"Attacked From Within" @ Kuro5hin, Anaesthetica Rules Da Web Options Rating

[EricBarbour]

Attacked from Within

Quote:

This article attempts to fundamentally rethink what constitutes community and society on the web, and what possibilities exist for their maintenance and reconstruction in the face of scale and malicious users. The recommendations reached, after analyzing the weaknesses of the web forums we all know and love, are:

* User anonymity should be forced.
* Barriers to participation should be as low as possible.
* Moderation should not focus on users or on comments in isolation, but on the relational quality of comments.
* Passive moderation filters can mitigate problems of scale.
* Preservation of community must shift from being based on exclusion to being based on demonstrated constructive interaction.
* Forums should discriminate between content types: original content, links, and personal content.
* Story promotion and front page position should be driven by conversation, not voting.

Discuss.

(Sounds like a formula for Digg or Fark to me. Some of these ideas don't seem practical.)

This post has been edited by EricBarbour: Thu 9th April 2009, 3:50am

[Somey]

It's a bit lengthy, but it looks like the author is trying to suggest that completely eliminating usernames, forcing everyone to become totally anonymous, would be preferable to existing models of online community interaction - Usenet, Wikipedia, Digg, etc., in fact practically every Web 2.0 site in existence.

What's more, I believe the "Shii" being referred to (and being agreed with) may actually be our very own WR member Ashibaka, aka Shii (T-C-L-K-R-D) , a long-time Wikipedia Admin who rarely visits us these days.

If you got rid of all usernames, though, wouldn't people just sign their posts with names (fake or otherwise) anyway? And ultimately create their own reputation system on an ad hoc basis? I guess you could make rules against that sort of thing, but then you're just trading one moderation headache for another...?

[EricBarbour]

Sorry about the long post....needed to make a few points.

If you got rid of all usernames, though, wouldn't people just sign their posts with names (fake or otherwise) anyway? And ultimately create their own reputation system on an ad hoc basis? I guess you could make rules against that sort of thing, but then you're just trading one moderation headache for another...?

That's the main problem. If you eliminate vanity, that is a major motivation for many people to participate. You can't reduce people to simple mechanisms and expect them to work very hard to create something, if there's no reward at all.

And how do you make

Moderation that better reflects quality, as opposed to simple agreement.
Moderation that lightens the load on admins.
Reduce the ability of users to game the system.

work in an highly- automated system? How do you establish parameters for "quality", given the primitive state of artificial intelligence available today? Do you let humans do the filtering? Then you'll just end up with another cabal....I've suggested a system that's wide open, but only allows the original creator of an article to give other editors access to it. Yet, that could be abused easily, too.

The solution that xkcd's Randall Munroe hit upon after reviewing the standard options faced by all rapidly scaling communities—restricted entry, moderators, user moderation, and sub-communities—was a system of passive moderation. Moderation would be automatically applied according to a predetermined set of criteria specifying what qualities a good comment would have. In Munroe's case, originality was the key, and any commenters attempting to say something that had already been said before would be penalized by increasing mute times. A similar project, the StupidFilter, being developed by one of our own, uses Bayesian logic to identify stupid comments based on a seed group of human-identified stupid comments. The criteria for stupidity include: over- or under-capitalization, too many text message abbreviations, excessive use of 'LOL' or exclamation points, and so on. Spam identification systems for email and blog comments (e.g. Akismet for WordPress) do much the same thing, identifying commonalities in junk messages and containing them in a junk/spam purgatory awaiting moderation.

Yeah, right. Soooo easy. My ISP has Postini spam filtering, and even though it works really well, about 10-20 blatant spams make it into my inbox every day.

This article keeps talking about group blogs like Slashdot and Digg. If your intention is to create a knowledge database and not a bunch of opinion snark, how can you make group-bloggy setups work without producing gibberish? It's great to talk and talk about filtering out "retarded" content, but then you're cutting off a giant chunk of your userbase. The writer of this article is far too optimistic about a "reasoned" solution. How do you reason with millions of self-selected people, who are all over the map in personality, interest, emotional stability, etc?

I like this:

Don't make the user learn a markup language to format their post correctly—Google's rich text composed in Gmail is a good example of avoiding the problem of making a user learn HTML, BBCode, or Wiki markup. Don't prevent the user from posting by making them jump through hoops such as the impossible-to-satisfy Slashdot "Lameness Filter" or the mute-banning Robot 9000 from xkcd's IRC channel and 4chan's /r9k/ board. Obvious and strict wordfilters encourage users to game the system rather than work to write better posts. The result is a commenting system that favors those that spend the time to master technical details over those who write useful contributions without knowing the intricacies of the site's parochial commenting system.

That IMO is a major problem with the wiki system. If you want nicely-formatted, consistent and well-referenced articles, given the primitive editor in Wikimedia, you have to train people somehow. They need a wizard or autoformatting system, to take a layer of bullshit off the editor's back and make the job easier. AND to remove system gaming--a major reason why Wikipedia grew so quickly in the first place. It generates obsessive behaviour, not good written material. The Wikimedia developers have a lot of (unintentional) crimes to answer for.

Also noticed another familiar WP name in this text:

For example, new users might be asked to find sponsors among existing users in order to preserve some level of trust and social links. Ta bu shi da yu (T-C-L-K-R-D) detailed the perverse incentive structure that sponsored users would create (at the time, Shirky called Ta Bu's opinion "hysterical"). Rusty, k5's deadbeat dad, eventually agreed with Ta Bu, admitting that sponsorship "was a stupid idea."

[GlassBeadGame]

Sorry about the long post....needed to make a few points.

Don't worry about the length. This is a wonderful post.

[A User]

Attacked from Within

Quote:

This article attempts to fundamentally rethink what constitutes community and society on the web, and what possibilities exist for their maintenance and reconstruction in the face of scale and malicious users. The recommendations reached, after analyzing the weaknesses of the web forums we all know and love, are:

* User anonymity should be forced.
* Barriers to participation should be as low as possible.
* Moderation should not focus on users or on comments in isolation, but on the relational quality of comments.
* Passive moderation filters can mitigate problems of scale.
* Preservation of community must shift from being based on exclusion to being based on demonstrated constructive interaction.
* Forums should discriminate between content types: original content, links, and personal content.
* Story promotion and front page position should be driven by conversation, not voting.

Discuss.

(Sounds like a formula for Digg or Fark to me. Some of these ideas don't seem practical.)

Sounds like a formula for anarchy. New users won't be attracted if you treat them all as numbers, and not as names/nicks.

[GlassBeadGame]

Attacked from Within

Quote:

This article attempts to fundamentally rethink what constitutes community and society on the web, and what possibilities exist for their maintenance and reconstruction in the face of scale and malicious users. The recommendations reached, after analyzing the weaknesses of the web forums we all know and love, are:

* User anonymity should be forced.
* Barriers to participation should be as low as possible.
* Moderation should not focus on users or on comments in isolation, but on the relational quality of comments.
* Passive moderation filters can mitigate problems of scale.
* Preservation of community must shift from being based on exclusion to being based on demonstrated constructive interaction.
* Forums should discriminate between content types: original content, links, and personal content.
* Story promotion and front page position should be driven by conversation, not voting.

Discuss.

(Sounds like a formula for Digg or Fark to me. Some of these ideas don't seem practical.)

Sounds like a formula for anarchy. New users won't be attracted if you treat them all as numbers, and not as names/nicks.

User anonymity should be forced.

I would understand this to mean not even permitting contributions over time to identified by any means, not even an impersonal but potentially stable identifier such as an IP address and certainly not by an account name, let alone an actual identity.

The proposal seems motivated by the belief that these steps would force all content to be evaluated on the basis of merit. But we are talking about "merit" atomized on a level that is hard to comprehend. It assumes some sort of self organizing force that will sort out endless disjointed, disconnect and utterly "unowned" contributions into high quality content. I don't see this as present and agree that this would result babel. It would attract new users but they would be the most unaccountable and irresponsible types.

[Jon Awbrey]

Attacked from Within

This article attempts to fundamentally rethink what constitutes community and society on the web, and what possibilities exist for their maintenance and reconstruction in the face of scale and malicious users. The recommendations reached, after analyzing the weaknesses of the web forums we all know and love, are:

1. User anonymity should be forced.
2. Barriers to participation should be as low as possible.
3. …

Two observations just off the cuff:

1. I am puzzled why Anaesthetica did not post as Anonymous Cowad. Perhaps someone should forcibly delete his/her byline?
2. If you wonder what the "5" in "Kuro5hin" is all about, it will cost you a fin to ask 'em.

Ja Ja

[Milton Roe]

The proposal seems motivated by the belief that these steps would force all content to be evaluated on the basis of merit. But we are talking about "merit" atomized on a level that is hard to comprehend. It assumes some sort of self organizing force that will sort out endless disjointed, disconnect and utterly "unowned" contributions into high quality content. I don't see this as present and agree that this would result babel. It would attract new users but they would be the most unaccountable and irresponsible types.

The question is what are we aiming for? A new merit system starting from ground zero? Or merit system which is forever erased and re-started, even day to day, like those people with the hippocampal memory problems ala 50 First Dates (a pretty good film, even if it did have Adam Sandler).

Because we can screw it up even more than what's above: not only no ID, but the computer re-randomizes everybody's names between one log-on and the next, so all you see is the content of the moment, COMPLETELY free of a-priori expectations. Then you really find out how much that Baysian pre-expectation reputation makes you listen more sympathetically to one person vs. another.

There have been lots of experiments with this-- anonymous and label-stripping peer reviewing of science papers, etc. It turns out to be very difficult, because people use all kinds of tricks to try to get a handle on how much somebody else knows. Do they use the right language, talk the talk? Use the right "technical shibboleths"? The correct notation? Think of Eliza Doolittle and her Cockney accent.

Some systems have tried to do at least a round of that kind of stuff, since somebody or other discovered (in what's called the "Delphi System for the utilization of many experts") that you have to sometimes try to eliminate the background repetitious noise caused by the fact that whenever you get a lot of experts together, people seem to waste a lot of time repeating themselves and their opinions again and again, until everybody else acknowledges them. So you take the labels off, so they have to reason to do that, since they have no "rep" to protect or push. And also, there is no "advantage" to kissing up the right person, since nobody knows who anybody else is. Gen. George S. Patton once opined that heros in the planning room, who refused to let bad plans past them just because of who was pushing them, were as rare as heros on the actual battlefield. Maybe more rare.

Of course it's necessary to pre-select the experts before you even try this, else dire things happen in the way of wasted time.

[GlassBeadGame]

The question is what are we aiming for? A new merit system starting from ground zero? Or merit system which is forever erased and re-started, even day to day, like those people with the hippocampal memory problems ala 50 First Dates (a pretty good film, even if it did have Adam Sandler).

Good imagery. Here's one you might enjoy: Combining the worst elements of the Great Cultural Revolution and Pol Pot's Day Zero. The continuous re-boot.

[Jon Awbrey]

¤ Sigh ¤

What Millennium Is It?

I must be the victim of a time travel experiment gone mad.

We are seeing the reductio ad absurdum of a pre-literate e-pistemology.

Herme, Herme, Did You See Where I Put My Neutics?

Ja Ja