The word curation has become a bit precious in the last decade. Law libraries have been curating collections – making selections, choosing one thing and omitting another – forever. This recent post by the founder of Medium was very positive to see. He contrasts what he calls the relational and transactional approaches to consuming information. It’s important because the transactional – disintermediated, often algorithmic – approach treats all objects as equal. Faux curation.
The post is mostly about changes Medium, the company, is making to its app. The whole post is interesting but the meat starts at the heading “Relational media“:
Think of (or imagine, if you’re not old enough) when you got your morning newspaper or your favorite magazine and read articles because they were in that newspaper or magazine. … [W]hat you read was very heavily influenced by where it came from. You picked the source first and then you picked the articles. You built affinity and trust for the sources ….
– Ev Williams, Toward a More Relational Medium
The benefit of dedicated news sources is that they select content. Implicit in this selection is bias, so selectivity isn’t always optimal. Law library collections can be selective based on use patterns or jurisdiction that are less biased (an Oregon law library focused on Oregon – and not Montana – law makes sense) but still reflect choice.
As with everything, curation has a cost. A law library doesn’t self-generate and self-maintain. You pay people to make selections – and deselections – and you get the benefit of those choices. A lawyer may not agree with the choices, in which case they can purchase additional content to supplement the curated collection, but someone else is doing the knowledge work to get there.
At the other end is the generated collection. The curation cost still exists. But the modern information approach is to shift it to the reader or researcher. The transactional approach, again from Ev Williams:
In contrast, it is much more common today to choose to read something before you even know where it came from, based on the headline. Sophisticated readers will pay attention to the source, but there are too many (down to millions of individuals) to be familiar with or have a point of view on.
Navigating from homepage to homepage — the digital equivalent of picking up a magazine — feels very antiquated and inefficient when the headlines can all be aggregated and presented to you, personalized. Or, depending on your need, when you can search for a specific topic and be led directly to a reasonable set of answers [from a web search]Ev Williams, Toward a More Relational Medium
Law libraries are small. The legal profession is a tiny information audience. We curate for the tens, the hundreds, the thousands. Rarely do we curate for even the tens of thousands. A newspaper can curate for hundreds of thousands. A web site, for millions.
Key to the transactional problem is the personalization. In a law library, a staff person can know what a particular lawyer is watching for, their practice area and potentially specific cases. This doesn’t scale, though. A single law librarian can only juggle so much personalized attention.
So we, the world, have tried to save on curation costs. We’ve offloaded some of that work to technology. And it hasn’t done a great job. But it has created the perception that it is doing a good job for anyone who is not curious enough to look.
Perhaps the most frequent example for me is this little one, from the Google News app. I’ve written about how I use Google News for serendipitous discovery and Flipboard for dedicated sources. They are both very weak but they are a step above random discovery and the work that entails. When I fire up Google News, it promises me the Top 5 stories personalized for me:
The Top 5 Stories on Google News are frequently only 3 or 4. It could be because they initially want to show a story but I’ve put the source in my blocked list. That begs the question: why would the algorithm not know to remove blocked list content from my top stories?
It’s not entirely clear how Google News chooses content. It’s not like they offer a collection development policy; they certainly have zero standards in selecting sources. In fact, they select a lot of sources that do not seem reliable or that it would take a lot of time to confirm their reliability. I have a long block list.
My best guess is that they use some sort of taxonomy with their algorithm that lacks any depth. If I look at any news article that relates to game then I must be interested in all articles that involve game, even though I might only be interested in certain types of games, or on certain technology platforms. I have zero interest in wireless phones, so the Xiaomi phone camera article is probably surfaced due to a keyword like technology.
Study 1 tested these propositions with a news-aggregator Website that was either personalized (system-tailored), customized (user-tailored), or neither. Power users rated content quality higher when it had a customizable interface, whereas nonpower users preferred personalized content.
Personalization versus Customization: The Importance of Agency, Privacy, and Power Usage,
Human Communication Research 36 (2010) 298–322 doi:10.1111/j.1468-2958.2010.01377.x
We can see a lot of personalization attempted in web search. Results become localized, which is helpful if you’re looking for a local resource (restaurant, hospital) but less helpful otherwise. This is particularly true if your locale is based on where your ISP pops up, rather than your physical location. This sort of personalization can create filters for researchers who are funneled into a narrower universe based on previous, unrelated research.
I had to laugh at this Youtube channel created based on my musical interests. I am, in fact, a frequent watcher of orchestral Youtube performances. But it’s ironic that a Violin channel would contain only winds.
To that end, I have increasingly tended towards source-based selection. That is how I use Flipboard, to make specific source selections and then read what they provide. There are three levels of curation here and they are not all helpful.
The first is to make the source selection, which is entirely within my control. That suits me fine. The only caveat is that sources come in flavors. Flipboard offers the publisher’s source but can also offer user-generated collections of the publisher’s content. Youtube offers something similar: you can follow Taylor Swift, or you can follow the Taylor Swift topic. The user-generated collection is a grey area, relying on that curator to maintain it and also to act – like a law librarian – to curate within specific collection parameters.
At the next level, the source specifically limits the content that makes it into Flipboard. Unlike a newspaper, where I can read cover to cover, a channel online is only offering a content subset. From what I can tell, it tends to be general interest media rather than localized, which suits me fine.
The interesting contrast within Flipboard is the last curation step. There is a For You tile that aggregates content from all of your sources within the app. Unfortunately, it represents neither all the content that is aggregated on the app nor even a balanced selection. I didn’t notice this until I subscribed to a high volume source – International Business Times – and a low volume source – MediaRedef.
I chose the former for breadth, if not quality. The latter is one of my absolute favorite channels. It is highly selective and the curators pick related articles from across the web, reaching far beyond where my filter bubble would normally go.
Flipboard uses some sort of mathematical formula to create For You. If the news source is fairly typical, about 50% of the content appears in For You. A lower volume source like MediaRedef will only have about 20%-25% of its content appear there. My percentages are based on going back to the source channel and actually reading all the content.
On the other hand, IBTimes generates a ton of content. And the For You tile is overwhelmed. It’s not clear to me what percentage, but I can have page after page of content and it is all from IBTimes. The flood defeats the purpose of an aggregate source. I tend to avoid For You now because I know it will take me forever to even see the subset of content from other sources.
Our journey towards transactional sources is obviously still ongoing. It’s hard to see how this is a quality journey for anyone who isn’t actively looking beyond the headline to determine the source. I don’t like the term power user, because some people are just busy. But they may be trusting the transactional approach in the same way they used to trust their newspaper, without fully thinking about how the sourcing has changed.
I was glad to see the change coming to Medium. I’m not a Medium follower or user – I’ve mostly read stuff on their platform after it’s highlighted on MediaRedef. Similarly, I’m hopeful that we’ll see changes in other news sources like Twitter and Facebook. These environments lack curation and the way in which people use them amplifies the likelihood that someone will interact without actually reading the item or determining a source. It’s a hopeful sign that there are publishers and platforms out there who are trying to rethink how to do aggregation and curation.