This page argues that “tags” as found in typical 2007 SocialBookmarking software should more properly called “mnemonics.”

Then true “tags”, tags that are assigned in efforts to identify EmergentStructure, are introduced, and differentiated from mnemonics.


By RashmisTheoryOfTagging, we understand tags not so much as categories, but rather, as mental associations.

…For example, I come across the book “Snowcrash” in my library. Immediately a number of related semantic concepts get activated: “book”, “science fiction”, “Neal Stephenson”, “Zodiac”. Other concepts might be more personal; e.g., “favorite author”, “airplane trip”. Still other concepts activated might be more about the physical characteristics, e.g., “paperback”, “bad condition”.
…With tagging there is no filtering involved at this stage, you can note as many of those associations as you want. This is how tagging works, cognitively speaking. Yes, it’s that simple.A Cognitive Analysis of Tagging, Rashmi Sinha


I call these mental associations “mnemonics,” to put it into a single word. (Mnemonic, memory, association.) Things that just sprang right out of our mind.

“Mnemonics” don’t even need to make any sense at all; If I have a page about robots, I might put down “vanilla,” because I had a memorable dream about a vanilla ice cream serving robot that looked sort of like the one on the page. It could be anything.

The word “tag” seem to imply some greater formality to me: like the name of a folder, or some other kind of deliberation.


Most, but not all of my tags in my SocialBookmarking are mnemonics. The few that are not, I call “bona fida tags,” or “Tag tags.”

Here is an example of a set of tags on an item in my UnaLog collection:

The following tags are mnemonic tags:

After a bout of StrengtheningMnemonicTags?, I had also added:

But the following tag is a bona fida tag:

It’s a “tag” because it’s:

Bona fida tags are often hard to remember, as well. The idea of the tag may just “pop in there,” (because I’m actively tracking for it,) but the exact spelling can be a lot harder to remember (because it’s specific- and therefor long, and also unique.) I often need to simulate TagsForTags, by looking for prior articles using mnemonics, and then scanning along the tag list for the relevant bona fida tags.


My subjective sense of tagging, is then something like this:


That is, I feel like I am me, (symbolized by a brain,) sitting in my mind, surrounded by little squiggles (“mnemonics,” “associations”, drawn to represent EEG signal) constantly coming out from me.

If I want to find a web page, I can find it’s URL by way of mnemonics – they are the interface between my brain, and the SocialBookmarking service’s retrieval of the web page’s URL.

From the SocialBookmarking service’s locating of the page, I can track it from there to a BonaFidaTag. From that bona fida tag, I can track to all the other resources on the web I’ve seen and tagged.

Mnemonics alone are too inexact for reliable rediscovery; This is the way to do it.


No SocialBookmarking service actually permits TagsForTags, or can distinguish between MnemonicTags and BonaFidaTags. So I simulate it in UnaLog, by just listing them together, and then pretending to search for links, when I’m actually searching for the tags. But I am drawing up designs for a social bookmarking service for just this.

You can, for example, have entry-time suggestions (recommending tags, based on observed mnemonics,) clarity over tags, relationships between tags, and very specific queries. You should also be able to start making top-down maps showing you the whole, and relating the article with several others.

Once you recognize this difference in software, you gain access to a middle-range between “folksonomy” (just collecting mnemonics,) and “top-down ontology” (or “pre-determined categories.”) I believe that EmergentStructure lies in this middle-range, and that it’s a crown jewel of the HiveMind.

See Also


I’m starting a set on “higher order tagging.” (Perhaps I’ll just call it: HigherOrderTagging?.)

This is all in order to respond to a post made by SamRose on SocialNetworkSecurity, about how the HiveMind can track conversational spaces searching for points of difference and structure and so on. How it can “see the whole,” as in SharedAwarenessSystem, and other stuff.

Ok, wow. This is great. I can see how this can make “resonance” with real human brain activity, because Mnemonics is kind of an outward relflection of that individual-inner brain activity.

A couple of thoughts about this:

DiversityInTechnoAidedSenseMaking If you look at sites, like del.icio.us, UnaLog, etc, you will see that many different people use the same set of tools in a wide range of different ways, presumably because they have unique processes going on in their brains for Sense making.

Take a look at HwoStyle?, for another, which is tough for many people to decipher, yet obviously highly useful for HansWobbe, and actually an amazing creation once you wrap your head around it. Also, by the same token, you can look at what MattisManzel is doing with WikiHive configurations. totally different but still useful. Different people solve similar problems in vastly different ways.


I (SamRose) believe that there are brain processes, and internal dialogue that “activate” when we are trying to make sense of things for ourselves, and different internal dialogue that happens when we are trying to make sense of things both for ouselves, and for others.

I think TagsForTags addresses these differences. I think it is a sound workable and useable system, that definitely addresses SenseMakingForSelfVsSelfAndOthers. But, I think it might be too confusing for people outside of our “niche” of people who even have some degree of understanding of what we are doing when we “tag” things.


This is why I think we need to look not just at MnemonicTag, and not just at TagsForTags or BonaFidaTags?, but also at the “objects” being tagged, whether they are our own MnemonicTag ceations attached to URI’s, or whether they are TagsForTags or BonaFidaTags?. When I am building these SocialBookmarking databases, I am building some of them for myself, and some of them for myself and others. The databases being built by me for both myself and others are purposefully trying to make a way to “connect” my sense making to others.

For instance, in del.icio.us, with people from http://cooperationcommons.com project, there are tags that we have all discussed and agreed to use, and we include these tags. In P2P foundation, we are using a similar tactic, but with wholly different tags (many for the same “objects” that are being group-tagged by people in CoopCommons?)

One question I have is: How can we join together knowledge bases like these, without having to radically restructure the way people are doingthings? So, naturally I thought about looking at the “objects”. The same thing could happen between just 2 people.

Key questions:

  • What if we could “signify” when we are creating MnemonicTag content for ouselves, vs for ourselves and others?
  • How many same or similar URL’s have we both (or all) tagged?
    • How do we detect similarities in “objects” when they are not the same exact URl,but content of 2 sites is similar?
      • How are we individually connecting same or similar tagged objects to other objects? How does the way that I am doing it look, compared to the way that you are doing it?
  • What patterns are emerging in your Mnemonics, in mine?
    • What if you could see these patterns (which is possible to a crude extent with “TagClouds?”), and tag, or “mark” them, instead of just tagging tags?

I intend to post more about HigherOrderTagging?, so, I am committed to this conversation.

For instance, in del.icio.us, with people from http://cooperationcommons.com project, there are tags that we have all discussed and agreed to use, and we include these tags.

To my mind, these are all BonaFidaTags. They are unique, intentional, deliberate, actively tracked (by a whole community, even!), and so on. They aren’t tags that “just popped in there.”

I love your questions; They’re the sort that I’m really excited about.

I can envision software that identifies likely synonyms, or similarities, and suggests connections and unifications, and perhaps even noting important differences.

If I enter something, I’d like it to look at the website, the mnemonics I gave, and then suggest BonaFidaTags from my own (that seem relevant) and from other people, and so on.

If I could say, “If I use this tag, also recommend these,” so that I can go through a checklist procedure, that’d be awesome. I’ll put some more skin on these ideas later.

Too busy immediately.

My ideas above are really, really not an argument against your ideas about TagsForTags. I owuld use TagsForTags today, and actually I think if TagsForTags were presented in systems like del.icio.us et al in a way tat makes sense, many people would yuse them (actually, if you look at, for instance if you look at my http://del.icio.us/network/srose a couple of people listed, http://del.icio.us/mbauwens and http://del.icio.us/hrheingold actually do apply “categories” to their tags. (Come to think of it, Drupal allows for “sub taxonomies” too). But, del.icio.us does not do anything with those categories to make them “findable” like the (usually) MnemonicTags? themselves in my friends’ tag-lists. So, this would be a useful feature, I definitely agree with you. Plus, you could eventually include other people’s MnemonicTags? under your BonaFidaTags?, for instance, if some of these services allowed for cross-site sharing of data (possible with MicroFormats “rel-tag”, for instance).

understnad you are too busy, and no problem. Just thought I would get thoughts out of my head while they are fresh.

Plus, it seems to me that SemanticWeb code, like http://simile.mit.edu/ has already addressed the simple idea of allowing a use to sort by “tagged” object from many sources, but currently most of these services only compare “object” from within their service.

Oh, sorry – if I seem un-enthusiastic, it’s because there are some things going on in my life that are real downers, and it’s probably showing up in my interaction. :( (Don’t want to talk about it.)

But I’m actually really excited and inspired by your energy here, and I want to continue.

I think TagsForTags addresses these differences. I think it is a sound workable and useable system, that definitely addresses SenseMakingForSelfVsSelfAndOthers. But, I think it might be too confusing for people outside of our “niche” of people who even have some degree of understanding of what we are doing when we “tag” things.

One thing that perplexes me is that most people put only a single tag, at most, on their links, on UnaLog.

So, I see that my use of it, and the way that we use it, is substantially different than most people.

That’s not a show-stopper, to me; But I would be excited to see more clearly what the difference(s) are or could be.

It reminds me of the difference between people who use wiki as a posting board, and people who use wiki for TheoryBuilding.

We are doing something like TheoryBuilding or pattern recognition, I think, in what we’re talking about here.

How can we join together knowledge bases like these, without having to radically restructure the way people are doingthings? So, naturally I thought about looking at the “objects”.

Can you explain about the objects?

Are you talking about the tags themselves (the underlying data structures?) If so, I’ve been thinking about that as well – TagsWithMetadata?, AnnotatedTags?, or UniqueTags?. Perhaps that is what I should write next; Because it’s fundamental to a lot of other stuff.

My thought is: Not to radically restructure the way most people do things, but rather, to build tools to help people who are doing TheoryBuilding (or whatever we’d call it) with SocialBookmarking software.

It might happen that people who are doing theory-building might be able to build structures, that people who are just casually tagging stuff might want to plug into. Then again, everybody might just say, “That’s too complex.” We can’t know until we try! :)

Actually, by “objects” I was talking about the URL’s that the MnemonicTags? are pointing to.

But, now that you clarify that TagsForTags is most useful for TheoryBuilding, I have to really, really agree with that. This is a method that can work really well for TheoryBuilding. If TagsForTags were also a link, it could give many individuals ways to quickly join their existing tag bases together. all they would have to do is agree on what the “TagsForTags” should be. You coudl even import tags from existing services into this new service. Mayeb you could even somehow grab “Categories” from wikis, which are a form of BonaFidaTag?

The other thing that I was talking about with “objects” is basically starting with the URL tagged, and using it to show who all tagged it, and with what, from multiple tagging services. Also, being able to compare URL’s we’ve both commonsly tagged, could be even more meaningful than comparing “tags”. Or, using URL “objects” we’ve tagged (the original links to sites themselves) as a basis to see where our areas of interest are similar. Even comparing “people who tagged this, also tagged” and then comparing that ‘also tagged’ data, or perhaps other useful data mining.

I cam imagine:

I’m browsing the web. I stumble upon the article How children lost the right to roam in four generations. I see at the top of my browser the BonaFidaTags people have applied to it.

“kidscantroam” shows up in my browser, and other BonaFidaTags (that I can’t even imagine right now – perhaps something to do with child development, social development, social perspectives) but not the myriad mnemonics (distracting.)

I mouse over them, and some have descriptions.

If it’s progressed to the point where there are communities of people tracking BonaFidaTags, then I can get a sense for what communities are tracking those tags, and connect with them, and learn what else they’re tracking, and how they see it fitting into their world story.

Lion, a side note, do you think this would be worth experimenting with in OddMuse wiki?

How would it work? (-?)

I have always been envisioning SocialBookmarking as the vehicle, but perhaps there’s some way to do it via wiki?

Perhaps something similar to WikiAsCollage? (A hybrid form?) Using OddMuse as the base?

I’m not sure if I should feel “smart” or “stupid” as a result of JUST discovering that the oddmuse calendar function makes it very easy to tag pages. For example…

 [[month:+0 pageName .tag1 .tagN .tag1.tagA ...]]

makes it very easy to create a page name that can then be easily found.

I’m absolutely amazed at how well this works in conjunction with the new (pre release version3) of DiiGo, when you standardize tag sets between the two environments. Add to that the effective use of a WikiHive’s namespace per Audience and the WikiSyllables? concepts and (at least for me) the result are OUTSTANDING!

Now I really want to help Lion expand his LocalNames ideas to mesh with these insights.


Yes, these days I keep thinking “This big complicated application? Wouldn’t it be better to do it with a wiki?”

Flight of fancy: a social bookmarking wiki using the BigBucketsFirst technique, perhaps it would work like:

  • people see something they like, so they paste the exact URL (for example, “Gever Tulley: 5 dangerous things you should let your kids do”) on the “new sites” page of the wiki, with a bit of description and a few tags. (Perhaps a DatePage ?)
  • Other people check out the URL and add a few more tags (“childrencantroam”) and commentary.
  • If the section dedicated to that URL gets big enough, someone makes a new wiki page just for that URL (using some standardized URL-to-wiki-page-name system, so we don’t get lots of different wiki pages mentioning the same URL?), and moves that text off the “new sites” page, leaving behind a link to the new wiki page.
  • Some of the tags get popular enough that people create wiki pages for the tags, describing exactly what they really mean on this wiki, with manually-edited links to the top 5 or so “most relevant” wiki pages (or perhaps transcluding their first paragraph) – both wiki pages describing a URL and wiki pages describing other related tags – and an automatic list of every page mentioning that tag.
  • Some of the tag pages will name people interested in that specific subject, searching for more web pages and other information on that subject (so these are “bona fida tags” ?)
  • Other tag pages will discourage people from using that tag (Please use more specific tag X, Y, or Z instead), or from posting URLs in that specific category (We already have a bajillion web pages in this category; don’t bother posting yet another unless it’s at least as good as one of these top-10 sites in that category).
  • … people pages, with 2 groups of tags: tags that describe the person, and tags that person is interested in watching …

Could we / should we turn the AboutUs wiki into this sort of social bookmarking wiki?



I’m engaged by this line of thought.

WikiAsCollage keeps coming back to me, and I think again about SemanticRegions?, UserInterface issues.

I often think that tags (and Categories) are not “kept” nicely, because it’s an involved operation to change the tags on just one page. If we want to interact with tags and categories, then we need (it seems to me) CrossCuttingEdits as a part of the UI.

But we can have separate UserInterfaces? for different purposes; What matters is that we use the WikiEngine as a database server. The database is transparent and open to the public.

Will CrossCuttingEdits overload RecentChanges? Perhaps the difficulty can be papered over by even more sophisticated RecentChanges visualizations, that know how to look for these cross-cutting edits. We’d also need mechanisms for rolling back mass changes. (“replace tag X with tag Y for those 17 links.”)

I believe in publicly maintained databases.

Hans, I don’t understand about how tags work with the calendar..? (I need to poke at that a bit more; I don’t think I understand the mechanics well enough yet.)


Define external redirect: StrengtheningMnemonicTags TagsWithMetadata HwoStyle AnnotatedTags SemanticRegions UserInterfaces BonaFidaTags UniqueTags TagClouds WikiSyllables CoopCommons HigherOrderTagging MnemonicTags

EditNearLinks: WikiEngine DiiGo AboutUs UserInterface OddMuse