Technorati uses both del.icio.us and furl.net to populate tag based search results. Wouldn't it make sense to use both bookmarking sytems to promote your meme?

Lee:

Makes a lot of sense to use both. In the end, I think you pick the one where the crowd you want to influence hangs out.

One of the motivations in my strategy (not stated in the post) was to use a method that was easily automated. It winds up that both FURL and del.icio.us make it pretty easy to add bookmarks and get them in RSS. However, del.icio.us is a little better supported with third party apps.

That notwithstanding, it winds up being easy to get a bookmark into either FURL or del.icio.us and then automate making a blog post, but you would have to do double hand entries to get the post into BOTH FURL and del.icio.us.

Probably, you could write a javascript to automate the FURL and del.icio.us entries.

At any rate, from what I can see, the method I used with del.icio.us could work with FURL.

Bud

Using tags to spread a meme is like pushing string, it is silly and fruitless. The benefit of a folksonomy comes from many people tagging one item with terms that make sense to the person doing the tagging for their own retrieval. Out of this comes some patterns of use that are of benefit, which follow the net effect or power curve as it is called.

Technorati's approach to tagging supplies none of this and has far less value than their extremely helpful keywords approach to finding information. Technorati is improving there methodology behind the tags as they realized their errors (rather quickly) and they now will follow the rel="tag" that points, not to their site, but to the object of the tagging (just as del.icio.us does. This is a far less efficient tool than del.icio.us as it requires properly building the tags for every entry rather than the quick tagging in an external tool. Technorati tags does have a nice interface, but the their own tags are the least valuable items on the page.

Content creators have been and still are the poorest link for tagging in an open system (a reliable link in a closed system that has training). Why? Content creators have the poorest view of their own subject, as they can guess on the value of their content, but others with different perspectives will add richness.

The richness of tagging that make a folksonomy comes from the work of many, like del.icio.us, flickr, furl. etc. These are broad or narrow folksonomies (broad has many people tagging and narrow only has a few people tagging the same items). Non-folksonomies like Technorati Tags do not have richness and other methods of search can actually produce broader and better results.

Wow Thomas, your comment could have been a post in and of itself. There are two focal paragraphs, I would like to take issue with:

"Using tags to spread a meme is like pushing string, it is silly and fruitless. The benefit of a folksonomy comes from many people tagging one item with terms that make sense to the person doing the tagging for their own retrieval. Out of this comes some patterns of use that are of benefit, which follow the net effect or power curve as it is called."

We agree that people are using folksonomies as individuals for their own individual retrieval. From there, our perspectives part. Let me lay out some scenarios. First, folksonomy can be viewed as a way of collecting data about how people perceive a web artifact (i.e., a web page at a given URL). Indeed, your notion that the distribution of tags may follow a power law distribution (i.e., there is a dominant view of the group) is likely to hold with frequently cited URLs that have been "processed" by the group. However, with new, unprocessed URLs it will not.

Second, your model of folksonomy leaves no mechanism for changing the group consensus. In your world, we arrive at the group consensus (power law distribution in your terms), and then are there until the group changes its mind, who knows how. I suggest a mechanism for change is to start to relabel web content. In a post last night, I noted that many people do this with short web log articles pointing to specific URLs. The point is that some people are changing and recrystalizing the group opinion by effectively labeling the web world we live in.

Third, folksonomy can just as easily be used as a data collection tool to find out more than what the consensus opinion is. You can see what the whole distribution is saying and possibly identify influence points. Your power law notion essentially throws out a lot of data. Why stop at consensus?

Thomas you then note:

"Technorati's approach to tagging supplies none of this and has far less value than their extremely helpful keywords approach to finding information. Technorati is improving there methodology behind the tags as they realized their errors (rather quickly) and they now will follow the rel="tag" that points, not to their site, but to the object of the tagging (just as del.icio.us does. This is a far less efficient tool than del.icio.us as it requires properly building the tags for every entry rather than the quick tagging in an external tool. Technorati tags does have a nice interface, but the their own tags are the least valuable items on the page."

We agree that technorati is an incredibly inefficient way of spreading your meme. You have to write a whole blog post. However, to sway people, you usually have to engage in some effort at putting forth your reasoning. Blog posts can be efficient enough for this purpose.

My understanding of how technorati updated their tagging mechanism was simply to say you did not have to point at them. You are however still essentially just labeling your own post with the tag.

Further, however, the type of tagging technorati is doing could be viewed as simply joining a conversation on the topic in question. Your cut at the topic mixes in with various other posts folksonomically classified on services like delicious and furl. In some sense, adding your posts to this (perhaps, in your view, unholy and uninteresting) melange helps define the meaning of the already existing tag. The recently introduced tagsurf pushes along nicely in the conversational view.

Personally, I would rather introduce a neologism instead of attempting to redefine an existing tag people were already applying.

I do, however, find the technorati interface better for retrieval than delicious. I would like an interface that allowed me easily to see how a particular URL was tagged across all of the people in all of the tagging services.

Bud --

There is no group mentality that can change. People tag with their own individual interests and perspectives. These individuals when viewed as aggregates appear as as groups around tags, but the groups of individuals do not know they have formed so treating them as a group is a tough endeavor. I think that should cover a few things in your response, but I will respond by item.

Item one: Agreed objects that have little tagging do not have power law effects, that is part of the definition as it is part of any trending with statistics you need a population that is large enough to define anything. Rough trending seems to appear (with the current set of people tagging in del.icio.us with about 25 to 40 tags, and definite trends are filled in with around 100 people tagging. As the make up of those tagging changes and it will as most currently are early adopters, or the next bunch that rides in, the community will broaden as will the perceptions and opinions. This was made clear in my econometrics and statistics work in grad school years ago.

Item 2: There is no group so there is no consensus only a snapshot of what trends are at any time. Essentially Google and other search tools that pay attention to the content within the anchor tags and weight their algorithms with that metadata were the first folksonomy tools. You ask how will people change the spikes in the power curve, first the spikes in the power curve are not the definitive answer to anything, it is just what the people call something when the snapshot of tags was taken. Secondly, terms change over time in many disciplines and cultures (particularly sub-cultures). Generations have different terms for identical things. Media will promulgate certain terms as favorites (folksonomy is one current flash in the pan that will last its 5 minutes) that then spread through out the culture. It does not necessarily take our own retagging objects (I say objects because it is not only the web that is being tagged but physical space too with mobile devices has been the rage in some places for a year or two now) to change trends it is others that come along and tag the exact same object behind us. Vocabularies emerge and change overtime and vocabularies are different across disciplines and cultures, as they get involved the folksonomy gets really rich and helpful for finding things across these boundaries.

Third point, you hit on my reason for coining the term. Power laws are not about the spikes they are about the whole curve. The spikes are not that important to me, they are of interest, but I find the long-tail very interesting. I can't use the yellow pages of a phone book because I grew up in too broad of a culture and I name things differently than the yellow pages, so they don't work for me and never have. If you go read Clay Shirky's posts on Corante and my comments you will see that this is my point from day one. The spike in the power law curve is only one point on the curve and the rest of the curve is made up of such a rich melange of views and perspectives. I love the long-tail far more than the spikes. The patterns of use of interest and value are all through out the power curve.

My problem with Technorati Tags is that they are only one perspective for that object. That perspective is helpful for organizing on a site, but not organizing externally. Technorati's keywords are a far more valuable tool than their tags are at this point, no matter which direction they point. The changed direction of the pointing must now point to a URL that contains that tag term in the URL. That is just a tag and not a folksonomy that is rich. That does not allow any tail on the power curve as there is no curve to have a short tail or long tail. It is Google without any relevance.

Spreading meme's is hard work if that is one's intention. It takes years, or just dumb luck. Most of the time those pushing memes look foolish, with folksonomy I could care less. Coining a word has no value. Understanding systems and human cognition has value and only in that you are making things work more easily. The folksonomy is not what make Flickr great, it is a part, but most of it is a fantastic interface to a tool that does what other similar tools do, but with far greater ease and a lot more fun thrown in. Del.icio.us how ever is all about the folksonomy and the richness it provides.

I've decided to move my response to this conversation up to a blog post which is available here:

http://thecommunityengine.com/home/archives/2005/02/folksonomy_for.html

In summary, these arguments rely too much on the power law distribution, something that occurs only under certain conditions.

Bud