Two Tales of Taxonomy Ignorance
Maish Nichani has an absolutely brilliant post dissecting the taxonomy issues around the reorganisation of TV channel listings on local provider Starhub.
Previously Starhub had created a flat numbered channel-list – within which it appears, they had originally done a preliminary clustering on the original creation of the list. But as channels are added, related channels obviously get separated as they are added at different times to wherever the current end of the list might be.
So Starhub quite sensibly “did a Dewey”, creating ranges for different genres of content. Seems like a sensible move in principle, but it was not well executed. As Maish clearly points out, (a) they don’t seem to understand how to create categories that create the ability to predict where content will be found – the current genres are full of overlaps and ambiguities; (b) they thought only of creating room for their own internal expansion needs, and do not seem to have thought through at all the issues for customer usability in browsing or remembering channels, nor the immense opportunity for discovery and new sales created when you categorise in a way that enables browsing and discovery.
And what makes Maish’s post brilliant is that he has gone beyond the critique and thought through the mechanics of how Starhub could have done this much, much better. If I was Starhub I’d hire him to fix this.
It’s amazing how taxonomy awareness can make or break usability design for the navigation of large amounts of content. Maish also references Mike Cane’s ruthless take-down of the abysmal tagging and metadata support for Apple iPad’s e-books on iTunes.
We might argue that Starhub can be given a bit more latitude if even a fabled user-design company like Apple can make basic errors through lack of taxonomy expertise. But it’s not up to us to forgive. The market – in the shape of customers – decides who it will forgive, and wherever there is competition, companies that make their products easier to find and buy will always win out. There’s been a lot of hot air spouted about the promise of simple tagging versus taxonomy design, or of how they can be complemented. Apple seem to have taken the easy “tags are all we need” route (very poor discipline in the use of metatags). Thinking about it, an online product catalogue is a great way to test for the monetary value of using taxonomy principles. And using controlled experiments in the marketplace might actually help taxonomy work get the understanding and appreciation it deserves.
5 Comments so far
- Patrick
Well there was the Raw Sugar approach of creating a simple structure to put tags into, so that some association of meaning is created.. I have started thinking that captions might be better than tags, but that solution would serve navigation better than keyword search. Another approach would be for a system to infer a relationship between objects based on context, origin or usage, and thereby establish relationships between tags. But I haven’t seen a great example of this!
- Don Sheridan
I just got an email that states MISQ stopped updating their keywords taxonomy in the early 2000s as keywords according to the then editor were going out of fashion. Apparently search engines were to take over! http://www.misq.org/roadmap/codes.html
Don -
Knock me over with a feather!!!
That’s the first time I’ve seen the Henri Barki/MISQ work referenced.
I stumbled (brute force going through the stacks in a library) across it in pre-Internet days & thought it might have promise.
I asked Barki if he intended to continue the work—I consider terminology “drift” a significant challenge that is not well addressed with classic fixed taxonomies—and his interests had moved on… a common habit of academics. Discover the high ground & then move on to the next mountain top. Don’t get involved with the messy details of implementation.
So much for that promising idea…
I can vouch for the perceived power of search engines over keywords.
In 1993 I was working on a naming standards enforcement tool (getting off the mainframe where it had been working for 10 years) and down to the desktop.
And then this thing called the Internet appeared out of nowhere. And there were these things called “search engines”... that were going to be so powerful that “well formed” names wouldn’t be needed.
Yeah, right.
The developer & I looked at each other, agreed the idea of super-duper search engines was idiotic… & tossed our product in the the trash.
Don’t get me wrong here… Google is wonderful. But it works on documents written by humans for humans. People write web pages to be found. Plus there are 100s of millions of documents to find… so when was the last time you went below the fold on a Google search?
But there’s a whole other universe inside the company where documents are NOT written for human consumption & are therefore far more difficult to find. Inside the company you need to search for “social security number” and be nudged to also search for SSN, TIN, EIN, SIN, emplID, emp_id, tax_id, SOC-SEC-NO, soc_sec_nbr…
Bottom line… keywords do have a place in the search equation.
- David
- Don Sheridan
I just checked the price of a simple taxonomy from http://www.taxonomywarehouse.com and its nontrivial. Agree the ‘call it what you will’ and the search engine will find it is current practice. Folksonomies are orphans looking for their taxonomic parents?
Page 1 of 1 pages
Comment Guidelines: Basic XHTML is allowed (<strong>, <em>, <a>) Line breaks and paragraphs are automatically generated. URLs are automatically converted into links.
Patrick -
I’ve been tagging (in a long dead Mac contact management product) since 1990… but ONLY as a single user. Long before I hit a peak of 800 tags (aka keywords) it was very evident more robust management capabilities were needed.
Example: the tool allows me to put “SaaS” and “Cloud Computing” in as separate, unrelated tags… but in my own mind (when it’s actually turned on) they are direct synonyms.
A major benefit from my experience is that tags do NOT require any sort of hierarchical/taxonomic structure to be effective.
As far as I’ve seen (it’s a big world, perhaps there’s a breakthrough solution out there...?) an inherent limitation of tags/keywords—if used beyond single user as I do—is the implicit assumption that these strings actually mean the same thing to me as to you. Have you seen anyone push beyond this inherent limitation of tags?
- David
Posted on April 17, 2010 at 11:02 AM | Comment permalink