# User talk:OrenBochman

## Re : CipherBot

Hi. Thanks for your interest in CipherBot. Yes, I use Pywikipediabot for programming some functions, and I also prepared a few templates using AWB to run some other minor maintenance tasks. Unfortunately, I am not publishing the code right now as it is still underdevelopment, and it is not yet optimized in a good shape. What does the new pages review bot is simply to go over the list of all new pages and check their texts. It checks again some criteria (five by now) and defines problems with each article, next it adds maintenance template in the article page, notify the user who created it about these problems, and make a log entry based on its topic, and problems. The community liked the idea as we spend a few months debugging and testing it, however I often receive messages from new users asking me to stop the bot from "ruining their articles". By the way, I like your Simone proposal, and I might borrow some ideas for developing my code and testing it soon. Let`s keep in touch. --Ciphers (talk) 05:02, 4 April 2012 (UTC)

## Coordination costs

I think that you've got a challenging problem to define coordination costs. I've left some notes in reply to you on my the Talk page of my research project at meta. Regarding the User:OrenBochman#Coordination parts, I figured it would be better to reply here.

First, even though I'm a little familiar with game theory, I hadn't heard the phrase "stag hunt game" before. So I was interested to learn about that. Based on this: "The stag hunt differs from the Prisoner's Dilemma in that there are two Nash equilibria: when both players cooperate and both players defect.", and since I think the prisoner's dilemma is very well known, I think that a more effective way to introduce it would be "a variant on the Prisoner's dilemma called the "stag hunt", which has two w:Nash equilibria. One occurs when both players cooperate; the other occurs when both players defect." Then I would go on to explain what the payoff matrix means; some people won't have seen these before, so a symmetric example (which is simpler) might be easier for people to digest first. (I assume in this that you're trying, in part, to explain -- if you're just trying to work out your theories in public, these didactic suggestions may not matter!)

When you say "this does not look like it is modeling the coordination challenge."--do you mean, you don't think that this is a good model? I agree.

Thakns for reminding me about this mistake - the w:stag Hunt is indeed a model of coopration not of coordination - it was part of the previous section. I've revised and expanded comunication and coordination material and moved it to the begining since it is universal issue in other wiki games. There is now an better definition of communication.
Coalitions are a more advanced area in Game theory and I am still reviewing it - but I'll expand this material since it does look like a more natural model in larger conflict. The N-agent cooperation game shows a simple coalition forming

The distinction between MicroCosmic Coordination and Political Coordination is interesting. First I'd point out that the typical AfD is 7 days (though these can be extended for multiple 7-day periods, and some discussions close earlier, particularly under CSD, snow keeps, etc.). Do you think that establishing coalitions is necessary/useful?

I am pretty sure that:
• Coalitions either form or they don't. (Tautology)
When they don't it seems to form then something is happening -- it could be a clear coordiantion failure or it something more complicated but I don't have enough data or research on this to say what it is.
• Coalitions simplify debates and simplify group decisions (consensus is one of the hardest to reach - with the exception of unanimous decision).
• I am leaning to the opinion that they are necessary to block consensus.
• I think they are very useful to their members - especially in political arguments which are less about facts and more about opinions or representation of facts.
• Looking at some of research on massive debates on Wikipedia would be illuminating to further define these concepts. As would more field work and experiments.

What you say here is *particularly* interesting. Examples would be helpful: Afd discussions - seem to have 2+ modes of closing. Fast (closely related and very similar to CSD) - do not require MicroCosmic Coordination Slow with a good likely hood of repeated coordination failures - MicroCosmic failures oft lead to relisting (extending the deadline), while political failures leads to no consensus at closure, multiple Afd of the same article over long timeframes.

• The size of the coordination is key indicating that two people's cost is an order of magnitude more difficult than one person and that a new coalition is an order of magnitude harder still. (a theory in need of testing) but would be explained by a .1 probability of scarcity in a coordination game and a two round model for coalitions.

Is there any chance that you'll be at WikiSym? The call for posters and demos is open until June 8th, and I think this could be really relevant.

I am really exited with this research and would like to pursue it further and possibly publish it. WikiSym sounds interesting option. I'm going to the Berlin Hackathon and Wikimania2012. I have had minimal support from WMF so far, though WM.HU has generously helped out with WikiMania. However I am based in Budapest so WikiSym is in driving range - so it is an option. However I'd need some help drafting a submission and getting it accepted.

I've considered making a poster or a demo - If you can helping me with this I'd be very greatful! It's also interesting to look at the points that require more looking into in your list.

I think that the kind of coordination cost that needs to be avoided is when a discussion gets reopened without new information. In these cases people articulate the reopening as a problem: "The original AFD was closed earlier this month, and absolutely nothing has changed." (e.g. w:Wikipedia:Articles_for_deletion/The_Best_of_The_Velvet_Underground:_The_Millennium_Collection_(2nd_nomination). Yet reopening is not always a bad thing, when the article has changed significantly or when there were insufficient comments the first time around.

Who can't envision a more prefect process but would it really be better all around??? (work, results, openness etc). e.g. Some kind of formally structured argument summary with votes. (you must add something for your vote to count) This would be inherited at a future discussion. But this might:
• contribute to the learning curve which would be bad.
• reduce the scope of required coordination which would be good.

I think under the cap of a game theoretic social scientist I can see how both approaches balance out in practice (I think there is even a result that says that in Evolutionary game theory - a mixed strategy is under certain conditions is the only stable ESS). This is how Wiki Game Theory resolves many paradoxes.

Note: There is a parallel attitude in Open source development that a duplication of work required due to imperfect coordination - is not always a bad thing.