Algorithms Vs. Group Intelligence
I saw a Presentation/Q&A session with one of the Google founders who, when asked about his opinions about the (then) new trend of tagging responded with (and I paraphrase):
It struck me that this statement revealed a potentially growing philosophical divide between companies like Google who believe in machine based indexing, ranking and searching vs. the latest 'web 2.0' creations that seem to focus on explicit user/community interaction (and a belief in Group Intelligence) in order to achieve the same thing.
As far as I can tell, Google has resisted most forms of tagging in their applications while Web 2.0 apps continue to integrate it as one of their primary content classification methodologies.
While the question and answer was about tagging specifically, I think it applies more broadly.
For example, Google uses computer algorithms to determine which search results belong on the front page of a search and their news page uses yet more algorithms to work out what's 'front page news'. In fact, even some meme trackers have the same philosophy and use algorithms to discover and rank news and blogs – these include Tailrank, Techmeme and Technorati.
Digg and Wikipedia (as well as other similar sites), on the other hand, use direct human interaction on a mass scale. Group Intelligence.
However, I don't recall anyone ever saying 'I just got Tailranked'. Tailrank (and similar sites) do not seem to generate the level of traffic and interest as Digg has recently.
However, as most people know, Digg is in the middle of some controversy over their ranking systems which is causing many to wonder if ‘Group Intelligence’ (particularly when it comes to voting/popularity/value judgments) is actually just another name for Herd Mentality'?
With Touchstone, we have mainly made a bet on Algorithms in order to determine the 'Personal Relevancy' of an incoming item - but that's not to say that those algorithms can't take into account a broader set of factors including direct user rankings and feedback.
At Google we have always thought that computer algorithms should be responsible for indexing and classifying information for people rather than the other way around.
It struck me that this statement revealed a potentially growing philosophical divide between companies like Google who believe in machine based indexing, ranking and searching vs. the latest 'web 2.0' creations that seem to focus on explicit user/community interaction (and a belief in Group Intelligence) in order to achieve the same thing.
As far as I can tell, Google has resisted most forms of tagging in their applications while Web 2.0 apps continue to integrate it as one of their primary content classification methodologies.
While the question and answer was about tagging specifically, I think it applies more broadly.
For example, Google uses computer algorithms to determine which search results belong on the front page of a search and their news page uses yet more algorithms to work out what's 'front page news'. In fact, even some meme trackers have the same philosophy and use algorithms to discover and rank news and blogs – these include Tailrank, Techmeme and Technorati.
Digg and Wikipedia (as well as other similar sites), on the other hand, use direct human interaction on a mass scale. Group Intelligence.
However, I don't recall anyone ever saying 'I just got Tailranked'. Tailrank (and similar sites) do not seem to generate the level of traffic and interest as Digg has recently.
However, as most people know, Digg is in the middle of some controversy over their ranking systems which is causing many to wonder if ‘Group Intelligence’ (particularly when it comes to voting/popularity/value judgments) is actually just another name for Herd Mentality'?
With Touchstone, we have mainly made a bet on Algorithms in order to determine the 'Personal Relevancy' of an incoming item - but that's not to say that those algorithms can't take into account a broader set of factors including direct user rankings and feedback.
16 Comments:
I think the whole Digg situation requires us to take one step backwards.
The underlying priciples that Digg encapsulates are real examples of group inteligence, as are wikipedia's and many other pieces of collaborative software. The problems comes from when people don't play the game fairly. Digg asks you to think as an individual not as a member of a group, bloc actions are one reason the system fails.
This is a classic debate within political philosphy, particularly within deliberative democractic theory. To cut what could be a 6000 word essay short, do we have a truely democractic model (such as Diggs) when a bloc of people acting in unison can overrule the voice of the few. Everyone has heard of the saying 'tyrany of the majority' right?
I digress. I suppose my point is that Digg could be a beacon of Group Inteligence but the social force of 'herd mentality' as you call it is causing problems. The irony, with regard to Digg, is that their management have suggested that more intuitive algorithms could be a way to weed out these activities.
I was pleased to stumble across this site as group intelligence (or the lack of it!) has always fascinated me.
Unless and until there is a real "cost" involved in expressing an opinion then Web2.0 techniques are flawed.
Some of the more interesting new techniques of examining the web involve an ability to weight others views yourself - I suspect that this is the way forward....
This is very true.
Expressing an opinion in a political forum always had costs incurred with respect to time, money and knowledge.
However, arguably this system was to closed and one, particularly those less well off couldn't express their opinions easily enough.
The Internet, I suppose you are suggesting, is too open, people can express mindless opinions cheaply, in a very small amount of time with possibly no knowledge over their given subject.
Maybe we have gone from one pole to the other?
I think another point of interest is that of annominity, much like your comment. You could be Bill Clinton for all I know. In you case it is not too much of a problem beacuse your comments are not particularly controversial. If they were however you would not be held to account or forced into providing any justificantion for what you have said at all.
This is why I always use 'danielbower' as my moniker where possible (my full name) and always leave a link to my site (www.danielbower.com) as a means to continue any debate.
Google DOES use Group Intelligence. If you link to a page you give it a vote (similar to digg it for Google). This Group Intelligence made Google so popoular so you may consider Google being ahead of the Web 2.0 movement.
Group intelligence only functions if the group is diverse. Otherwise you have groupthink (aka herd mentality or mob mentality). The problem at Digg is that a small, influential group was allowed to override the opinions of the majority. It was a tyranny of the minority.
I agree with Tihy..
The Pagerank system (and hence, the overall rankings in search results) is affected mainly by the number of incoming links to that page. Since these links are created by people, we can safely conclude that Google uses Group Intelligence too. This was the property that allowed Google jump light years ahead of the competition back in the days..
Tihy and Can, I am glad you bring this up. So if incoming links can be counted as votes, then can we not conclude that systems like Digg that require explicit votes in order to determine value are actually a step backwards and just a fad rather than any sort of innovation?
But isn't the advantage to Digg is about "like mindedness"?
It's not that I'm searching for "organic toilet paper". I use Digg in order to browser for things that I wouldn't have otherwise known I even wanted to see, and I go to Digg because there are PEOPLE there who are LIKE me. I think that's a growth limitation for Digg, but it is the reason for example I have noticed I don't like Netscape. What people choose there doesn't interest me as much.
There is also the unconscious vale of thinking that this stuff is "vetted" somehow by the hordes, mobs and "group think" that came before me and pushed it to the top.
-- dave
You say :
As far as I can tell, Google has resisted most forms of tagging in their applications while Web 2.0 apps continue to integrate it as one of their primary content classification methodologies.
But google vidos use tags....
(although i havnt got round to reading rest of article yet)
As far as "group think" and "herd mentality" you are both right and wrong. The technology divide already thins the heard with regards to this synthetic selection you are discussing. By the sheer fact that you are reading your news, dispersing stories/papers/games/articles over the internet you have already selected a "minority" pool from which to draw water. Most folks dont even know about Digg and dont really even know how Google really works, or how to really use it. You have to take into account the intelligency quotient you are already dealing with as far as internet users.
Plus, by just a quick look over the last month of articles that make it to the Digg "headline", its easy to tell that most Diggnation netizens are liberal minded and leftward leaning. Gamers, techheads and the internetizens are usually highly intelligent, pretty macro-suspicious users. This isn't a criticism! But hell, the internet exists because of this "minority" that was fed up with the way things worked. Its wonderful, but you also have to be careful you don't mistake yourselves for the majority, because you are not. The majority doesnt even know who you are.
google Images also uses tags - I think that they are embracing tags more and more anyone played the googles image game. free tagging from the masses. democratic or slave- labor matrix Style? ;)
I am not talking about Google Search Vs. Digg though (that is like comparing apples and oranges), I am talking about Google News Vs. Digg - or for the long-tail - Tailrank/Techmem Vs. Digg.
The point being that they're generating a 'front page' of news using implicit rankings based on incoming links - isn't this more innovative than Digg's explicit vote system?
Google is good for search, digg is great for finding out whats new and upcoming. We need both.
Google does track clicks, not every click but it takes a sample and this obviously counts towards future ranking.
I think that herd mentality has a place. *Especially* in searches. Many times when I go to find something on the internet, i *want* to find the 'normal' pages for that. That is to say, the pages that the herd considers to the the authoritative page. Even if it doesn't have the best information, which may be discovered by an algorithm search, it will have the 'normal' information that is consumed by the 'herd' of the world.
It would be cool if google would allow you to tag items and then maybe give you a choice to search via the herd or via the algorithmic indexes. Hmmm.... i get to use the word algorithmic *so* infrequently. haha!
Do you think then that the herd can be trusted.
And then, how do you stop yourself being trambled by the opinions of people whom are possible (and generally) uneducated on any given topic. The point is, Google tries to determine validity and authority of authors, while Digg is "just another popularity contest".
The argument is: are we ok with that?
I don't see the devide. When someone creates a link to another page that person is actually "tagging" that page. (The name of the link being the name of the "tag".) And just like other "web 2.0" websites people search for information based on those tags. So it seems Google has the best of both worlds!
Post a Comment
<< Home