Freewheeling on “Filtering on the way out”

I said I would post further on David Weinberger‘s Four Strategic Principles as outlined in his new book, Everything is Miscellaneous .

David’s first principle is to filter information on the way out, not on the way in. I’m still working on it, masticating it, there’s some work involved, but I like the early flavours I can taste. So I thought I’d share with you the kind of stuff that went through my head when I saw that sentence and read what followed. Humour me.

1. In order to filter on the way in, we need to have filters, filters which can act as anchors and frames and thereby corrupt the flow of information. We’ve learnt a lot about anchors and frames and their effect on predilections and prejudices and decision-making. With David’s first principle, we reduce the risk of this bias entering our classification processes too early.

2. I think it was economist Mihaly Polanyi  who talked about things that we know we know, things that we know we don’t know and things that we don’t know we don’t know. Again, filtering on the way in prevents us gathering the things that we don’t know we don’t know.

3. The act of filtering is itself considered necessary to solve a scale problem. We can’t process infinite volumes of things. But maybe now it’s okay to be a digital squirrel, given the trends in the costs of storage. [Sometimes I wonder why we ever delete things, since we can now store snapshots every time something changes. We need never throw away information]. Filtering on the way out becomes something that happens in a natural-selection way, based on people using some element of information, tagging it, collaboratively filtering it.

4. I like the idea (proposed by David) of there being no need to throw stuff away. You just have to not-find it. If you can’t find it you might as well have thrown it away, and if it all costs the same then who cares? Reminds me of the Douglas Adams definition of flying: jumping off a tall building and missing.

5. Collecting information this way is fine, but it has no value unless someone tends to it, someone looks after it. So maybe I shouldn’t be thinking ‘not-find’ and instead I should find ways of incentivising people to clean up their information. Maybe there is a Silent Spring for information. I somehow like thinking of bad DRM and proprietary tools, methods, structures and standards as weeds that strangle the life out of good information. But then I would, wouldn’t I? Walled gardens have the worst sort of weeds.

Just musing. Comments welcome.

19 thoughts on “Freewheeling on “Filtering on the way out””

  1. JP, under the traditional belief that every idea is but a reworking of a previous idea, I think you would learn a lot by consulting Heinrich von Kleist’s essay “On the Gradual Fabrication of Thoughts While Speaking” (if you have not already done so) to elaborate your thoughts on “filtering on the way out.” As I explained when I wrote about Kleist in my first blog, we really have no way of knowing how serious Kleist intended this piece to be taken; but, in our present context, there is much to be gained from taking it very seriously indeed!

    http://blog.360.yahoo.com/blog-Mff23hgidqmHGqbcv.lfskakEtS6qLVHUEMFUG4-?cq=1&p=118

    The basic thesis is that ideas only begin to reify when we are compelled to talk about them, and it goes without saying that the construction of coherent speech is basically an act of filtration. To steal from Moliere, we have been filtering on the way out for as long as we have been communicating with others!

    Now here are a few comments on your numbered thoughts:

    1. According to the current reputable brain models, perception is all about “filtering on the way in” and about the ways in which the filters interact to create perceptual categories.

    3-4. On the other hand, if we are to appeal to biology, then we ought to entertain the hypothesis that FORGETTING is a biological phenomenon and (if you believe in selective processes) that it has survival value. We have made a lot of process in studying memory at physiological, psychological, and even social levels. We need to study forgetting just as assiduously, because it, too, has implications for our information systems!

    5. I read this one as a variation on: A library without a librarian is just another collection of books. Librarians add value by, in your language, tending to the collection. They do this is all sorts of social ways that “digital libraries” (not to mention search engines) are still inclined to overlook.

  2. Stephen has a good point – often ideas only ‘solidify’ when we need to communicate them, which is one reason these conversations are so useful (and your ‘provisionality’ is so good – ‘fail fast and fail early’ is a filtering mechanism as well!) – this is why learning via teaching is so effective (for the ‘teacher’ anyway – maybe not for the ‘student’).

    ‘Filtering on the way in’ is difficult to avoid because of our inbuilt and developed norms and biases – overcoming those is often as high a hurdle as the scalability issue. However, overcoming our bias is probably more useful than attempting to scale any further – as Stephen points out, being able to forget stuff is vital to survival too. Not-filtering-due-to-bias will allow us to see things we don’t know we don’t know.

    As for the rest – chuck it all on the Web, and tag it. We may never want to see it again – or we might later, or somebody else might at anytime. Somebody (sorry – I’ve forgotten who!) mentioned recently the possibility of (e.g.) Yahoo combining the del.icio.us tagging with existing search algorithms (using links) to add some richness/meaning/relevance to the results – is that a useful starting point for librarianship of digital knowledge?

  3. Haven’t received the book yet… but I have it on order. :) The “filter on the way out” principle seems related to the sound engineering approach of initially recording the input “dry” (without effects) so you have the full waveform to play with as you edit and mix the tracks down. Applying effects on the way in necessarily limits what you can do later.

  4. JP, isn’t “the gradual fabrication of thoughts while speaking” in a “collective environment” called “brainstorming?”

    Ric, my wife teaches middle school and has been experimenting with getting her students to teach some of the classes. This means that they get to fabricate their thoughts while speaking, too! (There is also a lot of group discussion and project work in her classroom.) She has been pleased with the results, but she teaches in an independent school! As to the “librarianship of digital knowledge,” I think that, in the interest of thought-fabrication, we should not discount the value of conversations with human librarians!

  5. Stephen, my concern was about groups filtering on the way in, which leads to groupthink, narrow-mindedness, heresy and suchlike. A collective mindset filtering similarly on the way in is scary. La la la we’re not listening is the stuff of nightmares and genocides.

  6. JP – in a collective environment, I think that (and this seems non-intuitive to me) you end up with the aggregate of all the individuals’ filters, rather than perhaps the intersection of them – that’s where groupthink comes about. Everybody takes on all the biases of all others and you end up with a megafilter. How extensive this effect is will depend on the makeup of the collective – the more diverse the group, the more varied the biases, the less likely they are to accumulate? The ‘wisdom of crowds’ principle seems to counter this to some extent, by aggregating individual responses rather than a single collective one, removing some of that ‘filter accumulation’ and bias reinforcement.

  7. Ric, my “educated” guess is that there are so many different social elements that come into play in a “collective environment” that there is no “context-free rule” (or even set of guidelines) that determines whether the filters aggregate or intersect. As a matter of fact, social interactions are so dynamic that, over the course of a single engagement, biases may come and go in a variety of different ways that can only be explained (if at all) by a POST HOC examination of a documentation of the engagement (which would probably have to account for not just text but also paralinguistic features, such as body language). However, our individual communicative skills still take care of “filtering on the way out.”

  8. How right you are Stephen – what ARE we thinking? This is certainly a big bite to take – and you point out a few of the difficulties in chewing on it. I was attempting to keep it simple to focus a little (oversimplifying, probably) on the thinking behind JP’s concerns – maybe I should be filtering on the way out more! This is certainly an area of discussion in which a ‘context-free rule’ is non-existent (I see it as a theoretical construct).

    BTW – I’m sure your “guess” is better-educated than mine … in this sort of discussion I’m a dilettante at best.

  9. Meaningless Filtering?

    Can there be meaning without filters? Or are filters implicit in any effort to assign/perceive meaning? The brain filters assiduously, and actively fills-in holes from memory and extrapolated edge effects. This is/must be true of animal perception, also. So do animals experience meaning?

    At some point there must be consensus on word use and referents, so filters must be shared — so that when I say “mother” you don’t think of an SUV, or a mollusk. E.g.

    When we argue, we are pushing for alternate boundaries between filters, perhaps. Or accuracy of certain overlaps: Iraq war/noble/evil. Many such filters have heavy emotional investment, which is a clue that they access basic biological, conceptual, and personality functions.

    So filtering is not optional, but it would be well not to take one’s own current filter-set too seriously, or overlap the contents permanently with the eternal/perfect concepts!

  10. Brian, “meaning” is one of those “hot-button” words that is usually best left out of the conversation. (Alan Turing said the same thing about “intelligence;” and it may be the wisest thing he ever said!) However, you are right to look at “wet brain” behavior for more substantive data. By now we know that the brain does a lot more than GESTALT-style gap-filling and edge-continuation. Much of Gerald Edelman’s recent work has gone into developing and validating a brain model (based heavily on networks of filters and an ongoing process of their reconfiguration) that accounts for how we delineate the very objects we see and recognize them as instances of different universal classes. He calls the mechanism behind this “perceptual categorization;” and, in his more speculative writing, he has explored the hypothesis that perceptual categorization is the fundamental building block without which consciousness (another “hot-button” word) would not exist.

    Needless to say, these are filters that are too deeply embedded in brain biology to be shared. However, part of Edelman’s consciousness model includes the emergence of language. That is the primary medium through which all of our “experiences of consciousness” can be shared. This particular idea predates Edelman by about three-quarters of a century, since most of the details (albeit hypothetical) were worked out by George Herbert Mead, who called the resulting model “symbolic interactionism.” Some of those details can also be applied to Kleist’s fabrication-of-thoughts-while-speaking model, which was in my first comment in this discussion.

    Your final comment is critical to the whole Edelman story, which is why I cited the element of reconfiguration. The “current filter-set” is a very evanescent thing. Both the repertoire of filters and how they are configured are constantly in flux (possibly even during sleep, which brings up one currently fashionable dream model). Heraclitian universal flux rules!

  11. SS;
    Yes, I was aware of much of that, and am in particular impressed by the discovery of the intimate role played by memory access midway through the processes of perception.

Let me know what you think

This site uses Akismet to reduce spam. Learn how your comment data is processed.