Wednesday, November 28, 2007

People I Know

There is a very sad film about the "high spheres" life in New York. It is about the politics, the games people play, the artificial and the indifference. It is very sad - but so true. It is so sad because of the contrast between our imaginations based on the artificial glitter and the exalted words and the true life. And it is important because it reminds us how much of our social communication is just conventions and games.

We should remember this when designing social applications and not call the people one know as his friends.

Thursday, November 22, 2007

Don't be evil Google!

Google becomes more and more scary - add to that the Android and all the other initiatives - and the picture becomes quite paranoid. It's not a new thought that power corrupts (and absolute power corrupts absolutely). If Google does not want to make farsa out oftheir 'Don't be evil' motto they need to change. Drop the secrecy! Engage into the public discourse!


Don't love algorithms more than people!

Tuesday, November 13, 2007

Scale Free, Zipf Law and Christopher Alexander living structures

OK - so this is just a brain dump. Some analysis from Software Libraries and Their Reuse: Entropy, Kolmogorov Complexity, and Zipf’s Law suggests that the reuse of software components follow the Zipf law (that is the nth most popular component is reused in c*n^(-1) other places). My intuition is that this is not just a law about software - but about just any system - the economy in the program length would have a direct analogy in the economy of physical system parts. And I would guess that Christopher Alexander patterns and structure preserving transformations are result of the same entropy maximizing principles.

Tuesday, November 06, 2007

Myspace or Facebook? I put my bet on Couchsurfing.

Why? This is the social networking site that creates real trust. It is a bit specialized now - but this will change.

Friday, November 02, 2007

Spam in comments - reposting

I was spammed and I had to delete the whole post (as there is no way to delete just the spam comment), I repost it here with all the comments. Now comments are moderated.

Online Debate

When there are more than a few participants or the subject is highly political the free form face to face conversation is normally replaced by a more structured form of communication like a debate. The rules of this more structured form can be implicite, coming from common culture, or explicite like 'The Roberts Rules of Order', but there needs to be some rules to govern the discussion that otherwise would turn to chaos. In current online settings, and especially in asynchronous communication like email lists all the immediate cultural feedback - like seeing the angry faces of other participants, that would in face to face situations push the participants to obey the rules, is lost. This leads to the question - how can we compensate for that loss? My first answer was by setting explicite rules coded in the list server algorithm, for example restricting the amount of emails to one per day, but perhaps there are other possibilities?

Update: I need to add that after a face to face meeting, even when it involved only a few participants from the mailing list, the list much become much more productive.

At 8:53 AM, Kaunda said...

Various online discussion forums seem to have different implicit understandings about the the tenor of conversations there.

Perhaps not directly relevant to your question posed in this post, but consider the differences between Usenet and Chat. In Usenet "Prove it!" and a very rough and tumble discourse is the norm. Chat developed in the opposite direction: how many stories of love on the Internet. The extremes of both forms of online communication are in part a result of lacking immediate feedback of face to face interactions. But these features are also reinforced because they're "fun" for lack of a better word.

Blogs which allow comments are different from Usenet, Chat, mailing lists, and other online forums; all forums develop certain implicit norms even without explicit rules. Implicit norms, the tone of the setting can be influenced informally.

Some bloggers have explicit rules for comments. Here's the law Will Bunch lays down:

PLEASE COMMENT WITH PASSION...
...but not with racial slurs, potentially libelous allegations, obscenities or other juvenile noise. Such comments will, at our discretion, be deleted in their entirety, and repeat offenders will be blocked from commenting. ALSO: Any commenter advocating killing any government official will be immediately banned. Thanks.

Recently in the blogosphere Juan Cole reacted to a piece by Christopher Hitchens at Slate. At issue was that Hitchens had excerpted--out of context Cole avers--quotations from work Cole posted in a private mailing list. That list serve had the rule:
"It has a strict rule that messages appearing there will not be forwarded off the list."

Your algorithm to restrict to one email a day is clever. Certainly there may be other "technical" fixes to the problem of feedback present in face time not present in Internet discussion. But technical fixes seem less important to me than agreements between participants.

It seems to me that in online discussion forums a space for discussing the rules overtime will be more effective in promoting useful debate.

Kids around the age of 10 learn to play differently. When they play "baseball" at that age it's as much about arguing about the rules of the game as playing ball. There's a similar phase in the development of various online forums. The "rules" are determined both explicitly and implicitly.

At 9:58 AM, zby said...

Thanks for the comment. The subject matter is really complex and there are many solutions. What I wanted to propose is to develope some 'technical' features aimed directly at improving some concrete problem with a communication forum. Perhaps the 'One email per day' rule is too strict - perhaps what would be enough is for example a visible counter (in the email subject?) of the number of emails a participant send in last 24 hours? I do agree that the legitimacy of all of those changes can be reached only via a democratic process - and since our organisation is democratic there is no question that we would introduce them using democratic methods.

At 1:57 PM, phil jones said...

Well, my proposed "solution" was a dead loss :-)

http://www.nooranch.com/synaesmedia/wiki/wiki.cgi?TypedThreadedDiscussion

I still think that it's worth experimenting with the idea of requiring (or encouraging) contributers to a discussion to explicitly add some kind of tagging or type metadata to their posts. And then allowing the reader to sort and filter according to tag. But the technical and UI details need some rethinking.

Aside : it *is* a very remarkable thing that in an age of a dozen different Ajaxian / web 2.0 to-do lists etc there still doesn't seem to be the web 2.0 version of Slashcode or similar threaded discussion system.

At 3:15 AM, zby said...

Hi Phil, recently I am thinking that there are two kinds of discussion - one that I would characterise as the exchange of ideas and the other which is about deciding about some collective action. When there are only few participants those two can work without any additional rules - but when 'scaling' those two need different techniques. What you propose is more about the 'filtering of ideas' side what I was talking about in 'One email per day' was more about the political deliberation.

At 4:42 PM, phil jones said...

good point zby.

Yep. TTD is definitely to support argument for generating knowledge rather than as a decision-making process.

I wonder what the differences you need to support are. Is it that you need to formally make explicit what everyone's current preference is (a central count of votes for each proposition).

Or are you just worried about taking the heat out of disagreement?

What about combining this with the "thinking hats" thing? For example what about a discussion board that allowed each person to post one message of each hat-colour per hour?

At 5:12 AM, zby said...

I think the first goal would be to concentrate on the things that are common for the participants instead of discussing over and over the differences and trying to convice everyone to one view.

It really seems that there is some deeper subject in that. Currently it is not clear for me when the discussion starts to be political. I believe there are some hidden interest conflicts that start the vicious circle. For example if I am an vi user I would like more people to use vi so that the vi culture thrives, so that there are many new vi developments etc. And also there is the feeling of being right - I chose vi - so when you choose vi too it shows I am a smart person. I am not sure what is the interplay between those two ingredients.

Finally when we go to our personal relation - I know you are a Python programmer, I chose Perl - how come we don't try to convice each other to choose the right programming language?

At 1:33 AM, zby said...

Some related link: http://www.buzzmachine.com/index.php/2006/05/20/herding-cats-domesticating-rats/

At 4:29 PM, phil jones said...

Well. I'd argue that everything is political. You can't take politics out of human-life.

I don't really think it's a *problem* when politics gets in the way of decision-making, I think it's a *warning* that there are issues that haven't been addressed properly.

Your vi example illustrates this perfectly. What could be more stupid than a paralyzing argument between intelligent people about which editor is best?

But people argue.

Not, I say, because they're stupid, but exactly because of all those *externalities* that you mention. The appearance is that the decision between editors is arbitrary. But the reality is that, due to social *context*, it does matter.

So I'm a bit suspicious that we want social software which tries to "fix" the problem of political argument by excluding it. I think we need social software to help resolve the political arguments by helping people make more of their contextual assumptions *explicit*.

At 4:33 PM, phil jones said...

Oh yeah, and we don't have problems about Perl vs. Python because we trust that Parrot will solve our interop problems :-)