The best time to plant a tree was 20 years ago. The second best time is now. - Chinese proverb (supposedly - but lots of difficult-to-source good quotes seem to be attributed to a Chinese proverb) Building the digital commons is a tricky thing. It’s not all that different from doing a startup. If you’re at the right time, people are telling you you’re crazy. People have failed at your task. Big companies have yanked projects. The market isn’t ready. But that’s when you have to build. Because once a market gels, inserting competition from the open is very difficult. See: Diaspora. It is far better to ensure from the get-go that credible competition exists from the open source side, so that the entire market is affected by the credible threat that customers can bolt to open source, open data, open content, open systems. I would argue we’re at this point in the personal data revolution. We’re awash in devices that collect data, which will look paleolithic in five years. We’re on the edge of ubiquitous electronic health records for everyone, which will actually start to be useful in five years. We’re already at the $1000 genome, which will be $100 in five years. It’s easy to say, just wait. After all, Google exited the health records business. None of the big players in personal genomics or health has more than 150,000 total users (I don’t have a link, but this is what I’ve been told by management at a variety of companies) and a tiny fraction of those actually log in each month. Don’t plant the tree yet. Wait til the ground’s ready. Now is the time to build a commons out of our personal data. Not in five years. In five years, the walled gardens will be so big most people won’t notice the walls. But walled gardens don’t lead to the kind of innovation we need. They lead to monopolistic, bullying behavior when the walled garden needs to hit quarterly results. It’s not about shutting down the companies. It’s about changing the overall composition of the market’s ecosystem through openness. Openness means that the non open players have to treat us better, because then we have an option to simply leave. Openness in web browsers and web servers changes the closed browsers and closed servers, in ways that we don’t have in social and mobile (despite the promise of android, it’s faux-pen, not open). Now’s the time to plant the trees in personal health data.

Lee Dirks died yesterday with his wife in a car accident in Peru. They leave two daughters behind. Lee worked in Microsoft Research, where he was a large force in the open access world. He was also a very good friend. Someone I trusted to talk to in confidence. Someone I called the first day I changed jobs. When his name was on the agenda, flying twelve hours in a cramped plane to a mediocre hotel in an impossible time zone was a lot more fun. We’d have 30 minute work calls that ran 80 minutes to talk about balancing fatherhood and the travel schedules we shared. We drank together, cussed out our favorite sports teams together, complained about mediocre mexican food together. We sent each other pictures of our kids. And we’ve spent more than one night talking about some awesome place one of us heard about. Nowhere near the mediocre hotel, just a place you can reach in an poorly maintained cab, on a road with crumbling features, that great local version of the overrun tourist attraction with no crowds and the best food stall. Nearly every time that ride is ok. Sometimes something goes wrong but it always ends as a minor thing, a sideswipe, a funny story to tell at the next bar night. But I guess it doesn’t always. This ending sucks. Lee deserved, more than most I know, to go out fat and happy with a smile on his face. He was a mensch. I’m going to miss him fiercely.

So, big news today in the DTC genomics world was that 23andme has decided to publish an API. The blog announcement optimistically announces that there is “no monopoly on great ideas”and encourages developers to apply for access to write applications to the API. I’ve gotten quite a few requests for my thoughts on the announcement. I’m torn between wanting to praise 23andme and wanting to scream DANGER DANGER DANGER at the top of my lungs. So, I’m going to do both. First, bravo. Publishing an API is smart business move for 23andme, and holds the potential to create a lot of genomics apps on the data they hold. Second, DANGER. Publishing an API is a smart business move. Let me repeat that: it’s a business move. It’s not because they’re nice. There is a very serious difference between “we’ve opened our API to you” and freedom. Developers must apply for permission to make applications. Developers must tell 23andme, before developing, what their apps will do. And there is nothing that prevents the strategic shrinking of an API, or the subtle or not so subtle pressure to turn off applications that compete with core business functions or revenues of 23andme. ’ The fundamental problem is monetization. If you write an awesome app to a closed API, it makes good business sense to give you the Sopranos treatment. You make the app, it gets popular, you get the bust out. Twitter started with a wide open API and is basically phasing it out. Sucks if you invested your time there. Or if you’re like Dalton Caldwell, being strongarmed by Facebook when his app competed with FB’s App Center. So that’s the fundamental thing. Develop to a corporate API knowing full well that if you are successful, there’s a better chance that you wind up Scaltino and not Zuckerberg. And despite all the press that App.net is getting now, remember that social networks achieve network effects before they shut down functions in the API. And once a social network gets to that network effect, it’s very hard to build a competitor from the open. Anyone out there use Diaspora? Didn’t think so. Code carefully, my friends. And choose your monopolies wisely.

Sam Arbesman’s new book, The Half Life of Facts, is coming out soon. I’ve had the pleasure of reading both a galley proof and an advance print, and I can tell you it’s a great book. Sam demonstrates, quantitatively, something we all know implicitly but probably don’t think about very much. Everything we know, or think we know, has an expiration date. For a long time, we “knew” that the sun revolved around the earth. For a long time, we “knew” that atoms were the smallest part of matter. But we learn. And we learn more. And what we “know” decays, over time. The Half Life of Facts is a great read for those who are interested in science, or philosophy of science, or who just need a reminder to have a little less hubris about how smart we are. Because to the future, we’re all phrenologists. Disclosure: Sam works at the Kauffman Foundation, where I am a non-resident Senior Fellow. But I like his book anyway.

I’m at the tail end of a wicked burn of travel and talks, which means that I’ve been pummeled with input (in a good way). The more time spent in airplanes and hotels, the more reading I get done. And the more I give talks, the more I get introduced to people I don’t know, hear others speak on topics I don’t know well, and learn. Because it’s politics season, and I can’t stand televised political coverage, I sought refuge in Future Perfect by Stephen Johnson and Ill Fares the Land byTony Judt, which each in their own way address the gap from which the current political system in the US suffers. Because I’m addicted to William Gibson in all forms - right down to staying at the Mondrian in LA, at least in part because it was the start of Spook Country, Distrust That Particular Flavor. It’s also easy to consume in small parts. Because travel is lonely and makes me melancholy, but also helps me learn and is my job, Dark Star Safari by Paul Theroux. I also just turned 40 and this is his book about turning 60. Because sometimes you need a f*cking drink, Every Night’s a Saturday Night by Bobby Keys. He played sax on everything that was awesome for a long time, and the back story of Exile on Main Street alone is worth the price of the book. Because David Byrne is David Byrne, How Music Works. Bonus points for making me see connections to the re-entry of music into culture to the re-entry of science into culture. Because I’m curious about the new Wachowski film, Cloud Atlas by David Mitchell. Because I like re-reading things, Reamde by Neal Stephenson. Also, because I was going to a conference with a keynote on games, and although I’m not a gamer myself, I’m fascinated with what power they hold. Those are all on my kindle. In real life, I have been lugging Symbol by Steven Bateman orThinking With Type by Ellen Lupton in my rolling bag. I’m experimenting with web redesigns for Consent to Research and these are great books for when I’m worn out, jetlagged, sleepy, and can’t focus on narrative. Symbol is a collection of symbols, just like it sounds, hundreds and hundreds of pages. I tag ones that move me with post its and come back to them later. Thinking with type is like an ethnography of fonts and typefaces. Both books are visually gorgeous and feel good in the hand. Anything I’m missing or should be reading?

(after much complaining, I’ve moved my meme photo to the bottom of the page) I woke up this morning and found that my twitter feed had exploded with the news that GSK is promising to start making individual level data from clinical trials “open” to independent research. There’s not a lot of detail to go on. But there is enough to do some inferential analysis, and a little bit of historical context as well. I’m not surprised to see this. Perhaps that it happened now - I’d have bet on 2013 - but not that it’s happening. Sanofi is doing DataSphere (toll access, sorry for sucking), Eli Lilly is running a Clinical Open Innovation portal, etc. Pharma has already figured out that the creation of precompetitive space will be part of whatever new business models wind up supporting drug discovery. As for the swipes against academia…well, I’ll only speak from my own experience: I have had pretty much all of the top 20 global pharmas call me at some point over the last decade to explore data sharing. I have had maybe three academics call me. There are some good - indeed great - shared academic projects, but the relentless funding requirements and sharp elbows in grantmaking mean that data sharing is hard. There is a business case against sharing, from an economic perspective, in academia - someone else finds a golden ticket in your data, and you lose. That business case does not exist in biology data or in clinical data if you’re a pharmaceutical company. It’s about getting compounds into the patient and getting reimbursed. OK, on to the open part of all this. First, based on what the press release and news reports are saying, we don’t know if the data will be “open” or not - because the press reports don’t mention intellectual property status of the data. So, to the first point - is it open? We don’t know. Honestly, it doesn’t matter very much for a company like GSK, so I’d expect that they will indeed use a CC0 public domain waiver - they did so previously with their Tres Cantos malaria compound set, which is now underpinning open source drug discovery work. I helped, just a little bit, with that project while I was still at CC and I can tell you the hard part was getting GSK to agree to make the data public, not getting a good license onto it. Second, frankly, open from a copyrights or patents perspective probably doesn’t matter very much compared to the other mechanisms of control that are explicitly referenced in the reports. The release intimates that a select panel of judges will review a select panel of applicants and grant access. This points out the failure of “open” definitions to adequately grapple with data in my opinion. It’s easy to meet an open definition with this kind of data but only allow an elite group of scientists in to touch it, using strong norms that say “if you share it, you’re out of the pool forever” which don’t violate the various definitions - because of the obsession with intellectual property as the source of openness. Who will be the judges? Under what terms will they allow access? Will it be a liberal policy - one that says, by default, any reasonable request will be granted, and here’s what reasonable means? Or one that says, by default the answer is no? If GSK wants this to work, they need to go all the way and make it not just legally open, but accessible and usable. I’d like to see the data deposited in Synpase at Sage Bionetworks, as well as at a federal repository like NCBI or EBI. I’d like to see it under a set of terms that specify that any researcher willing to comply with certain terms - not attempt to reidentify, not attempt to harm, perhaps agree not to use the data to bring class action lawsuits (hey, for me, it’d be worth it if we could actually build a map of tox and ADME that worked - sort of a truth and reconciliation commission approach). And I’d like to see the world take a whack at it, not have to apply to have their research ideas judged. Anything less than this will be “half” open. And being half open is like half learning how to break a board with your hand in karate. You don’t break the board. You break your hand. It’d be a shame to have the first big pharma that tries this fail because they didn’t have the guts to go the distance. File under:

Guest post by Carolina Rossini (@carolinarossini)

My TED talk went up yesterday. And whatever snark you hear about TED, or the TED brand, know this. The conferences are amazing in person. I go to to conferences. All the time. There is nothing like TED. Part of it is the seriousness of it - no laptops, no tablets, no phones, except in the very back of the auditorium. So if you want to be close to the speaker, you have to pay attention. And people do. It’s the most attentive audience I’ve ever seen. I spoke at the end of the week, on the last day, at the end of the first morning session. The crowd was tired, hungover from the grand party, and it was probably the only session where there were seats available from beginning to end. Imogen Heap gave a surreal performance just before I went up. Suffice to say I’m not used to that as a warm up.

I got a lot of reaction to my slides today at Strata Rx, but there was one part that worried me a bit, which was the reaction to my comments about filing for developer credentials for 23andme’s API. I applied during the first day of the conference. I’m asking to develop an application that sits just after the Portable Legal Consent process, so that people who have gone through the consent and are 23andme customers can just zap their data over into the open commons. I immediately got testing credentials. It’s so far a very professional setup and I commend them on the ease of the process. I’ve written about 23andme’s getting patents. And I really hope they’ll make it easy for people to export their data, starting with the genotype but ideally at least including some basic phenotypic information. If we could even get just the equivalent of the form that you fill out when you sign up - like when you go to the doctors’ office… But I always want to come back to this. I tweak 23andme but they deserve enormous credit. They show up, time and again, where other companies in the space don’t. They talk to their critics. They fight the fight at the FDA. Were there any other DTC genomics companies even in the room at Strata today? The vision is one of real change, and one that has room for a commons in it, even if it’s not all we want yet. I’m going to keep tweaking, but that’s because I think 23andme can set the tone for an entire industry by facilitating data portability through their API, not just through a lengthy manual download process. I don’t think anyone who wants to export wants to leave 23andme - I sure don’t. I just want to send a copy to a place where there’s more eyeballs because there’s lower barriers. Most people won’t. The secret of the commons is that it takes a shockingly small number of people to pull one off…though it does certainly help to have seeds. And the success of the more eyeballs space will only increase the value of the coherent, internal data set at a place like 23andme. It’s a long look kind of investment. But it would cost nothing other than a small number of people’s data sets becoming non-rivalrous, and could potentially bring enormous returns.

There’s a ton of activity around the web for Open Access Week 2012. There’s eleventymillion events in person and online around the world for you to engage in. I’m not doing any, for the first time in a long time. The OA movement is running downhill now, for real. Despite the squeals of those who would defend a broken business model tied to a broken access model, this one’s pretty much over. The question isn’t “if” open access, it’s “how” open access. OA isn’t going to be pretty, just as the old system wasn’t pretty. Using the flaws of the publishing industry itself as a rationale to ration access is a weak strawman - conflicts of interest and hucksters flacking bad journals aren’t exactly an invention of open access. They’re part of publishing. But Open Access means vastly more readers, and wider distribution of knowledge driven by science. And in the end, that point trumps all - especially given the public investment made by taxpayers in science. There is no constitutional protection for industries disrupted by network effects. Just ask Borders. Anyhow. This year’s been a weird year for me in OA. I’m not working in the heart of the movement, having moved over to try to make it easy to donate your data to science instead. But I did help poke a stick in the White House’s eye with the Access2Research petition, for which we are still awaiting a reply. Honestly, once August came and went I gave up hope of hearing anything before the election. The news cycles in a presidential election, especially one as taut as this one, magnify everything. An OA policy would’ve been jumped on by the Romney campaign, quoting those Elsevier stock hits as American job losses on the way. I remain very hopeful that the answer when it comes is a policy extending the NIH public access policy across all federal agencies but I’ve always been willing to believe. The extension of the policy is the natural end game for OA in the US from a taxpayer perspective. It’s not the end of the movement though. Now we need to avoid fragmentation. We need ORCID to actually not just mint identifiers, but itself act like an open organization (give us the freaking source code already, ok?). We need more entrepreneurs starting new publishing businesses, and testing new business models. We need big publishers to recognize the opportunity - to be the IBM/open source of the space. And we need some acknowledgment that licensing, in the end, is really important. If it ain’t CC-BY, it ain’t compliant with the community definitions of OA. It’s better than no access at all, but we should never, ever trade away our rights in return for free stuff, even if that free stuff is knowledge. Congrats all on a great year, and here’s hoping I get a good excuse to bust out the Snoopy Dance sometime in November.

Pete Stark lost his House seat last night. He’d held it for 40 years. He is a divisive figure to some. He’s a hero for his legislative work to others. And I’m not going to address any of that, though I believe history will judge him very kindly on the merits of his record. I just want to thank a man who did more for me than I did for him. Pete hired me in 1996, when I was so wet behind the ears I could water plants just by sitting next to them. I was a legislative assistant in his Washington office, which is a fancy way to say I read through the legislation that the new Republican majority was ramming down our throats and tried to find some of the nastier bits and get them taken out. One little anecdote from those days. It was my first few weeks on the job and I was overwhelmed. A previous staffer dropped a massive set of files on my desk and said, “this is the Headwaters file - you’ll want to start reading it soon.” That night I got a call at home, after midnight, that a clause had been inserted into an appropriations bill that would allow the logging of Headwaters. I had to taxi into the office, start writing Dear Colleague letters, and help whip a campaign to get the clause taken out when the bill came to the floor. I had no idea what I was doing. I was 23. When the floor debate came, I was watching it on CSPAN, which ran 24 hours in the office. We were just across the street. Pete grabbed his 2-minute comments that I’d drafted off the printer and walked over. I saw him stride up on the TV, and to my horror, he started reading the press release, not the speech for the chambers. The press release was completely over the top. It called Charles Hurwitz, the then-owner of much of Headwaters, a “blood sucking corporate vampire” (not my words, but the product of a gifted, and awesome, press secretary). You can’t attack an individual like that on the floor of the House. Pete was immediately called for censure. I nearly wet my pants. I was surely going to be fired for this mistake. The time it took for him to walk back to the office was epic torture. He came back, clapped me on the back and said, “Aw hell, I decided to read the release because I thought it would be more fun - relax, kid.” He was a great boss. Pete gave me my first big professional break. Pete’s office is where I learned how to run a database. His Mac is where I wrote my first webpage, in plain text in 1996. He gave me my first bottle of great wine. His chief of staff got me backstage at a Jerry Jeff Walker show at the Birchmere, and took me to the National Democratic Club on the night Clinton was re-elected. Pete, I left before I ever got good at working for you. But you were good to me, and you served your country well. I hope somehow this gets to you. Thanks for your service, and thanks for giving a kid a lot of breaks a long time ago. ps - we won the debate on Headwaters, and I look forward to taking my son to see those trees, when he’s old enough, and telling him this story.

Yale University published a draft “Open Data Access policy” on November 15th to make available rhBMP-2 clinical trial data that they created in conjunction with Medtronic. Let me get this out of the way before I go negative: the fact that this policy exists, is on the web, and is open to comment is an unalloyed Good Thing. It’s progress. It’s to be commended. For those who aren’t initiates in the arcane world of open data or pharmaceutical testing, this is data related to a clinical trial of a kind of biological chemical found in our bodies that plays a role in bone development. Most of the time, this kind of data is collected once, then locked away, never to be seen or used again. One of the reasons drugs cost so much is that we never re-use the data from clinical trials, so we never learn anything from failures, or from secondary uses of data. It’s an incredibly inefficient system. This project at Yale is an attempt to address that inefficiency by making the data available. But that’s the extent of my nice words. What follows is a point by point review of the policy. The short version: this is not an “open data” policy. It’s a data access policy, and if they’re not going to fix it, they need to rename it. Because those of us who care about the definition of “open data” actually meaning something are going to criticize, persistently and loudly,  if there’s any attempt to claim the title for data under this policy. I apologize in advance for the length of this rant, but it’s a long policy. Also, they’re using the name of Yoda in vain.

I’m more than 20 years away from my last math classes. And though I spent my startup experience around some very smart mathematical types, I didn’t learn enough at any point to actually use data in models, make models, or otherwise become one of those “big data” types you read about…well, fucking everywhere. But I did learn a healthy dose of respect for what data, and math, when put together by the right people, are capable of. And since signing up with Sage Bionetworks, first as a Board member, and more recently as a member of the management team, I’ve learned a lot about what data can do now that processors, storage, bandwidth, and sequencing are all brutally cheap. The key is, as per @DCDave, to reset your expectations. Cheap, plentiful data changes the epistemology of fields. It did so in baseball, which went from trusting a scout’s “gut” instinct to an intensely data driven science. It did so in weather. This year, it did so in politics. But it’s very hard to reset your expectations when you’re at the top of a traditional industry. The punditocracy’s dismissive handwaving towards Nate Silver is all I can think of when I go to traditional events on science and health these days.

Olin Hyde noted a nifty survey on patient control of electronic health record information the other day. But he interpreted it in a dour way, at least for my own worldview of openness. Here’s his tweet: Bad news for #opendata advocated by @wilbanks “@berniemd31: Study: Patients Want To Control #EHR Information - bit.ly/ToOglB” — Olin Hyde (@olinhyde) November 29, 2012 The article talks about how patients want control over the privacy of their data, so I can see why this is an easy interpretation to make. You’re a patient, someone tells you how much crap is happening with your data, and your gut reaction is to protect it. Makes sense.

Have a lot of work to do. Then a little bit of vacation to do. Be back here in the new year.

On May 20, 2012, I joined Mike Carroll, Mike Rossner, and Heather Joseph to start a campaign for a federal policy to require free access to scholarly articles emerging from taxpayer funded research. We decided to launch a petition on the We the People website. We hit the 25,000 signature number pretty quickly, and I got a few calls from DC people asking what the hell was going on. This was before every maniac who wanted to secede started a petition - it was, and still is, a backwater, and we figured we could make some noise there. It worked. Since then, petitions on beer recipes, gun control, and the death star have received official responses from the White House. We have not. I am told there is a conversation going on. I am asked to be patient. I am tired of being patient. I’m tired of the power of publisher money carrying the day, delaying the policy, blocking the flow of knowledge. It’s time to make taxpayer funds turn into taxpayer goods. It’s up to the businesses to figure out how to make money in the new world - if you can’t figure out how to adapt  your business to the network, you will merely be the latest in a long line of dinosaurs dating back to Digital Equipment Corp and running straight through to Newsweek. I do not cry for you. It’s time to bring an end to the barriers to new businesses. Time to enable new entrepreneurs, who make money from increasing the free flow of information rather than by restricting it. Time for search engines that make it easy to find and integrate the literature. Time to enable people outside the institutional system who might take scholarly knowledge and turn it into products, into policies, into something of use. It’s time for an answer. And if we don’t get the answer that says, “you paid for it, it’s yours,” it’s going to be time to escalate this from something on a sleepy backwater petition site to a protest that the whole world will notice.

If you’re new to the idea of open access and public access, brought by the sad energy of Aaron Swartz’s suicide, I urge you to try to be inspired by him and learn about the space. He was a voracious reader, an insatiable scholar. Follow his example. Do your research, read the overviews, follow the links. Learn is the first part of his four-part method. This is a well developed field, and though we need your energy, we need it very much in a learned and focused manner. There is already legislation. There are already organizations, advocates, resources. Get involved. Learn. Then Try. Then Gab. Then Build. But always, always, start by learning.

Yaniv Ehrlich’s lab at MIT has a new paper out in Science today, with a companion policy piece from the National Human Genome Research Institute at NIH. Apologies for icky paywalls but these are important papers. The gist is that a savvy computational scientist can find enough breadcrumbs in a genome to figure out the surnames for participants in supposedly de-identified studies. The methods for the paper are reminiscent of the re-identification approaches against the Netflix database, the AOL search database, and the Massachusetts health records database: cross-referencing de-identified information with other public information, and then using that against other records associated with surnames. And it gets bigger - 135,000 or so records may potentially identify millions more through inheritance of short tandem repeats. I’ve probably mangled the approach. The short version is perhaps better summed up by a picture: