Lessons and big-a** failures

In my last post I walked through the challenge of curating content (digital music) on a tight schedule. I responded by developing a content strategy with a taxonomy and system of ‘likes’, to build up useful metrics and deliver accordingly. My content strategy worked a treat, until cops shut us down. Here are are the rough-patches along the way, and the wider CS lessons learned.

People Don’t Like What They Ask For

Curation is a skill. Think about WIKIs. Without a strong curator/strategist keeping things going you often get content soup, and even the people who created it don’t like it. Deliver according to the metrics! Many organisations I’ve seen have released this or that content on a hunch, or because this or that manager used their weight to push it through.  The result is never as good as when a combination metrics and good content professionals are used. In my applied content science experiment, I used my DJing taxonomy and generally stuck with my model of how the content should be structured. However, later in the evening, I started opening up user-generated content, aka, straight-up “requests”. It was ok at first, but soon the users started taking over and crowding out the curated content. Result – they hated it. Individuals may have liked this or that song, but when the structure and coherency of the taxonomy was gone, the genres started mixing and content that wasn’t high quality started surfacing to the whole user group. So, I learned giving people exactly what they wanted can create a mess.

Structure and control feedback channels

Organisations sometimes take on more channels than they’re actually ready for, without a strong underlying strategy and plan. The result is a lot of messy conversations where the content and its provider end up looking worse for their half-assed presence. When I stopped using my taxonomy board and let the users submit directly, feedback become overwhelming and, worse, contradictory.

Planning for reuse is fun and effective

When planning content, whether content marketing or technical manuals, structuring for modules and reusable assets as early as possible can get you more bang for your buck.  We often forget the simple idea that a bit more up-front work in your writing and structuring will smooth later leveraging of assets later. When I came up with my approach for the party, blogging the outcome was always in the back of my mind. In the true spirit of designing reusable assets, I crowd-sourced not only the evening’s content list and structure, but got them to actually be the assets so I could hopefully teach a fun lesson later on my blog. After the party, I realised I didn’t have a DJ Facebook page, and of course the assets I’d created for the blog could be reused in a new, previously unconsidered context. This is living the dream of content reuse. This was all happening with the Content Strategy Google Groups discussion “To Tech Or Not To Tech” going on in the background (related discussions in LinkedIn and Google+) where we were discussing how much CSs can or should relate with the deeper, more ‘technical’ aspects of what we do.  My believe is we can and should, we have to discuss such things in a way people can relate to.  And serve cocktails.

Proven again: the best tech is humanist

In all my workshops and classes I try to get people to approach new fangled techniques and technologies with from a “humanist” perspective. Thinking of ourselves first as people, then as users and technologists. In my opinion, most of what a content professional needs to know about content technology is a simple 1-2 punch:

1. Make semantic meaning explicit: Make content structured and meaningful through explicit semantics and metadata (if you don’t know what this means, ask, as it was 3 chapters in the book).  This structure and meaning makes it so that computers can understand content in a way that’s more similar to the way people do.
2. Design experiences leveraging semantics: When computers can understand it more, they can create meaningful and engaging experiences with the content downstream.

When I teach semantics, structure, XML, adaptive content, and DITA, I try to explain them not as new things, but as the common interchange languages between humans and systems. Life is structured and semantic, and our brains operate naturally in taxonomies, building relationships and narrative on top of them. They exist in the digital realm because they mimic our natural understanding of the world. When I explain content technology from the human-centric perspective, I get a lot further than a decade ago when I started off lessons with “XML and HTML are based on their predecessor language: SGML…”. Snore!!

…For example, taxonomic metadata is innately human

We don’t remember every song we hear and tuck it away for instant recall (if only!).  We group by genre, speed, rhythm, culture of origin, etc.. We explore them in our minds by traversing the indices. It’s not a skill.  We don’t have to think consciously “I’ll store this in my brain under ‘Easy Listening’,” that’s just how we work. (Remind me to curate some links to TED talks about this!) Think of when you when you can’t quite remember a song.  What do you remember? Some of the content, like melody or snippets of lyrics.  But aren’t you always able to rattle off something like this? “It’s a medium-fast song, dance-y, with a sort jazz feeling, and there’s a guitar part that’s quite rock-y and then the guy goes ‘Riiings! Pearls!’.  It’s a rock classic, man!” So we have just referenced the categories in your internal taxonomy:

• Mid-tempo songs
  • Dance
• Jazz
• Guitar music
• Rock
  • Classic rock
• Male singers

And ta-dah!  Led Zeppelin’s “How Many More Times”.  Implement this in computer code and ta-dah: Last.FM, Pandora.com, Spotify! These are systems that use taxonomy to tell you music you like before you knew you liked it! Note – it’s often easier to remember things by their taxonomical categorisation than something in the content, like, say the lyrics, or the f***ing title itself!

Amazon.com: always leading the taxonomy pack

On the same point, just last week I was considering a recent purchase. I thought, “I bought an Xbox Kinect on Amazon and it’s going to arrive soon. What games can I get now?” I had opened up the “Kinect” part of a product taxonomy – a category of games that previously I’d closed my mind to – and was starting to explore it. Amazon, in their taxonomy-driven brilliance, emailed me just this morning with the subject line, “Amazon.co.uk recommends “Kinect Sports: Ultimate Collection”.  Awesome.

More?

I’m hungry for more examples of this “real-life” content strategy applied, so please do submit yours in the comments!

PS – For sound tech geeks …

This bit is just for those who care what hardware and software I was using. I was spinning Mp3s in open-source Mixxx. It’s good, but not great, and I think 20k songs was too much for it as after a few sessions my database file died and I needed to re-index everything. I was running the whole lot through a Yamaha home hifi amp (RX-V795RDS) 5 x 125 watts RMS to 2 channels 8ohms, running both A and B outs to 4 speakers (didn’t use the surround outs). I supplemented that with a 200w, 10-inch Klipsch subwoofer and was running out to Tannoy Reveal monitors (crappy as studio monitors, decent as speakers). The interesting bit I think was that I was running 3 sound cards. One was the one built into my (kinda crappy) Asus laptop and two USB soundcards (both by Creative) to run separate outs to my cue mix and mains. Worked great!