[teampractices] Good article on why we don’t like what we struggle to categorize

Discussion:

Grace Gellerman

2016-06-09 20:48:22 UTC

http://www.nytimes.com/2016/05/29/opinion/sunday/the-psychology-of-genre.html?mabReward=CTM&action=click&pgtype=Homepage&region=CColumn&module=Recommendation&src=rechp&WT.nav=RecEngine&_r=2

Favorite quotes from the article:

"This âcategorical perception,â as itâs called, is not an innocent process:
What we think weâre looking at can alter what we actually see. More
broadly, when we put things into a category, research has found, they
actually become more alike in our minds."

âSimilarity serves as a basis for the classification of objects,â wrote the
noted psychologist Amos Tversky, âbut it is also influenced by the adopted
classification.â The flip side holds: Things we might have viewed as more
similar become, when placed into two distinct categories, more different."

"Categorization affects not just how we perceive things, but how we feel
about them. When we like something, we seem to want to break it down into
further categories, away from the so-called basic level"

"When we struggle to categorize something, we like it less."

Kevin Smith

2016-06-14 17:17:58 UTC

Permalink

Back in 1990, things were almost always put in hierarchical taxonomies. The
idea of "tags" and non-hierarchical categories wasn't really a thing,
partly due to limitations of technology. The Dewey Decimal System (for
books) and biological taxonomies ruled the day. And it was really
frustrating, because so many things really can't fit into a single bucket.
Even today, it seems like rigid taxonomies remain overused.

I'm not sure where I'm going with that, but it seemed relevant somehow, and
I needed to rant.

Kevin Smith
Agile Coach, Wikimedia Foundation

Post by Grace Gellerman
http://www.nytimes.com/2016/05/29/opinion/sunday/the-psychology-of-genre.html?mabReward=CTM&action=click&pgtype=Homepage&region=CColumn&module=Recommendation&src=rechp&WT.nav=RecEngine&_r=2
"This âcategorical perception,â as itâs called, is not an innocent
process: What we think weâre looking at can alter what we actually see.
More broadly, when we put things into a category, research has found, they
actually become more alike in our minds."
the noted psychologist Amos Tversky, âbut it is also influenced by the
adopted classification.â The flip side holds: Things we might have viewed
as more similar become, when placed into two distinct categories, more
different."
"Categorization affects not just how we perceive things, but how we feel
about them. When we like something, we seem to want to break it down into
further categories, away from the so-called basic level"
"When we struggle to categorize something, we like it less."
_______________________________________________
teampractices mailing list
https://lists.wikimedia.org/mailman/listinfo/teampractices

Mukunda Modell

2016-06-14 18:05:37 UTC

Permalink

This is really an interesting topic to me personally because I spent a year
at deviantART working on a faceted ontology to classify art works by facets
such as Form, Medium, Technique and Genre. It was a really hard technical
problem to solve and we never really finished that work before deviantART
burned through their budget and the project got put on the back burner.

I really enjoyed the article.

A key insight that we were keenly aware of, and struggled with at
deviantART is called out in the article:

"Even though a computer might struggle to see a strong difference between
two forms of music, humans seem intent on finding one. As an example of two
genres where the music sounds quite alike but isnât, Mr. McDonald cites
âvegan straight edgeâ versus âhatecore.â If you didnât know the categories,
you might have trouble telling them apart â but meet the two groups of fans
and the difference would be pretty apparent.

This also tells us that, very often, these distinctions are for social
purposes: People label music, music labels people. As with the rainbow, a
country song and a rock song might be musically closer than two songs
*within* either genre, but in our minds, the genre threshold takes
precedence."

We weren't classifying music as that is the one and only art form which is
excluded from deviantART. Nonetheless, we had to deal with the social and
emotional meanings of genre and that's where it got very difficult to build
a system and structure that captures the subjective aspects which are very
personal, subtle but incredibly important to the people who identify with a
given genre.

I think it's still a largely unsolved problem in computer science. There
are a lot of unstructured tag-based classification systems that achieve
acceptable levels of usability but I don't know of any that do a good job
of capturing the essence of a genre in a meaningful way. Art, and
especially genre, defies classification and it's difficult to come up with
a good solution using a scientific approach.

By now I suspect that the folks working on Wikidata probably have better
ideas and solutions than anything I came up with when I was working on it.

Thanks to Grace for sharing the article and Kevin for sharing your thoughts
on it.

Post by Kevin Smith
Back in 1990, things were almost always put in hierarchical taxonomies.
The idea of "tags" and non-hierarchical categories wasn't really a thing,
partly due to limitations of technology. The Dewey Decimal System (for
books) and biological taxonomies ruled the day. And it was really
frustrating, because so many things really can't fit into a single bucket.
Even today, it seems like rigid taxonomies remain overused.
I'm not sure where I'm going with that, but it seemed relevant somehow,
and I needed to rant.
Kevin Smith
Agile Coach, Wikimedia Foundation

_______________________________________________
teampractices mailing list
https://lists.wikimedia.org/mailman/listinfo/teampractices

Kevin Smith

2016-06-14 18:45:29 UTC

Permalink

That's true that even coming up with the possible tags can be
controversial. I sometimes describe smells in terms of colors ("This smells
bluish"), which completely confuses my wife.

It's trivial for computers to manage the tags. And it's trivial for humans
to spontaneously add a million tags as they pop into their head. Creating a
clean system of tags is a difficult human problem.

The article also mentioned that people want to deeply classify things they
love or feel strongly about. Which is not surprising. To me, for the most
part, beer is just "beer"[1]. To beer fans, there are a thousand subtle
variations. Meanwhile, I can distinguish at least dozens, if not hundreds,
of types of boardgames. (Thinking in terms of tags, not taxonomies.)

[1] I mostly put beers into 2 categories: "eww", and "OMG NO get this out
of my mouth!"

Kevin Smith
Agile Coach, Wikimedia Foundation

Post by Mukunda Modell
This is really an interesting topic to me personally because I spent a
year at deviantART working on a faceted ontology to classify art works by
facets such as Form, Medium, Technique and Genre. It was a really hard
technical problem to solve and we never really finished that work before
deviantART burned through their budget and the project got put on the back
burner.
I really enjoyed the article.
A key insight that we were keenly aware of, and struggled with at
"Even though a computer might struggle to see a strong difference between
two forms of music, humans seem intent on finding one. As an example of two
genres where the music sounds quite alike but isnât, Mr. McDonald cites
âvegan straight edgeâ versus âhatecore.â If you didnât know the categories,
you might have trouble telling them apart â but meet the two groups of fans
and the difference would be pretty apparent.
This also tells us that, very often, these distinctions are for social
purposes: People label music, music labels people. As with the rainbow, a
country song and a rock song might be musically closer than two songs
*within* either genre, but in our minds, the genre threshold takes
precedence."
We weren't classifying music as that is the one and only art form which is
excluded from deviantART. Nonetheless, we had to deal with the social and
emotional meanings of genre and that's where it got very difficult to build
a system and structure that captures the subjective aspects which are very
personal, subtle but incredibly important to the people who identify with a
given genre.
I think it's still a largely unsolved problem in computer science. There
are a lot of unstructured tag-based classification systems that achieve
acceptable levels of usability but I don't know of any that do a good job
of capturing the essence of a genre in a meaningful way. Art, and
especially genre, defies classification and it's difficult to come up with
a good solution using a scientific approach.
By now I suspect that the folks working on Wikidata probably have better
ideas and solutions than anything I came up with when I was working on it.
Thanks to Grace for sharing the article and Kevin for sharing your
thoughts on it.