Filters: Part 6: Trading Places

Note: This is the sixth in a series of posts I’m committed to writing about filters; I started with the principles of filtering, and will proceed to blow up each of the principles in as much detail as makes sense at this stage. Earlier I looked at network-based filters, and then spent time on routing, went on a tangent to look at bringing responsibility into publishing, then looked at “designing for serendipity”.  And now I’m going to spend a little time thinking about what it would mean for a person’s collection of filters to be transferable, even tradeable.

“The future of search is verbs”

There was a time, it must have been around 2 or 3 BG, when I used search engines with funny names. Names like Dogpile and Mamma and Copernic. I found each of them useful in different contexts. I moved off all Microsoft systems when OSX arrived, and that meant no more Copernic. If I remember correctly, Dogpile searched; Mamma searched across a number of engines, and let you choose which one(s) to use. But Copernic did some other stuff, really useful stuff. It let me put my search results into a folder, which I could save if I wanted. I could add one folder to another. I could subtract one from the other. I could mail the folder to others. And I could ask for the folder to be “updated”. Retain the old results, see the new ones that came in; compare the search output between two or more dates. I could even ask for the links in the search results to be revalidated, and mark some of the historical links as broken.

Those were the days.

Around 2008, I remember Esther Dyson quoted Bill Gates as saying “the future of search is verbs”. I have a lot of time for Esther, and if she thought something was important, it gave me pause for thought. And she thought that what Bill Gates had said was profound. I was then still stuck in my search/subscribe/converse/fulfil model from 2003, so that was the lens I placed on her comment. Aha. So the future of search is fulfilment. People don’t just look for things aimlessly: when they search, they hope to find something that they can do something about. For a search to be valuable, the output has to be actionable. Since that was why I had consistently included “fulfilment” as my fourth “pillar”, I was content.

Now, instead of relatively static web pages that need spidering and indexing, we’re in the business of filtering firehoses. The same rules apply. Filtering eases comprehension. But that comprehension is usually for a purpose, an action to be performed related to the filtered stream.

Maybe I’m warped. But sometimes I think of search and subscribe a bit like we’re taught to think of add and multiply. Multiplication is repeat addition. Subscription is repeat search. With an understanding of what’s changed between “searches”.

Acting in the stream

None of us can deal with firehoses. So we filter. When we filter, we do so in order to act. So filtering is part of learning how to do something. When someone watches you do something, that’s teaching. So much of what I learnt, at school and in later life, came from observing someone else who knew how to do something.So when it comes to a time when we’re all living in the stream, a person’s ability to do something depends on her having learnt how to do it. Which may come down to knowing about the right filters to use.

A person’s collection of filters becomes some sort of toolchest, where the instruments that allow that person to do the job are kept and looked after. This is as true for knowledge workers as it was for artisans and craftsmen.

I’ve been fascinated by craftsman’s guilds for decades now; the history of the guilds in London is incredible, particularly when you see just how they were committed to social change via education, training and apprenticeship.

Which brings me to my final point, how filter-sets can help transfer knowledge and learning.

Learning in the stream

Many years ago, as I began to lose all interest in email, I tried an experiment. I opensourced my mailbox to my team, allowed them access to all my mail. [In point of fact I set up a separate mailbox for “private” mail, off the beaten track when it came to spam, corporate or otherwise, and proceeded to make my “official” mailbox open to my team.]. And then I sat back and watched. Something very strange happened. Rather than spend time looking at my incoming mail, I saw my guys spend far more time looking at my “sent items”. Most of them had gotten used to what was in my incoming mail anyway, I would always involve one or more of them. So what they wanted to do was to look at how I handled the things they may not have been involved in. In effect, some of them chose to learn not by watching what came to me but instead by watching what I did as a result.

In the world of stream/filter/drain, filtering is part of how we do something. So there is something to be learnt by looking at how the filters are set and chosen.

Since filtering is something we do on the “subscribe” side, it would mean each of us had our own set of filters, hand-crafted to our needs. Filters we knew how to use, how to refine and improve. This makes for some interesting possibilities for new hire induction and training, and in fact for many types of role-related training. You could take the filter set of an exemplary knowledge worker in a specific role, and make it available to others you sought to train in that role. A transferable set of lenses.

There are other interesting possibilities; in a hierarchy I could try and “wear” the filter-lens of a colleague, a peer, a subordinate, a boss, “see” the stream from their perspective. In a flatter, networked organisation those  labels may not have the same meaning, but the principle remains. Make it easy for you to see the world from someone else’s perspective. Make it easy for someone else to see the world from your perspective.

We’ve seen some of this before, initially with stuff like blogrolls, then with bookmarks, then with sets of subscriptions. Filter-sets have value.

And if they’re in digital form, and they have value, then it’s only a matter of time before people find a way to standardise inventory exposure. Discovery processes follow, and before you know you have valuations and negotiations and trade.

More later. Have you had enough? Or should I go ahead and complete the set, I have only four more to write. Please keep your comments flowing.





12 thoughts on “Filters: Part 6: Trading Places”

  1. keep writing … the world needs more thought rather than automated processing of ‘stuff’ … quality of output does not equate to volume of input

  2. Thank you for the article, and I concur with the essence of the article. In my experience too, verbs and adjectives catch behaviour and attitudes rather nicely and without too much fuss.
    When it comes to text analysis, grammar rules!

  3. @joachim I think we have a long way to go before we understand the importance of tacit knowledge, relationships and patterns

  4. @jp I have, over the last couple of years, been experimenting with extraction, categorization, query making and summarization using PoS tagg/RegExp methods, and I must say the data bends to yield when the word is treated as a word, and not a number.

  5. I like your line of reasoning, JP, but in some cases I think the filters become part of a subconscious process, a heuristic that is quite hard for its owner to communicate.

    Fortunately, there is quite a body of evidence that you don’t need to formalise these heuristics to allow other people to learn them: simple observation will often pass them from one person to another, without either party ever being able to articulate exactly how they are filtering. Kathy Sierra gave a nice summary of some of the science around this in her Business of Software talks from 2012 (from about 36 minutes in) and 2013.

  6. @hermione thanks. I agree, I’m also a big fan of Kathy’s. I don’t see this as an either/or: sometimes we can do something to aid the filtering process, sometimes it happens without our being able to articulate what we did in the process.

  7. A fascinating series indeed
    In the world of market data – vendors like Thomson Reuters, Bloomberg, FactSet etc that cater to investment & cap market professionals have been in a race to keep adding more content sets to their terminals and feeds. Like a race to who has the fattest hose-pipe. In that context, how does filters work with quantitative and textual data? By sources, by topics or other entities? Currently most market data vendors use “screening” at the consumption end as a metaphor for filtering. Is it a close substitute?

  8. On learning from JP’s sent box. We built an enterprise workflow application with web widgets workers could easily customize for collections of knowledge and transfer to ramp new hires. Unfortunately we built it out of Microsoft “web technologies” and found they didn’t understand latency in deep collections (as they do with Windows) so we hit a full stop. We contributed our results to a study McKinsey published on knowledge transfer at work.

Let me know what you think