Slashdot Mirror
Cooperation Works if Majority Can Punish Freeloaders
plasmid writes: "Some Swiss economists ran an investment game... they found that if the majority could punish freeloaders, cooperation flourished. I think this has implications for cooperative peer-to-peer systems and, to a lesser extent, for open source development. I'm so inspired I plan to go out an punish someone right now, as a matter of fact." I had just read this article the other day (go memepool), so this Nature piece seems oddly apropos.
A similar set of ideals has been previously applied in psychological and darwinian non-zero sum games where there is a reduced personal gain but higher group gain from cooperation. These games challenge participants in finding an optimal outcome for both in cases where there are multiple iterations of choices to cooperate or 'defect' from cooperation - the website details only a new variant of these.
One model is that of the cold war. If both countries cooperate in an arms reduction treaty, they both win some, but for the individual country, a win can be made if their competitor cooperates and they 'defect' and build more arsenal.
This game has a matrix of possible points scored by each side depending on their individual choices.
. . . . . coop . . defectcoop . . . 3,3 . . 5,0
defect . . 0,5 . . 1,1
In the above situation, the two scores delimited by commas indicate the score for each country. If the countries both cooperate, each receives three points. However, if they disagree, one country will win, but the sum score is less. The interesting situation is if both defect - the value placed on these scores may also determine how the game is played through multiple iterations by two players.
Another variant is the prisoner's dilemma game. Two criminals are captured, and the DA will cut one of them a deal if they squeal on the other. Of course, if both squeal on each other, both loose big. If both are quiet, they will get a lesser charge. The dilemma is that the best group outcome is that they will both fare better if they are both quiet, but they don't know what the other will do.
The article listed is similar to this, but different that there is a cost involved in punishing the 'bad' player that doesn't pay into the investment pot. Here the game asks you to punish the uncooperative player with costs now, but the punishment might make them more likely to contribute in future rounds of the game. Interesting.
If the other sources somehow become slashdotted, NewScientist also has an article up on this.
It's up under the title "Anger plays key role in human cooperation".
--R
http://www-personal.umich.edu/~axe/research/Evolvi ng.pdf
I'm not an academic but I've become really interested in Complex Adaptive Systems research recently (I was interested in this before I knew what it was but that's another story). One of the books I came accross was "The Complexity of Cooperation" by Robert Axelrod. In it he discusses much of the research that led them to Tit-for-Tat and many other strategies for the Iterated Prisoners Delima. Very good read, check it out.
(including the words "Five words")
./configure --with-vxworks; make ) were not compelled to release their changes by the XFree license. Result - a less than stellar server, that locks me into a buggy and feature-lacking OS (Don't say it - as soon as I have the manpower my project will be converted to Linux.)
I didn't see anybody at +3 making the analogy to the GPL vs. the BSD-like licenses.
In a sense, the GPL "punishes" freeloaders by denying them resources - "If you don't share with us, then we won't let you have a share of the pot." If you won't contribute to the shared codebase, you cannot take from the shared codebase.
Compare and contrast that to the BSD-like licenses that don't have the "Release the source" requirement - a freeloader (certainly Microsoft, possibly Transgaming, possibly Lindows) can take from the public pool, not give back, and incur no "punishment".
I used to think that RMS was a crazy, extremist bastard. Then something happened to cause me to revisit that thinking. I work professionally with a product called RtX, which is an X Windowing System server for the embedded operating system VxWorks. RtX is derived from XFree86. I've had several problems with RtX - it won't recognize certain graphics chips, it doesn't support font server use, it won't do anything but 256 pseudocolor, I cannot easily add key bindings or LEDs to the keyboard routines, and (most importantly) it won't work under the newer versions of VxWorks. None of these would be insurmountable problems if I had the source, but the folks that did the conversion of XFree into RtX (and it isn't a trivial conversion, not just
I know I just enraged the "GPL is tyranny, BSD is freedom" crowd. But please, think about this for a moment. If you wish to continue to use the BSD license for your code, wonderful. However, any code I do off-hours will be released under the GPL, for the reasons stated above.
www.eFax.com are spammers
I object to that article, and to the next reply.
There is another method for ensuring cooperation and fair behavior in peer networks. And it works the same was as the method described.
It is called social discovery, and it works by having each and every peer create a view of the network that suits their interests and needs. In such an environment, the freeloading peers will not be viewed as valuable peers and will be dropped from your peer group(s); no longer used, and no longer using your resources.
On the flip side, there is a strong incentive to become a better, more reliable peer yourself, as the quality of peers you can associate with is directly related to how they perceive *your* quality to them.
If you want to be able to tap better, higher quality peers, then you should keep your node available longer, more often, and also share more resources (whatever they may be).
The project I am working on that implements this social discovery mechanism is called the ALPINE Network and there is also another social discovery based project called NeuroGrid.
I am biased towards this kind of approach, but I think it provides the best long term solution to resource discovery / searching in large peer networks.
There's a long article (which unfortunately is not available online) discussing experiments like these and coming to the conclusion that the vast majority of people value "fairness" over material success in this particular case. I was particularly fascinated by the experiment known as the Ultimatium Game; the article says that only 4% of people, IIRC, choose what the mathematically most beneficial solution. (In other words, in 96% of cases people would choose the "fair" outcome over one that was objectively better for BOTH participants.) Worth checking out if you are interested in this kind of thing.
"Biped! Good cranial development. Evidently considerable human ancestry."
Art. 13 Right to Privacy
e ga ff/swilaw/fconst.html
1 All persons have the right to receive respect for their private and family life, home, and secrecy
of the mails and telecommunications.
2 All persons have the right to be protected against the abuse of personal data.
http://www.eda.admin.ch/washington_emb/e/home/l
-s
It's not available online, but the January 2002 issue of Scientific American has a very relevant article titled "The Economics of Fair Play". It discusses the nonrational dynamics of how groups of human expect and enforce fairness. Definitely worth a read for open-source economic theorists and fans of intriguing behavioral-psych experiments.
When all you have is a hammer, everything looks like a skull.
If you've got two machines, one running FreeBSD or some other OS that has support for Sparse files in the file system, you can fool DC into thinking you've got 60 gigs worth of data on a 100 meg share.
Share a directory with Samba to the windows machine you're running DC on.
Create a bunch of sparse files in that directory with reasonable sounding names (Like Matrix01.mpg or something). The following program will create sparse files:
#include <sys/types.h>
#include <sys/stat.h>
#include <fcntl.h>
#include <unistd.h>
#include <stdio.h>
int main( int argc, char **argv )
{
int fd;
if( argc < 2 ) {
printf( "Usage: %s filename\n", argv[0] );
exit( 1 );
}
fd = open( argv[1], O_CREAT | O_WRONLY );
if( fd < 0 ) {
printf( "Couldn't open file.\n" );
exit( 1 );
}
lseek( fd, 1073741824, SEEK_SET );
write( fd, "x", 1 );
close( fd );
return 0;
}
Share that directory with DC and enjoy.
This
eDonkey uses a "punishment" method in it's peer to peer networking scheme. Basically, until you're sharing at least 10k/s, you're limited to 4 times your upload speed for downloads... Yes, that means if you're sharing at 0k/s, you can receive at 0k/s. Also, it shares partial files, so, for what it's worth, you are almost always sharing at least a part of a file...
It works really well in small groups (reference DAPCentral ), and from what I can tell, it really makes interpersonal cooperation a lot easier than, say, Morpheus...
Of course, it'll never be as popular because it's not a single central server... That and it's got a linux interface, and we all know that anything that gets on linux dies the next week (as a hax0r tool)...
Hasta luego,
/Ex