Comments on: are we hardwired to hierarchy?

By: Valdis

Valdis — Sat, 23 Jul 2005 22:40:35 +0000

Peter writes " We'll never catch terrorists by tracking info on them in single, static hierarchies." Yup. Yet, that is what we keep doing -- repeating old behavior and expecting different results. I riffed on this in a brief white paper -- hierarchy vs network in the war on terror. http://www.orgnet.com/orgchart.html

By: Peter Marshall

Peter Marshall — Fri, 22 Jul 2005 07:46:14 +0000

I agree with the “let’s not get all pissy” observation made in a couple of posts that BOTH hierarchy and tagging/association are useful, in their place… but I ALSO think that tagging is ALMOST always better.
Granting that this is partly a personality test thing, and many people will like hierarchy more than I do. Some may possibly like it less (hard to imagine).
But I think a fair test of this is to compare the utility of X1 (www.x1.com) vs. MS Explorer. I started using X1 as a desktop search tool about a year ago, and I now use Explorer about 10% of the time to go to stuff.
For non-X1ers, it essentially provides an interface where you type in keywords, and as you type, it instantly retrieves matches based just on indexing of ALL contents of ALL files, along with a “preview pane” showing the contents of any file you select. The list of matching files updates instantly with each keystroke, so no need to submit, look at results, go back and try again, etc. It’s instant feedback — just keep adjusting the search terms till you get what you need.
This has the useful side-effect of helping you find related useful things you hadn’t thought of or remembered… but even without that, it’s just a better, faster way to find almost anything EXCEPT images or music or other non-text content. Occasionally it’s faster just to go to a folder that has everything for a specific project.. but 90% of the time, in my experience, X1 is better.
A second confirmation of the same thing is how well Gmail finds stuff I need, again with no folders, just with keyword search. I never miss folders for my mail — it’s all in one big bucket, and I always find what I need right away.
Of course, I’ve never understood people that have millions of Outlook folders with hierarchy — I prefer even Outlook’s anemic search to the overhead of categorizing and then traversing the categories, not to mention the problem of needing multiple pathways and changing pathways to the same content.
In my view, hierarchy is RARELY right, but our systems and processes use it MOST of the time. This should be flipped on it’s head, and it’s important to recognize this, not just to say “well, both are valuable”. We’ll never catch terrorists by tracking info on them in single, static hierarchies.

By: Valdis

Valdis — Thu, 21 Jul 2005 11:45:49 +0000

Julian, meshes can contain trees but not vice versa.
The answer to your question is “it depends”. We can find trees in the mesh, but as soon as we find cycles/loops in the trees, they stop being trees. Kind of like life in an atomic collider…

By: Jay Fienberg

Jay Fienberg — Wed, 20 Jul 2005 08:14:12 +0000

Not sure if you'd find this relevant, but I've written about and made some diagrams showing how tags reinforce hierarchies. Tags stuck on trees (with pictures!) http://icite.net/blog/200506/tags_trees.html Tags remove a middle level of hierarchy, but they don't replace hierarchy--tags depend on trees. With tags, we get less access to hierarchy. This is a tradeoff--we get hierarchy out of our face and don't have to deal with its awkwardness, but we also forgo the ability to shape hierarchies on which we are being dependent.

By: Julian Bond

Julian Bond — Tue, 19 Jul 2005 05:31:10 +0000

If a tree structure appeared in the mesh and there was no taxonomist to see it, would it still exist?
Part of the problem with hierarchies is the world seems to be mesh structured rather than tree structured. There’s always just enough cross links to screw up the nice neat tree we’re trying for.
Which is probably why every so often big companies experiment with matrix management. Just before closing down the whole division.
Time for another article; “Outliners considered harmful”; or to ponder the difference between traditional XML and RDF.

By: HorsePigCow: Life Uncommon

HorsePigCow: Life Uncommon — Mon, 18 Jul 2005 10:41:28 +0000

Better than me: the hierarchies that bind

My mother once said to me, “If you are on a date with a guy who is nice to you, but rude to the waiter, he is not a nice guy.” My mom is smart. I’ve taken that lesson with me throughout whichever status I’ve been.

By: Smarter Stuff

Smarter Stuff — Fri, 15 Jul 2005 03:53:08 +0000

Your Brain At Work

Quick: umbrella, flying, nanny As Jeff Hawkins points out in On Intelligence, your brain just sits in the wet, warm, darkness of your skull and tries to make sense out of patterns of nerve impulses. Via gaping void (Not

By: Valdis

Valdis — Fri, 15 Jul 2005 01:39:29 +0000

“That’s not how it worked at ANY of the companies I worked for, back in my Madison Ave days ”
And that was why you got the fuck out of there!
BTW, hierarchies are a special type of network know as a “tree” — so it is ALL networks, patterns of connections.

By: Jon Husband

Jon Husband — Fri, 15 Jul 2005 00:17:00 +0000

You probably know what I think … philosophically.
I like Frank Patrick’s points, and I agree strongly with Valdis’ points .. theoretically.
Practically, it IS “both/and” in my opinion, but only when there are those higher up in the hierarchy who think like that, are predisposed in terms of personality, or are in front of problems where hierarchy is creating more problems than it is solving (although there are many examples where that is the situation but the hierarchs wouldn’t ever think of tryiong other ways.
IMO, the issue is that there are not any other *models* that have widespread awareness or conditioned trust … organizations have almost all of their management processes derived from heirarchical structural assumptions, and there are many cycles of reinforcement going on … job evaluation, compensation schemes, strategy development, objective-setting, accountability and performance management processes come to mind
It WILL become “both/and” over time (See Stan davis work in Chapter 2 and 3 of Future Perfect, from 1987, for a seminal discussion on this specific point), because the compexity and pace of change is too rapid now .. networks are necessary for effective response, and some forms of effectively functioning hierarchy are necessary for decision-making.
It makes necessary people and skills such as “social” and “organizational” architects, rather than big organizational / HR consulting houses selling hierarchical control methods for the knowledge-and-network based work of people in today’s organizations.
i guess I should probably get more active blogging on these issues, and participate more in the conversation, given that I’ve been thinking about and working with them for a while now.
Btw, the “hierarchy to wireacrhy” t-shirt design got rave reviews at Gnomedex.

By: pheloxi

pheloxi — Thu, 14 Jul 2005 23:37:00 +0000

http://www.thingamy.com/ the best kept public secret. some one is creating oxy thingany...uhm...I mean doxy moron (as di for digital). thingamy could be the next ginger! see: the "i was there" factor http://www.gapingvoid.com/Moveable_Type/archives/001716.html

By: frosty

frosty — Thu, 14 Jul 2005 17:24:07 +0000

I've also been working on ways to find, filter and organize information over the last year, including playing around with the tags idea a bit. There are many areas in which tags and tag-like organization offer huge potential, in my opinion largely because it gets a little closer to the way our brains seem to work. And there are other twists on the concept that haven't been explored much publicly, but when I get to public beta... ;-) I don't know if, in the realm of digital information, there would ever be a complete move away from hierarchical organization. I rather imagine nonhierarchical (tags etc) forms taking over an ever larger part of our digital lives where strict organization isn't a true requirement. For example, on sites like Slashdot there are periodic lively debates about Google's approach, particularly with Gmail: "don't file, search." That can be amazingly powerful, but only where 100% accuracy isn't an absolute imperative. You might want your e-mail working that way, but I don't think you want your banking system working that way. Of course there's also the data-mining angle, where any new way to find connections is very welcome, and is going to generally be overlaid on a defined structure (set of structures) anyway. I think we've realized by now (or should have) that with the flood of digital documents, information, etc., the/old/file/structure/concept isn't going to cut it. At least not from the user's point of view. And yes, soul-crushing hierarchies in business are stupid, but that has more to do with the egos and insecurities of people "in charge of" other people than with any serious attempt at efficient organization. I highly recommend Oliver Sacks to anyone interested in this topic. http://www.oliversacks.com/ Unfortunately, this article has gone to "pay mode" but it's great: http://www.nybooks.com/articles/16882

By: Alex

Alex — Thu, 14 Jul 2005 17:13:40 +0000

I couldn’t agree more with Valdis… free form tags are useful, but so are hierarchies.
Both Sig or you talk about tagging in such an OR oriented way. Which is very provocative, which is why you’re doing it I suppose.
This not a binary problem with a binary solution, it is not black and white, it is not cut and dried. Each has their place.
Why can’t Hierarchy AND Tagging they live side by side? What makes you think one is so useful it can ‘completely’ replace the other?
You use toothfloss, but surely that doesn’t mean you should stop brushing your teeth?

By: Keith Handy

Keith Handy — Thu, 14 Jul 2005 05:21:58 +0000

Hierarchies can be directly translated into linear lists — start with the first large branch, then the first smaller branch, and keep going like this until you’ve exhausted all the sub-branches of any one larger branch. This will always be useful for anal-retentives like myself who want to make sure we haven’t missed anything.
A free-floating web of association, on the other hand — like the human brain — might have “dark areas” that we never get around to. Is it safe to assume that if we don’t happen to find these, they weren’t important enough in the first place?

By: hugh macleod

hugh macleod — Thu, 14 Jul 2005 01:23:52 +0000

“jumping back and forth from one to the other until we get there…”
That’s not how it worked at ANY of the companies I worked for, back in my Madison Ave days

By: Valdis

Valdis — Wed, 13 Jul 2005 23:42:46 +0000

It is not an OR problem… hierarchy OR something else — like network
It is an AND situation… hierarchy AND network — prescribed AND emergent…
We traverse/surf whatever paths get us to the solution, sometimes hierarchy, sometimes horizontal/diagonal associations[network], and sometimes both, jumping back and forth from one to the other until we get there.