(blows off the dust since the last entry)
(Life trumped blogging; my first child was born in March)
Just before I went into the parent tunnel, which is awesome by the by, I attended a seminar conducted by Niels Windfeld Lund, General Manager of the World Opera.
Not my usual event. But music’s always been a passion for me, and I performed a lot as a kid – lots of trumpet, both the sort of american wind orchestra stuff (seated and marching…yes, a band geek) and some jazz, a little bit of drums. These days I plink around on an acoustic bass, badly, but well enough that I’ll be able to sing lullabyes to my newborn. Starting to play again made me realize how much of music is a conversation, just like science is a conversation (well, an argument). And so this world opera thing seemed like an interesting way to come at the problem from a field that is semantically a world away from science, but in design space is remarkably similar.
This was the second of two recent musical events for me that bear on open collaborative science. The first one we can draw a lesson from is the World Choir. It’s a collaborative choir of more than 2000 people and got tons of press about how it was a collaboration breakthrough, and was designed as an asynchronous request for videos, with a ton of post-processing to stitch them into a single video.
Then there’s the World Opera, which is all about actually performing an opera live with the performers in multiple cities around the world (it needs better marketing help – the first sentence in the website is “You may not have heard about World Opera”). There’s tons of dimensionality baked into that idea from the start. Niels got it funded in his northern Norwegian home of Tromsø after first pitching telemedicine, which has the same fundamental requirements as online opera: big fiber, low latency, great audio/video capability, and the ability to do meaningful real time interaction with remote sites. Surgery or string rehearsal, you have to be able to replicate an intense real life experience. The government apparently preferred opera, which is both wonderful and improbable to me.
They wrestled, or are wrestling, with technical and existential questions. How to use the inevitable delay between performers, turning it into something like the acoustics of a conference hall. Whether or not to use a live conductor as a distinct part of the performance, or to use a metronome, or something in between. Basic stuff, like how to practice together but apart. They are able to do it, but it takes much more work than it would in a regular opera.
It’s hard to get a group on the one. It’s hard when you’re all in the same place. That’s why good labs or departments (or startups) have regular journal clubs, regular lunch sessions, coffee machines that require a fair amount of time to prepare a drink. It helps create that extra time where the individuals involved fall into a rhythm together. Eric Schadt has called it the “clock gene” of a good lab. And it’s been hard to virtually create in the sciences.
My gut is that we have the two musical performances mixed up. A lot of what we mean by open science is the choir:we’ll do crowdsourced data collection, we’ll see a surge of data from impassioned observers into online groups like Sage Bionetworks, but that data will have to be painstakingly synced and organized before we get a beautiful model.
Real collaborative science is going to be hard, like the opera, because it’ll be hard to get on the one. Big questions, both technical and epistemic, have to get answered.
Collaborative opera is totally disruptive to regular opera. It will be resisted, its flaws will be evident with no post-processing to make it shiny. It’s not just sound, it’s a story, it’s acting, it’s interaction between the performers themselves and the audience. It’s going to suck compared to a purist’s opera – at first.
But as the group learns, and they will, it’ll suck a lot less. Then it’ll be really, really good. Incremental innovation will smooth a ton of edges. The performers will figure out if they want an avatar or a conductor. They’ll get used to the latency, and “hear” it when they play together. There’ll be a disrupting tech, probably made by a frustrated musician, that makes some vital but boring process suddenly either a) easier or b) stable or both. Online opera will become a vital part of opera.
These problems, this inherent resistance (in the electrical sense, not the political or incentive sense) is the sort of thing we have to get used to in open science. We can run a bunch of virtual choirs – that’s what 23andme is doing, and I’m a customer. But our infrastructure, and our design thinking, and most of all our expectation, has to support opera, because it is, like science, hard.