certainly try to implement a reasonable selection.

Here's a small suggestion to get the ball rolling:

Quote:
Allow # symbols in the free text of the official names section. For example, you could then say

1234-AOS1 Age of Steam Expansion #1

However, # still would not be allowed inside an actual item name.
 
 Thumb up
0.25
 tip
 Hide
  • [+] Dice rolls
Michael Leuchtenburg
United States
Cambridge
Massachusetts
flag msg tools
badge
Avatar
mbmbmbmbmb
cokasaki wrote:
Quote:
Allow # symbols in the free text of the official names section. For example, you could then say

1234-AOS1 Age of Steam Expansion #1

However, # still would not be allowed inside an actual item name.


Isn't # currently used for comments? Losing the ability to put comments on the same line as other things would be annoying. At the very least, some other character or character sequence should be assigned to "rest-of-the-line" style comments. //, perhaps.
 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Chris Okasaki
United States
White Plains
New York
flag msg tools
badge
Avatar
mbmbmbmbmb
dyfrgi wrote:

Isn't # currently used for comments? Losing the ability to put comments on the same line as other things would be annoying. At the very least, some other character or character sequence should be assigned to "rest-of-the-line" style comments. //, perhaps.


Right now, only whole line comments are supported, so the # is required to be the first non-space character. There currently is no way to put a comment on the same line as other things. I'm not sure whether or not "rest-of-the-line" comments would be helpful in practice, or if they would cause more errors in hand-generated want lists. I suspect the latter, but that's only a suspicion.

Changing # to be a "rest-of-the-line" comment would take care of the problem, which is that the moderators sometimes want to use # as part of a game name in the description part of the official summary. Right now, that causes TradeMaximizer to complain.
 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Jack K
United States
Shorewood
Wisconsin
flag msg tools
Avatar
mbmbmbmbmb
Chris, let me start by thanking you for creating such an excellent tool for us to use. I've run two events now and ran the TM for a third where the person running the event couldn't get it to run on their PC. I've found it to be really easy to set up and use.

Is there any way to have it be more forgiving of errors in want lists? JeffyJeff's OLWLG is not available for every math trade - there are so many now that he doesn't have the time to help with every single one - which leaves a want list cleanup chore for moderators of those events.

For instance, I've found that many times a user will forget leading zeros in an item number. Instead of putting 0137-PGRID, they might put 137-PGRID. Could the TM, on finding an error like that, check for a match with a leading zero or two and automatically use it? That would save a lot of time.

If I recall correctly, the TM does allow item numbers which are purely numeric. If that's correct, it would also need some sort of configuration setting to indicate if this "leading zero" feature were to be enabled, because while this would make sense for my example above, it would make no sense for trades where the moderator just chose to use numeric codes.

What are your thoughts on that?
 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Chris Okasaki
United States
White Plains
New York
flag msg tools
badge
Avatar
mbmbmbmbmb
anglotiger wrote:
Is there any way to have it be more forgiving of errors in want lists? JeffyJeff's OLWLG is not available for every math trade - there are so many now that he doesn't have the time to help with every single one - which leaves a want list cleanup chore for moderators of those events.

For instance, I've found that many times a user will forget leading zeros in an item number. Instead of putting 0137-PGRID, they might put 137-PGRID. Could the TM, on finding an error like that, check for a match with a leading zero or two and automatically use it? That would save a lot of time.


I've been a moderator and dealing with bad wantlists is indeed a major pain. It seems to me there are four basic ways to deal with it.

1. Use a tool like the OLWLG that prevents malformed wantlists from being formed in the first place.

2. Make the users fix their own mistakes before submitting. This would require a tool (presumably web based) that makes it easy for users to validate their wantlists before submitting. The tool could then append some kind of checksum to the wantlists, and the moderator can flat out reject any wantlists that don't have the correct checksum.

3. Have a tool that makes it easier for moderators to fix malformed wantlists. For example, it could use the same approach that spellcheckers do and, for each bad name, suggest a list of alternatives and let the moderator choose one. (Coming up with that list of alternatives is actually pretty easy based on something called the "edit distance" algorithm.) Such a tool would then output the corrected wantlists, which the moderator would publish for users to check.

4. Have TradeMaximizer simply pick the closest match for each bad name, and continue on its merry way.

To me, 4 is a non-starter. I guarantee that it would occasionally pick the wrong substitution. In your example, maybe the person who said 137-PGRID didn't really mean 0137-PGRID but instead meant 1437-PGRID. I agree it's much more likely that they 0137-PGRID, but even a 90% probability is going to be wrong 10% of the time. The resulting headache of cleaning up the mess after the fact would, IMHO, be worse than the headache of fixing the lists to begin with.

3 lends some automated support to what happens now, where moderators fix errors and then publish the revised wantlists to be checked. But you know not everybody is actually going to look at the revised wantlists carefully enough. Ideally the tool would somehow highlight where changes were made, but even so, mistakes are going to slip through. At that point, the moderator can say, "You had the chance to check your list, and you didn't, so you need to just live with the problem." This is doable, but less than ideal. (A variation on this approach just occurred to me. The moderator would make the corrections as described, but the tool would then email users about the changes to *their* wantlists for verification.)

2 might work pretty well as long as it was dead easy to use, and extremely portable across different browsers/platforms.

1 may be the best approach in principle. The problem there is that getting the interface right is HARD. Also, we need to find a way to get JeffyJeff out of the loop for everyday math trades, as I'm sure he would whole-heartedly agree.
2 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Burster of Bubbles, Destroyer of Dreams.
United States
Sunnyvale
California
flag msg tools
Just imagine the red offboard up here. I'll create it Real Soon Now...
badge
Yes, I know a proper 18XX tile should have a tile number.
Avatar
mbmbmbmbmb
I've been working on a post-processor for TM output, to generate per-person cheat sheets for in-person trades.

(I didn't quite have the guts to go into the TM code and do it directly.)

There are still a few things my scripts don't support:

1. People who use something other than their BGG userid as the prefix on their wantlists.

2. People with spaces in their BGG userids

3. An input table of BGG Userid + wantlist-prefix + Real.Name, so someone can look at their personal cheatsheet and see

Your trade partners:
joeuser // JoeJ // Joe Jackson

Your items:
0123-FLUXX Fluxx, Fairy Tale, and Coloretto (joeuser)

Give 0123-FLUXX to joeuser

(However, my per-person cheat sheets have been greatly appreciated even in imperfect form; I'll certainly be doing them again for all future math trades I run.)

[In my last trade I had:
-- A person with a space in their BGG userid
-- A person who submitted their wantlists with their name instead of their userid
-- someone acting as a proxy for a friend whose BGG userid she didn't know.
It was a nice added level of confusion.]

Chris, if you're interested in adding an option to TM to generate cheat sheets, please let me know and I'll see if I can code it up.

Math trade organizers, if you're interested in my imperfect scripts please send GeekMail.
 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Burster of Bubbles, Destroyer of Dreams.
United States
Sunnyvale
California
flag msg tools
Just imagine the red offboard up here. I'll create it Real Soon Now...
badge
Yes, I know a proper 18XX tile should have a tile number.
Avatar
mbmbmbmbmb
cokasaki wrote:


4. Have TradeMaximizer simply pick the closest match for each bad name, and continue on its merry way.

To me, 4 is a non-starter. I guarantee that it would occasionally pick the wrong substitution.


Agree 100%.

I've gotten wantlists where I couldn't be sure that I knew what the person intended. No AI is going to do better than me on that task.

JeffyJeff's tool is awesome, but it doesn't allow certain kinds of wantlist fine-tuning, and making it mandatory will discourage some traders. That's also not a good choice.

A Web tool that lets users proofread wantlists would be cool. Even just a little Windoze tool you could put a blob of text into to check if it is a theoretically-legal wantlist would be a big help, though mostly it would only catch the simplest formatting errors.
1 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Ernie Lai
United States
Forest Hills
New York
flag msg tools
Avatar
mbmbmbmbmb
if you changed the arrays.copyofrange line to use something else (maybe System.arraycopy) then it might be compatible with java 5 and therefore able to be run by mac users.

see this thread:
http://www.boardgamegeek.com/article/1993023#1993023
1 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Subhan Michael Tindall
United States
Portland
Oregon
flag msg tools
badge
Avatar
mbmbmbmbmb
I'd be interested in a command-line option to have it output all the solution sets it finds with running with multiple iterations, not just the most optimum one.
 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
James Perry
United States
Spring Hill
Tennessee
flag msg tools
badge
Ex-Term-In-Ate
Avatar
mbmbmbmbmb
subhan wrote:
I'd be interested in a command-line option to have it output all the solution sets it finds with running with multiple iterations, not just the most optimum one.


For what purpose?
1 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Chris Okasaki
United States
White Plains
New York
flag msg tools
badge
Avatar
mbmbmbmbmb
Morganza wrote:
I've been working on a post-processor for TM output, to generate per-person cheat sheets for in-person trades.


What is the information that you want on these cheat sheets? The name of the person, what games they are giving (and to whom), and what games they are receiving (and from whom)?

Right now you can get most of that from the summary, especially by using the REQUIRE-USERNAMES and HIDE-NONTRADES options. By default, the summary is sorted by username, so all the information for a single user will be together.

This definitely does *not* currently translate from a given username to somebody's real name, nor does it translate from a given item name to more human readable game names. I suppose the latter might be achievable using information from the official summary, but it could get very messy for complicated items.

I suppose the cheat sheet could hide the information from the summary showing that you are giving item X for item Y. If you are trading several items, then in some sense you care which items you're giving and which items you're receiving, but it doesn't really matter which is technically for which.
 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Jeff Michaud
United States
Longwood
Florida
flag msg tools
On-Line Want List Generator - Hopefully Making Math Trades a Little Bit Easier
badge
Captain Kirk, Captain Picard, Captain Sisko, Captain Janeway, Captain Archer
Avatar
mbmbmbmbmb
anglotiger wrote:
For instance, I've found that many times a user will forget leading zeros in an item number. Instead of putting 0137-PGRID, they might put 137-PGRID. Could the TM, on finding an error like that, check for a match with a leading zero or two and automatically use it? That would save a lot of time.

I think this could be a bad idea... you never really know what the user intended... for example what if there is also a 1137-PGRID or a 1370-PGRID... having TM auto choose 0137-PGRID would then be incorrect.
1 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Jeff Michaud
United States
Longwood
Florida
flag msg tools
On-Line Want List Generator - Hopefully Making Math Trades a Little Bit Easier
badge
Captain Kirk, Captain Picard, Captain Sisko, Captain Janeway, Captain Archer
Avatar
mbmbmbmbmb
Does all the class libs TM uses available from a java applet? An applet version may make it easier for those non-computer savvy users. They could just go to a web page which would load it as an applet and then the applet could have some kind of file chooser for the user to point it to the wants, and the results could be displayed in the window.

Else adding GUI support into TM to do something similiar.
1 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Jeff Michaud
United States
Longwood
Florida
flag msg tools
On-Line Want List Generator - Hopefully Making Math Trades a Little Bit Easier
badge
Captain Kirk, Captain Picard, Captain Sisko, Captain Janeway, Captain Archer
Avatar
mbmbmbmbmb
command line arguments (argv) to override options in input.

I'd like to be able to run the input through TM with different options, like trying the different type of priorities, but that requires editing the input file each time.

It should also allow specifying the location of input instead of using stdin. Even better would be if it would recognize the URL's and auto fetch the input from that location. Again this is where it would be nice to be able to override options as you proably won't be able to edit the input on the moderators web page.
 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Jack K
United States
Shorewood
Wisconsin
flag msg tools
Avatar
mbmbmbmbmb
cokasaki wrote:

I've been a moderator and dealing with bad wantlists is indeed a major pain. It seems to me there are four basic ways to deal with it.

1. Use a tool like the OLWLG that prevents malformed wantlists from being formed in the first place.


Which only works if (a) Jeff has the time to help and (b) you make it required for the event. Considering that math trades are greatly increasing in popularity and frequency, and that this isn't Jeff's full time job, that's not a reasonable solution (unless others can do the work in Jeff's place).

Quote:
2. Make the users fix their own mistakes before submitting. This would require a tool (presumably web based) that makes it easy for users to validate their wantlists before submitting. The tool could then append some kind of checksum to the wantlists, and the moderator can flat out reject any wantlists that don't have the correct checksum.


I wish that I had the time to work on something like that.

Of course there's the simple way - make users responsible for their own errors. Send errors back to them with instructions to fix them before the deadline. I hate doing that, but it's a reasonable approach.

Quote:
3. Have a tool that makes it easier for moderators to fix malformed wantlists. For example, it could use the same approach that spellcheckers do and, for each bad name, suggest a list of alternatives and let the moderator choose one. (Coming up with that list of alternatives is actually pretty easy based on something called the "edit distance" algorithm.) Such a tool would then output the corrected wantlists, which the moderator would publish for users to check.


Can you imagine the hard feelings when the moderator chose the "wrong" one and the user didn't notice?

It would still miss just as many as #4, but at least it would put the onus back on the users who made the mistake in the first place. The downside is that you would have to allow extra time for users to review posted want lists and send in corrections.

Quote:
4. Have TradeMaximizer simply pick the closest match for each bad name, and continue on its merry way.

To me, 4 is a non-starter. I guarantee that it would occasionally pick the wrong substitution. In your example, maybe the person who said 137-PGRID didn't really mean 0137-PGRID but instead meant 1437-PGRID. I agree it's much more likely that they 0137-PGRID, but even a 90% probability is going to be wrong 10% of the time.


I think it's much closer to 1% than 10%. The examples I've seen that are critical of the idea are all reliant on a game showing up more than once with just the right pair of numbers. That's not going to happen very often.

That being said, your point (along with those made by Jeff and others) is still valid. Getting it wrong just once could cause a big headache, a lot of hard feelings and probably isn't worth it. Kind of like some other things in our society which I won't specifically point out and start an off-topic firestorm...

Quote:
3 lends some automated support to what happens now, where moderators fix errors and then publish the revised wantlists to be checked. But you know not everybody is actually going to look at the revised wantlists carefully enough. Ideally the tool would somehow highlight where changes were made, but even so, mistakes are going to slip through. At that point, the moderator can say, "You had the chance to check your list, and you didn't, so you need to just live with the problem." This is doable, but less than ideal.


If you're taking that approach, you might as well just do #4 and say "you should be more careful with your lists when you make them - we warned you this could happen" and then not have the additional delay in getting results out to the 95% of people who submit with no errors at all. I guess this just wasn't as good an idea as I had thought when I suggested it, but at least it's spurred some interesting debate.

Quote:
1 may be the best approach in principle. The problem there is that getting the interface right is HARD. Also, we need to find a way to get JeffyJeff out of the loop for everyday math trades, as I'm sure he would whole-heartedly agree.


FWIW, I would be willing to be part of a team that helped out with OLWLG. Yes, I know that I've said that I prefer not to use it, but I've always agreed that it's a great tool. I would hate to be in Jeff's position - always being asked to help - but if the job could be spread between 4-5 people...?

Just a thought...
 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Michael Leuchtenburg
United States
Cambridge
Massachusetts
flag msg tools
badge
Avatar
mbmbmbmbmb
cokasaki wrote:
I suppose the cheat sheet could hide the information from the summary showing that you are giving item X for item Y. If you are trading several items, then in some sense you care which items you're giving and which items you're receiving, but it doesn't really matter which is technically for which.


I think the "what was traded for what" data is very interesting. I regret how difficult it is to see with the current commonly used readable outputs, though not (yet) enough to write my own post-processor. Having this data be more available would be useful to people who want to estimate how much a given game is worth in the trading marketplace.

Not really something TradeMaximizer can do much about, though.
2 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Subhan Michael Tindall
United States
Portland
Oregon
flag msg tools
badge
Avatar
mbmbmbmbmb
Regai wrote:
subhan wrote:
I'd be interested in a command-line option to have it output all the solution sets it finds with running with multiple iterations, not just the most optimum one.


For what purpose?

Post-trade analysis. I like to play at home & see the 'what if' scenarios. Sometimes the chains have some significant differences. You can get the data by bracketing iteration counts anyway, & it should be a pretty simple tweak to add the option.
 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
フィル
Australia
Ashfield
NSW
flag msg tools
designer
badge
I've got an 808 and a 303 and a record collection like the ABC
Avatar
mbmbmbmbmb
My top two requests pertain to making things easier for non-technical traders:

1. Distribute a version compatible with OS X, i.e. with Java 1.5. I know we can recompile the source ourselves, but Mac users playing at home might not know how to do that.

2. More human-readable output would be good. Many first time maths traders can't read the output and get confused. Something along these lines might work:

Alice
--------
You traded your Civilisation (123-CIV) for a copy of Age of Steam (999-AOS).
Send your Civilisation (123-CIV) to Charlie.
You will receive Age of Steam (999-AOS) from Bob.
1 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Daniel Corban
Canada
Newmarket
Ontario
flag msg tools
badge
Avatar
mbmbmbmbmb
While I am sure Apple will release a current and stable Java version for MacOS sometime this year, it would be nice if TM ran out of the box.

If anyone knows an easy workaround (other than installing the 1.6 preview), I am interested.
 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Stephen Waits
United States
San Diego
California
flag msg tools
mbmbmbmbmb
First, Chris, thanks for finding and publishing the algorithm. You're a true computer scientist, and hacks like me are in awe of your CS-fu.

Anyway...

I've noticed most of the discussion so far is about interface, or Mac compatibility. But I wonder if we've really thought about the big picture enough? That's a topic that I personally find much more interesting.

For example, it seems that MT moderators have mostly settled on using Linear priorities. But, why? Does it just "feel" right? Have we really figured out how to best minimize total "regret"? ("regret" is a poor word choice, but you get my drift)

Any other "big picture" thoughts? Should we take a step back and take another look? Or, is the algorithm really finished at this point?
 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Chris Okasaki
United States
White Plains
New York
flag msg tools
badge
Avatar
mbmbmbmbmb
swaits wrote:
For example, it seems that MT moderators have mostly settled on using Linear priorities. But, why? Does it just "feel" right? Have we really figured out how to best minimize total "regret"? ("regret" is a poor word choice, but you get my drift)


You're right. Virtually all prioritized trades have used linear priorities. I ran one local math trade using SCALED-PRIORITIES, and I don't know of any real trades that have used TRIANGLE-PRIORITIES or SQUARED-PRIORITIES. And hardly anyone ever uses the semi-colon marker to fine tune their priorities. Which all makes me wonder if maybe I should just simplify the interface by taking some of these out.

I think there are several reasons for the preference for linear priorities. First, it's the easiest to understand, and I think we've all been very concious of trying not to confuse math trade newbies too much (which on the whole is a good thing). Second, there is a tendency among moderators to just copy the rules from a previous math trade, so once one person used linear priorities, that started to get copied.

A third reason is that priorities don't actually make as much difference as it might seem. I ran some experiments with a big trade early on, and it turns out that most people who get anything at all get one of the top few items that it is at all possible for them to get. In other words, if you prune every wantlist to remove those wants that do not appear in any maximal set of trades, then most people get one of the first few of their items that are left.

Another pair of options that have been used a lot less than I thought they would be is SEED and ITERATIONS. (I think they've been used once.) Iterations make the cycles smaller just in case there's some major problem that requires canceling a trade loop. Maybe it's just that we haven't had any problems that serious recently.

Quote:
Any other "big picture" thoughts? Should we take a step back and take another look? Or, is the algorithm really finished at this point?


I think we're pretty much at a local...um..."maximum" Meaning I don't think we're going to get major fundamental improvements from minor tweaks. Whereas we very well may be able to get major improvements to the experience of users and/or moderators by tweaking the interface.

But if we're willing to consider major changes, then of course there's a lot of things that could be tried. For example, there have been discussions before about the possibility of trying to maximize "happiness" first, rather than trying to maximize the number of trades first, with happiness as a secondary criteria. Huber-style "tiered" trades fall into this category. I also ran an "economic" trade a long time ago. I'd love to talk to a real economist sometime about ways to improve that.
 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
James Perry
United States
Spring Hill
Tennessee
flag msg tools
badge
Ex-Term-In-Ate
Avatar
mbmbmbmbmb
Please do not remove the semi-colon based big step. The code is already there and it works. I have an expirment in the works that relies on that feature. (Plus I'm a firm believer in not reducing the feature set of an application.)


I would like to see tiered trades implimented. I'm not sure how often they would be used but it is something I'd like to see explored.


I'd like to see arbitrary priorities as well. Though those would definately be prone to user error and thus would require an interface for want list creation.

We could look at possible optimization of the app, but I didn't see any places for obvious enhancements when I looked at the code.

 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Michael Leuchtenburg
United States
Cambridge
Massachusetts
flag msg tools
badge
Avatar
mbmbmbmbmb
I don't use the big steps because the OWLG doesn't support them. Otherwise, I might.
 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Jack K
United States
Shorewood
Wisconsin
flag msg tools
Avatar
mbmbmbmbmb
cokasaki wrote:
I think there are several reasons for the preference for linear priorities. First, it's the easiest to understand, and I think we've all been very concious of trying not to confuse math trade newbies too much (which on the whole is a good thing).


Never underestimate this. I recently ran a local event in which about half of the people had never done this before and most of the rest had only done one other small one that I had run in the past. Just a few of us had participated in more events. About half of our live items traded and everyone seemed to really enjoy the experience, many of the first-timers finding it a lot simpler than they had feared. Ironically, the only mistake that anyone reported making was one by me! If this had been harder to understand, then I doubt there would be so many of them looking forward to the next time we do this.

That being said, I would also advocate not removing features that work. While I would not use many of them for an event I run, I think that's a choice best left up to the moderator of each event.
 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Jeff Michaud
United States
Longwood
Florida
flag msg tools
On-Line Want List Generator - Hopefully Making Math Trades a Little Bit Easier
badge
Captain Kirk, Captain Picard, Captain Sisko, Captain Janeway, Captain Archer
Avatar
mbmbmbmbmb
dyfrgi wrote:
I don't use the big steps because the OWLG doesn't support them. Otherwise, I might.

Just out of curiosity, in which case(s) would you actually want to do that? To me it seems lowering your priorities on any of your wants is just increasing the chance someone else will trade instead of yourself.
dyfrgi wrote:
Another pair of options that have been used a lot less than I thought they would be is SEED and ITERATIONS. (I think they've been used once.) Iterations make the cycles smaller just in case there's some major problem that requires canceling a trade loop. Maybe it's just that we haven't had any problems that serious recently.

I guess I didn't/don't understand what ITERATIONS does... if it really helps reduce chain sizes then I'd use it... but does it really do much of anything when you are using priorities? In fact aren't they mutually exclusive options? In the wiki page it says in the section for priorities it says it the result set with the "minimum total cost" (aka lowest "sum of sqaures"), while for the ITERATIONS option it says "How many times to run the main algorithm, keeping the solution with the smallest sum-of-squares metric".

I also assume SEED only applies when the ITERATIONS option is used?
 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
1 , 2 , 3 , 4 , 5  Next »   | 
Front Page | Welcome | Contact | Privacy Policy | Terms of Service | Advertise | Support BGG | Feeds RSS
Geekdo, BoardGameGeek, the Geekdo logo, and the BoardGameGeek logo are trademarks of BoardGameGeek, LLC.