A New Homepage for Brandwatch Analytics
By Jane ZupanJun 4
Published December 21st 2015
This holiday season, you may have noticed a few more stars have been added to the night sky.
Every year, we try and do something a little bit more creative than simply making a card (although we do that too).
Last year, we asked you all to make a pom pom using your Twitter avatar – it sounds complicated, but really it was beautifully simple – and this year we’ve created a whole galaxy of constellations from your Twitter handles.
The campaign, ‘Star Quality’, has resulted in hundreds of you making a constellation for yourself, a friend, or simply someone you admire.
But how does a concept turn into a reality? And how did we allow an algorithm to design the holiday card this year?
We thought we’d go behind the scenes to ask our Creative Coder Matt Pearson what goes into making a holiday campaign to inspire and delight.
Back in another life, before I joined the Brandwatch design team, I wrote a book called “Generative Art”. It’s an important subject to me, and is a keystone in my approach to design.
This is why, when the time came around to think of the Brandwatch holiday card again, we thought we might try pulling the generative art cracker to see if a festive design might simply fall out.
And it did.
Rather easily, I’m pleased to say. The only consequence of the method being that this year we have a card design for which no member of the design team, me included, can take credit.
This year the designer was an algorithm.
Generative art, to try to condense a rich and nebulous subject into a pithy sentence, is taking an algorithmic approach to exploring a design space.
Or, to allow a machine process to make some of the artistic decisions, rather than a human.
The concept can be upsetting to some of the more precious artistic souls, who don’t like to believe their human faculties can be easily automated.
Which is partly why I like it so much.
It raises a lot of difficult questions about what makes us human, what is art, what is authorship.
Difficult questions, like which of the team designed the Brandwatch holiday card this year?
The constellation on the front of the card was derived from the word “Brandwatch” and the pixels of the Brandwatch Twitter avatar.
The word, with a bit of processing, is used to derive a “seed” number. The avatar, again with a bit of number crunching, gives us a colour palette.
We then use that number and palette as the initial conditions for a pseudo-random animated sequence – a swirling starscape – which eventually settles into a fixed state – our constellation design.
I refer to the sequence as pseudo-random, because randomness is something computers aren’t very good at.
Being creatures of purest logic, they cannot simply pick a number out the air, they need to be told what to pick. So any “random” number is actually the (entirely deterministic) result of a mathematical calculation.
It is this machine foible that means we can always replay a “random” sequence by resetting the random “seed” at the start of the process.
The seed, which is the number derived from the Twitter handle, determines the order of random decisions within the animated sequence.
You can test the process for yourself here, by re-running the animation on the “@Brandwatch” Twitter handle. Or you can try your own handle, or a friend’s, or a celebrity’s, to see what they might produce.
Every handle gives a different seed, so every constellation, theoretically, will be unique. Meaning the number of possible designs is, essentially, infinite.
No. Not really.
The concept still came from a human brain (Katja’s), the site still had to be designed (by Chris), and the printed card laid out (by David).
It’s only the constellation form at its heart that’s generative. And even that still contains a human element.
Generative processes are always a collaboration, between human and machine. After all, someone has to write the logic in the first place.
Although, on a design level, the human part of the collaboration is closer to curator than creator. The machine does the work, while the human does the “hovering art director” thing over their metaphorical shoulder.
This division of labour is quite sensible. We allow the machines to do what they’re good at – complex calculation and mindless repetition – while the human plays to their strengths too, the “that looks nice” part.
Given the advantage of being of the same species, humans have a much easier job of deciding what might be aesthetically pleasing to a fellow human.
— Brandwatch (@Brandwatch) December 15, 2015
Appreciation of human aesthetics is, like randomness, another thing machines find rather difficult, and can only really be learned by studying and replicating human responses.
We don’t have the machine intelligence to trust this to an algorithm quite yet.
A big thanks to Matt for talking to us. Now, have you made your star yet?