Joho the Blogplatforms Archives - Joho the Blog

January 12, 2017

Olin Library: Library as place, as lab, as local theater

I went to see my friend Jeff Goldenson — we worked together at the Harvard Library Innovation Lab — at Olin College, where he’s director of the library. Jeff’s taken a library that was an under-utilized resource and, with full Administrative backing, turned it into a playground and a lab…by learning some lessons from community theater. Most importantly, he’s turned it into a place that the community feels it owns.

Olin’s got 350 students, all engineers, half of whom are women. It’s a school that stresses hands-on learning, which turns out to work well for Jeff’s approach. The library’s got two floors, neither of them particularly large, and 15,000 volumes. (Here’s a banana for scale: My local community library has about ten times that many. Yes, it is an affluent community. Nevertheless, please keep in mind that I’m still looking for work.)

Here’s some of what Jeff — who’s background is in architecture and design — has done:

First, he has done the expected things to make the library more inviting — a place as well as a resource, as Jeff puts it. These include a media tools library, maker spaces, coffee spots, some very cool events. (Ask Jeff about the Awkward Family Photobooth :)

Second, he has encouraged students to participate in coming up with new ideas for the library and, since it is a hands-on engineering school, building them.

Third, he has taken some fantastic steps to make the library re-configurable, well beyond the usual putting wheels on everything. For example, he is not only putting things on shelves in the stacks that you won’t find in most libraries, he’s coming up with ways of enabling shelves to be generally repurposable.

Fourth, Jeff being Jeff, everything he thinks of or builds is done in open, shareable ways. (Jeff undoubtedly doesn’t want me to be as cagey as I’m being in this post.)

Fifth, when you have a chance, ask Jeff about cardboard. And vinyl. And other materials that lets him and others alter the physicality of the library — the library as place — the way a local theater company creates sets. For example, once a week the Library turns a structure in the lobby into a coffee shop. It’s very popular, but it still looks like a library structure repurposed as a coffee shop. But with the magic of some cardboard, paint, and just a few inexpensive touches — e.g., some cheap hanging lamps — the structure and the space are transformed. It’s set design, with the library as the theater. This way of thinking lowers the cost and risk of altering the perceived meaning and feel of the place.

The result is not just a supercool library but a model for how existing libraries without lots of resources can give themselves over to their communities…and become a point of pride for them.

Be the first to comment »

November 27, 2016

Fake news sucks but isn't the end of civilization

Because fake news only works if it captures our attention, and because presenting ideas that are outside the normal range is a very effective way to capture our attention, fake news will with some inevitably tend to present extreme positions.

Real news items often uses the same technique these days: serious news stories often will have clickbait headlines. “Clickbait, whether fake or real, thus tends to make us think that the world is full of extremes. The normal doesn’t seem very normal any more.”Clickbait, whether fake or real, thus tends to make us think that the world is full of extremes. The normal doesn’t seem very normal any more.

Of course, clickbait is nothing new. Tabloids have been using it forever. For the past thirty years, in the US, local TV stations have featured the latest stabbing or fire as the lead story on the news. (This is usually said to have begun in Miami
, and is characterized as “If it bleeds, it leads,” i.e., it is the first item in the news broadcast.)

At the same time, however, the Internet makes it easier than ever to find news that doesn’t simply try to set our nerves on fire. Fact checking abounds, at sites dedicated to the task and as one of the most common of distributed Internet activities. Even while we form echo chambers that reinforce our beliefs, “we are also more likely than ever before to come across contrary views”we are also more likely than ever before to come across contrary views. Indeed, I suspect (= I have no evidence) that one reason we seem so polarized is that we can now see the extremities of belief that have always been present in our culture — extremities that in the age of mass communication were hidden from us.

Now that there are economic reasons to promulgate fake news — you can make a good living at it — we need new mechanisms to help us identify it, just as the rise of “native advertising” (= ads that pose as news stories) has led to new norms about letting the reader know that they’re ads. The debate we’re currently having is the discussion that leads to new techniques and norms.

Some of the most important techniques can best be applied by the platforms through which fake news promulgates. We need to press those platforms to do the right thing, even if it means a marginal loss of revenues for them. The first step is to stop them from thinking, as I believe some of them genuinely do, that they are mere open platforms that cannot interfere with what people say and share on them. Baloney. As Zeynep Tufekci, among others, has repeatedly pointed out, these platforms already use algorithms to decide which items to show us from the torrent of possibilities. Because the major Western platforms genuinely hold to democratic ideals, they may well adjust their algorithms to achieve better social ends. I have some hope about this.

Just as with spam, “native advertising,” and popup ads, we are going to have to learn to live with fake news both by creating techniques that prevent it from being as effective as it would like to be and by accepting its inevitability. If part of this is that we learn to be more “meta” — not accepting all content at its face value — then fake news will be part of our moral and intellectual evolution.

Be the first to comment »

January 22, 2016

Open Syllabus Project goes live—Yay for open platforms!

The Open Syllabus Project has just gone live with a terrific beta Web site and a front page article about it by two of the main people on the project. (I’m proud to be an advisor to the group.)

The OSP is an open platform that so far has aggregated over a million syllabi. At the beta version of their search site you can do plain old searches, or filter by a number of factors. Want to see what is the most taught work at Harvard? In the state of Texas? In the field of Biology? Lucky you.

The project is computing what it calls a “Teaching Score” for each work, a number from 1-100. This is along the same lines of the StackScore I’ve been pushing for, a metric we use in Harvard’s LibraryCloud Project and that will be used in the Linked Data for Libraries project. (The OSP used Harvard’s open catalog metadata as a main source for book metadata and disambiguation; that metadata is available through LibraryCloud’s API. It’s an intertwingly world.)

The OSP plans on making its data available through open APIs, which will multiply the good effect it has. Sites will be able to integrate data from the OSP through the API, developers will be able to create apps that use that data, and researchers will find ways to investigate it that we literally cannot imagine.

Now, you’d think someone would have done something like the OSP years ago. In fact, there have certainly been efforts. For example, Dan Cohen (currently head of the DPLA) scoured the Web and aggregated about a million publicly available syllabi. But the sad truth is that most academic institutions don’t make their syllabi openly available. In fact, many institutions and many professors copyright their syllabi. That makes sense to me if they have written little essays in them. But as a listing of topics and works, I can’t imagine why anyone would insist on asserting copyright. What’s the worst that would happen? Some other teacher copies your syllabus perfectly? That teacher has learned from you, and you’re going to teach your course differently anyway. Meanwhile, the potential good from sharing syllabi is enormous: We can learn from one another. We can see unintended patterns that may express wisdom or bias.

The OSP is here. It’s going to make a real difference.


December 17, 2015

The Library in the Life of the User: An open platform use case

OCLC has posted an excellent report based on a recent conference, looking at how libraries can participate in the life of users, rather than thinking about the user’s life within the library.

I like this a lot. I’ve been talking about it in terms of libraries now being able to participate in the appropriation of culture that traditionally has occurred in private discussions outside the library: The user borrows a book, takes it home, and talks about it with her friends, etc. It is in those conversations that the reader makes the work her own.

Now that many of those conversations occur online, the library has the opportunity to offer services that facilitate these conversations, learn from them, and contribute to the act of cultural appropriation. That’s a big change and a big opportunity. (I’d say it’s huge, but I can’t use that word without hearing it in Trump’s voice, not to mention envisioning the shape of his mouth when he says it. So, nope, that word’s gone.)

One of the points of talking about libraries in the life of the user–Lorcan Dempsey‘s phrase from 1973 (I am a Lorcan fan) [LATER: In the comments below Merrilee Proffitt points out that the report says that while Lorcan popularized the phrase, it was coined by Douglas Zweizig. Sorry!] –is that user lives are much bigger than their lives in libraries. The library’s services therefore should not be confined to the relatively limited range of things that users do in libraries. In fact, users’ lives are so big and varied and unpredictable that libraries on their own can’t possible provide every service or address every opportunity for engaging in their users’ many acts of cultural appropriation.

Therefore, libraries ought to be adopting open platforms, i.e., public-facing APIs that let anyone with an idea build a new service or integrate into their own sites or apps the ideas being generated by networks of library users. Open platforms are ideal where needs and opportunities are unpredictable. Outside of cats trapped in physicists’ boxes, there is no more unpredictable domain than how people are going to make sense of their culture together.

Therefore: Open platforms for libraries!


November 24, 2015

[berkman][liveblog] Robin Chase

Robin Chase is giving a lunchtime talk at the Berkman Center.

NOTE: Live-blogging. Getting things wrong. Missing points. Omitting key information. Introducing artificial choppiness. Over-emphasizing small matters. Paraphrasing badly. Not running a spellpchecker. Mangling other people’s ideas and words. You are warned, people.

There is a totally new organization paradigm that exists next to the Internet, she says. She calls it “Peers, Inc.” It changes how we shape the economy. It’s happening now. Her explanation will be in three parts:


First, platforms for participation that leverage excess capacity. E.g., Facebook, Skype, Meetup, YouTube, MOOCs, open source, Blockchain, etc. For example, Skype is a telecoms company built on the excess capacity of its users systems. Working with excess capacity means sharing.

Bed-sharing (couchsurfing, AirBnB) uses excess beds. “It took four years for AirBnB to have more available beds than the largest hotel chain”It took four years for AirBnB to have more available beds than the largest hotel chain (InterContinental): 650,000. Couchsurfing has more than a couple of million.

We invented big institutions to do things that we can’t do as individuals. E.g., large investments, projects that require intelligence in lots of different areas, standardized contracts. And there are things that individuals do better: customization, specialization, creativity, trust.

These two coexist, and the Net enables them to collaborate. She calls this Peers, Inc. (“Institutions and governments are also Inc’s in this world view.”) The Inc’s provide a platform for participation, and the individual provides creativity and specialization.

Robin “adores excess capacity” because it’s green and efficient. Excess capacity is something that’s already been paid for but contains unused value. How do you harness it? 1. You can slice it so only pay for what they use (e.g., ZipCar); this lets you avoid buying more car than you need. 2. You can aggregate (e.g., AirBnB, Waze). 3. Open up these assets, e.g., and GPS.

The Inc side builds platforms for participation. They organize lots of small parts. “They “Platforms give the power of the large to the small”give the power of the large to the small.” They can scale. She points to a French car-sharing company: BlaBlaCar. Four million people use it every month.

Peers bring diversity. E.g., smartphones and apps. Smartphones are far harder to build than the apps they enable. Over 2M apps have been developed since smartphones were invented in the past seven years. “We’ve seen more innovation than throughout all of human history” because people can build apps that are relevant to their own situations. App creators are free-riders on top of the $600 people spend on their smartphones.

2. Peers Inc give us new powers, which she thinks of as miracles.

“The most depressing thing I know is climate change.” By 2100, we’ll see a 4-6°C increase unless we take dramatic action. What does that feel like? “The last time we were minus 7°F was the last ice age.” Warming the planet that amount transformed the planet. We should expect the same level of change if we boost it another 7°F. By 2060, it will be really awful. So we have to address this.

“Banny Bannerjee says: “You can’t solve exponential problems with linear solutions.””Banny Bannerjee says: “You can’t solve exponential problems with linear solutions.”

The “miracles” give her some optimism:

a. “We can defy the laws of physics” by leveraging excess capacity. If she had proposed building 640,000 rooms in four years she would have been told that that’s not possible. But AirBnb did it by leveraging existing excess capacity.

b. “We can tap exponential learning.” Platforms can get millions of iterations in and can do a lot of learning. E.g., learning a language. A semester is 130 hours. Rosetta Stone teaches the same in 54 hours. But it’s expensive. “My new favorite company is DuoLingo.” They do a lot of A/B testing. They now can teach you a semester in 34 hours. They have 90M people using it. A year and a half ago DuoLingo opened up its processes: Russians learning Balinese, etc. Now 45M of the 90M are learning language pairs DuoLingo did not create. (DuoLingo makes money because they have humans translating sentences from organizations that pay them incrementally.)

c. “The right person will appear.” E.g., Obama raised the prospect of normalizing relationships with Cuba. Six months later, AirBnB had 2,000 listings there, thanks to the Net.

Her only hope for climate change is creating platforms that will address climate “at scale, speed, and locally adapated.” E.g., a platform for a house will remember to turn off the light when there’s been no movement. We’ll get smart cities through the Internet of Things. Distributed energy. Autonomous vehicles, which will arrive in force in the next 5-12 years. We’ll only need 10% of the cars because we’ll be sharing them. “Public transportation will be at the cost of a bus but the speed of cars”Public transportation will be at the cost of a bus but the speed of cars, transforming job opportunities. (But the Internet of Things means that everything is tracked.)

All of these miracles only happen because of both sides of Peers Inc.

3. “Everything that can become a platform will become one.” Old-style industrial capitalism put thick boundaries around companies. Today, what’s inside and outside is blurred.

Four reasons Robin is convinced we’re moving into the collaborative economy:

1. Shared networked assets always provide more value than closed assets

2. More networked minds are smarter than fewer proprietary minds.

3. “The benefits of shared open assets are always larger than the problems associated with open assets.” E.g., yes, some people put scratches in ZipCars, but the company nevertheless is doing very well.

4. What I get is great than what I give.

We are in a time of instability. “Peers Inc is the only structure that can experiment, iterate, evolve and adapt at the pace required.”Peers Inc is the only structure that can experiment, iterate, evolve and adapt at the pace required.

So, how can we structure things so we give up the least privacy necessary? “What is the least privacy loss that delivers a habitable climate”


Q: For me it’s not privacy loss but who we’re losing our privacy to. What about platform accountability? Aren’t we pushing out power into more abstract systems that we cannot see or address?

A: I was on a panel at the Platform Cooperativism conference. I pointed out that these platforms are incredibly expensive. ““He who finances the platforms creates the rules of engagement.”He who finances the platforms creates the rules of engagement.” “I want these platforms created by a distributed, autonomous us.” We don’t have time to just hope this happens. “I have real anxiety.”

Q: [me] Suppose we build protocol instead platforms…

A: I’ve put all of that into the same bucket.

Q: Shareable cars disrupt ZipCar. There will be user agreements. How do we disrupt that?

A: “He who creates the data owns the data.” Autonomous vehicles have a middle space, e.g., around safety and learning issues. It’s in the deep public interest to have this data. But we need to make the privacy issues understandable and parseable by ordinary users so they can choose.

Q: Isn’t privacy gone already?

A: We can still do some structuring.

Q: Why does trust work over the Web, which is mostly anonymous?

A: Ebay was the first to figure out you need ratings and commentaries. We use other people as our proxies for trust.

Q: iRobot’s Roombas currently don’t upload what they’ve learned about the layout of your house. But Nest knows everything. What should the rules be?

A: That’s what I’m asking you. We have to figure this out.

Q: It’d be great if we had more choice about which pieces of info we give to platforms. Is there any work on standard ways of parceling out pieces of our identity?

A: I know people are working on this. “It comes back to the amount of money, time, and marketing it takes to push great ideas into market.”

Q: What are we doing to educate the younger generation about privacy?

A: Maybe you can push Harvard to do appropriate role modeling. Maybe students here could push for an icon system that tells us what data you’re taking from us, etc.

Q: [me] What would you tell a student about the dangers? And would you consider addressing this by putting restrictions on how the data is used, rather than on its collection?

A: How about doing some pilots to see what works? You have to inform people about the dangers as well regulating the industry.

Q: How will we embed public safety concerns into software for self-driving cars?

A: Self-driving cars will always follow the rules. No speeding. No parking in no-parking zones. All the existing rules will be embedded. So we’ll embed the appropriate behavior for ambulances, etc. No siren required. Also: The auto industry always brings up autonomous cars having to decide which person to kill in an accident. But why would you bring up this stupid case? One in a million trips this might happen? There are more deaths than that now. “Right now, 80% of cars are single occupancy. We need to put a high price on that.”Right now, 80% of cars are single occupancy. We need to put a high price on that.

Q: It sounds like you’re describing a train: get somewhere, not park…Why not public transportation?

A: We’ll see how it plays out. It’ll be a complex ecosystem. It’ll be decided city by city. More important than who owns it are: Will they be electric? Will it be 10x more expensive for single occupancy? Will we have pharmacy cars or liquor cars that deliver their wares without having a storefront? Who will design the software?

Q: Practically, how do you combat zoning for selfishness, e.g., my own one-person gas guzzler?

A: I don’t spend a lot of time on local issues. When I have, logic and data haven’t had much effect.

Be the first to comment »

October 26, 2015

[liveblog][act-iac] Innovation in govt

Brian Nordmann (Senior Advisor, Arms Control, Verification and Compliance at U.S. Department of State) begins with the standard disclaimer that he’s not speaking for the Dept. of State. And here’s mine:

NOTE: Live-blogging. Getting things wrong. Missing points. Omitting key information. Introducing artificial choppiness. Over-emphasizing small matters. Paraphrasing badly. Not running a spellpchecker. Mangling other people’s ideas and words. You are warned, people.

Brian reminds us that “innovation” has become a tired term. There was a govt flurry to innovate, but no one said exactly what “innovation” means. The State Dept. built a structure so there could be quarterly reports on innovation. “If you want to guarantee that you don’t do anything innovative, create a structure for writing reports.” He says this was a shame because in the basement there was already the perfect place for innovation: The Foggy Bottom Cafe — Starbucks, ice cream parlor, etc.. Sect’y Colin Powell went down there every day because he knew in 45 minutes he’d get twenty-five innovative ideas. People sit there and share ideas. “This is how you get innovation done in the gov’t”: Give people the freedom to talk, and the freedom to fail.

A couple of years ago his group started doing public challenges. In the first one, they got 150 entries. They awarded $5K for ideas they would have had to pay hundreds of thousands of dollars for. But then the lawyers discovered they were doing this and wrote a EULA for the site — 28 pages, with buttons strewn throughout that you have to press. Twenty-seven people applied to that year’s challenge.

Brian’s job is simple: Get rid of all nuclear weapons in the world. His office’s job is to come up with ways to verify agreements. In the 1960s, the sensors were physically large: big, expensive, fragile, and now replacement parts don’t exist. A radar installation in the Aleutian Islands looking for nuclear missile launches uses vacuum tubes.

Now their challenge is to explore arms control in the information age. What can we learn from YouTube, Facebook, etc.? But the lawyers say that’s a privacy violation. So, instead they’re investigating the Internet of Things. People don’t mean the same thing by that phrase. Brian means by it: networks with sensors.

Are there things we can do to get the public involved in arms control? It’s a complex issue, but you can simplify it to: What can we do to rid the world of nuclear weapons. Brian holds up a small spectrometer that feeds into a laptop. It’s used to analyze water quality. Another instrument is a rolled-up piece of cardboard that attaches to your phone. And smartphones’ accelerometers can sense earthquakes…and nuclear explosions. They could alert agencies that they need to look at the explosion more closely.

Researchers in Hawaii bought 12 iPhone6’s to explore this, which tripped an alarm at Apple. Apple contacted the researchers. The researchers told Apple that their phones could be used as seismic detectors, and that the iPhone6 degraded that capability. The researchers are trying to broaden Apple’s sense that its phones can be used for more than app delivery.

For innovation, you want to talk to new people, not the same people all the time. By bringing in new people, you’ll get a lot of junk, but also some ideas worth exploring. Hobbyists and startups are generally better to talk with than large companies. Brian spoke with Tom Dolby, ex of MTV, who has a media lab at Johns Hopkins that is working with Baltimore youth. Brian works with a California group teaching Latino kids how to program. Imagine putting them together, along with people from around the world, and create a Teen Summit. Imagine they see what they have in common and what they do not.

Q: How are you communicating to device manufacturers that they are platforms for innovation?

A: People respond to a title that ends with “US Dept. of State.”

Be the first to comment »

September 11, 2015

[liveblog] Bringing Silicon Valley to Government

I’m at an event sponsored by the Shorenstein Center and Ash Center and the Center for Public Leadership on “Bringing Silicon Valley to Government?” (#HKSgovtech). Panelists are:

NOTE: Live-blogging. Getting things wrong. Missing points. Omitting key information. Introducing artificial choppiness. Over-emphasizing small matters. Paraphrasing badly. Not running a spellpchecker. Mangling other people’s ideas and words. You are warned, people.

Van Dyck, Sinai, and Martin

Van Dyck, Sinai, and Martin

Maria Martin begins. She had founded a company and worked there for ten years, but then applied to be a Presidential Innovation Fellow, thinking she stood no chance. (She lightly recommends sending in a resume printed on colored paper.) She got in, and moved to DC for six months, or so she thought. Then she was asked to be a senior advisor in the White House. One day her boss, Todd Park [former federal CTO], couldn’t make a meeting at the Veterans Administration. She went and fell in love with the problem because it affected veterans and because it seemed solvable through software. The other people around the room were policy folks and didn’t know how to use software that way. She was told by her friends it’d be frustrating, but you can learn how to get things done in government. She was 28 years old when she became CTO of the VA.

Nick introduces Tim O’Reilly as “the godfather of tech.” Tim begins by denying that. “The thing I’ve done most of my career is to watch interesting people and say, ‘Wow, there’s something there!’.” “I’m kind of like a talent scout.” The person who first inspired him about gov’t is Carl Malamud. “He was the first person to put any govt agency on the Internet.” Carl went to Eric Schmidt who at the time was CTO of Sun and licensed data from the SEC. After two years Carl said he was going to shut it down unless the govt took it over. (See Public Resource.)

Then Tim saw Adrian Holovaty‘s mash up of crime data with Google Maps and said, “Wow, techies are beginning to pay attention to govt.” This inspired Tim to want to get more govt stuff into his Web 2.0 conference. He realized that government as a platform had created great value. E.g., opening up GPS. “So much of govt has been focused on, ‘We have to build it or else it won’t exist.’ … My idea was that if you open data, entrepreneuers can build on it.” Not open data for transparency so much as open data for building things. Todd Park, who was at Health & Human Services at the time, “totally got it.” When Amazon launched its Web Services, it found the people illegally hacking on its data. Tim’s company was one of them, to get book data. So Amazon brought in the hackers and built the services they need. HHS did this too. There are now hundreds or possibly thousands of apps using the HHS data.

Haley Van Dyck was on track in 2008 to move to Beijing to work for CNN on Olympics coverage. She was at a dinner where speakers talked about the need to do more for race relations in the US. Obama was running, so she went to work for him in Chicago. When they won, they were asked to connect citizens to government. She started at the FCC building the first new media team to fix their interface to the public. [Wow, they had a terrible UI. Thanks for helping to fix it, Haley!] She thanks Tim for helping to build a like-minded tech community in building. She now leads the US Digital Service.

Nick Sinai went to business school, worked for Lehman, and then pivoted to go to work for Pres. Obama, at the FCC for 1.5 yrs and in the White House for four years.

Haley: USDS is a team embedded in the White House and distributed dedicated to transform the most important citizen-facing services. It was started after the rescue effort. “Let’s bring in a couple of hyper-talented engineers” and add them to the hundreds of consultants in order to change the environment tasked with fixing the site that the President’s most important initiative rested upon. “At any moment, there were only 5-6 people working on” and they were able to fix it. The President asked that this method be used more broadly, so the USDS was founded.

The USDS theory of change is that the best way to create change is to deliver results and to do it where it’s most needed. There’s a team at the VA and a Homeland Security working on immigration services. The immigration process is currently entirely paper-based. To apply you have to send in avery long paper docket that humans then look out. It’s difficult to put this together. The paper file get sents around to immigration centers. By the end (6-9 months) it will have traveled the distance equivalent to going around the world six times. The postal costs alone are $300M/year.

A seven year long procurement for $1.2B was begun about ten years ago [I think] which resulted in a process that was even longer. The Obama administration decided to fix this. A $1.4B procurement process put it into the same hands as the first time. “We can’t build an application process the way we build battleships.” (Tim quotes Clay Shirky that the traditional waterfall sw dev process is “a commitment by everyone not to learn anything while doing the work.”) Instead, the digital team — five people — released the first sw update four months later.

Nick: Traditionally, we spend years developing the procurement. Then years writing the requirements. Then years building the system. Then in year 7 or 10 sw would actually launch. We’ll spend billion of dollars on a single sw enterprise system and then it fails because sw changes, and the requirements are wrong because no one tested them. But instead sw developers rapidly deploy, test what’s working, iterate.

Tim: There’s a real cultural change. If you’re a supplier charging the govt a billion for something that only costs a million, you don’t have incentive to shrink your profts a thousand-fold. The contractors say the project is massive like a moon landing, but Silicon Valley people look at it and say, “Actually, it’s in the range of a mid-size dating site.”

Haley: The five people on the team couldn’t have done it on their own. They worked with the contractors who were there. It’s a tight partnership.

Nick: We spend $80B/year in IT in govt, not including intelligence. (HHS has $11B budget for IT.) But we’re not getting the value.

Marina: The VA has 330,000 employees and 8,000 IT managers. It took a year and a half to get the first hire of the new team in. She had to document that she couldn’t hire through the usual pipeline…which itself took a year. She had to show the value of hiring another 75 people.

A VA example: The President was going to announce a site where you could put in the number of years of service and see how many “GI dollars” [? – couldn’t hear] you have. The contractor spent $1M building this simple page and it wouldn’t even load on the Internet. Marina asked one of the PIFs [Presidential Innovation Fellows]to look at the page. S/he called three hours later it and had fixed it.

[Audience member:] The culture, based on the annual budgets of the agencies, is more complex than you’re saying. SW companies selling to the govt have to include complexities to meet the culture and requirements. [Not sure I’m getting this.]

Marina: You change culture by celebrating vendors who do it in new ways.

Nick: The Administration has a program to educate the contracting officers across all the agencies who do the negotiations.

Marina: There are 1000+ websites at the VA and maybe a dozen IDs for each veteran. 942 toll-free numbers. So, how do you change the veteran’s experience. I could argue the need for this for 20 years, but instead we exposed it to the veterans. We didn’t close down the 1000 websites, but instead created one website that lets you get to what you need. How do we get info to the people who support veterans? To the community? The answer to our most-asked question takes 17 clicks to get to, and that answer is “Call your RO” without telling you what “RO” means (regional office) or how to call it.

The VA also built its own Electronic Health Records system many many years ago; we’re going to launch a new, open source EHR platform. And it’s building the first apps to make sure it works.

Third, the VA team is working on the appeals backlog. You have to process them in chronological order so there’s no low-hanging fruit. One of the boxes you can check is “Do you want a local hearing?” That’s very attractive to users, but it doesn’t tell them that that means a judge will be flown into their city in 2019. Giving users more info would help.

Nick: You’re engaging in user-centered design. The shift is massive. How do you hide the complexity of govt?

Tim: You should all read The UK Digital Govt Design Principles. The message is: Users first. The big difference between govt and the Valley is that in the Valley if you don’t please your customers, you’re out of business. But in govt you can go on for years getting funded. “The feedback loop is fundamentally broken.” In Silicon Valley you test, you work on it incrementally, you add new features.

Tim also recommends Jake Solomon’s “People, not Data” about food stamps in San Francisco. Why is there so much churn in the system? People apply and then drop out. One big reason: Applicants receive incomprehensible letters. “That was just the first step in debugging the system for users.” Govt administrators should be required to use this systems. (This was addressed by Code for America, which was a model for PIFs, started by Jen Pahlka (Tim’s wife.)

Haley: We’re working on a big project for the President. His advisors had a feature they really really really wanted included in the project. We were running an agile dev project, and added the feature. But in testing it turned out that the users weren’t clicking on that feature, preferring to use the search engine instead. But one of the advisors was incredibly upset that the feature he wanted wasn’t included in the launch version. We explained why. “We saw their minds just shift.” The advisor said, “You’re not building it because users don’t want it! We shouldn’t just build sw this way. We ought to build policy that way as well.'” “I left the West Wing wanting to cry [with joy].”

Tim: “If you can build something and show it works, you can change minds” about policy.

Nick: How do we turn those feedback cycles into weeks or months…?

Haley: Here’s the second half of the story. We decided to release all of the data to the partners in a private beta. It turns out that the feature the advisors wanted has been implemented by a third party as a separate standalone product.” You can achieve so much more by opening up data than by doing it all yourself.

Nick: Regulators sometimes fight innovation…

Tim: See my “Open Data and Algorithmic Regulation.” Lots of people hate regulation, but we want, for example, credit card companies to monitor usage to prevent fraud. This is a type of regulation. The credit card companies regulate in real-time and adaptively. Maybe govt regulation should look this way. The Fed does this, because we’re judged by the outcomes, not by our adherence to policy rules like “Interest rates shall be set at X.” They continuously adjust the knobs and levers. Nick Grossman has raised this recently about the debates about Uber and AirBnb. He says that they ought to open up their data to govt regulators so we can figure out the actual impact. Is Uber increasing or decreasing congestion?

Nick: What are the career paths like?

Marina: I didn’t think there was any way I could be in govt. There are many routes in.



Haley: Just do it.

Tim: You can do a lot from the outside. We all have a call to public service. We need a fundamental rebirth of civic mindedness, that govt something we do together. We need govt, unlike the conservatives and Silicon Valley believe.

Haley: We don’t know how algorithic regulation works or how far it should go into the delivery of services. This is something we all need to work on.

Nick: Engineering and Design programs should be working on these sorts of issues.


Q: Traditionally govt was not a top option for grads. Now it’s becoming more attractive. There’s more of a sense in the Valley that it’s their duty to fix things rather than just complaining.

Q: What steps are you taking beyond getting shit done to address some of the fundamental issues around procurement,agile dev, hiring quickly…?

Haley: We began with hiring. We’re done from 9 months to 4 wks to hire someone, which make us competitive with the private sector. (We use Schedule A Hiring Authority. “This is super wonky.”) The procurement process: crucial. We need to enable the right kind of companies to do the work. First, we’re working on building better buyers by bringing in technologists. In 1.5 months we’re launching an agile procurement process. We’re starting to train contracting officers with a five month course to understand how to procure digital in a way that makes sense. We’working with 18f on a pilot to reduce the barriers. We’d love to rebuild Schedule 70 .

Nick: It takes about 9 months to get on that schedule. GSA has a goal of getting this down to 21 days. There are ideas about raising the threshold for purchasing. Right now it’s an 80-90 PDF you have fill out. There’s so much friction in the system and you end up with people’s core competency is navigating the bureaucracy of the system.

Tim: As part of the new process, you’re given a data set available through a public API and you’re given a working app within a week. One of the largest IT companies couldn’t do it so they failed the agile certification process.

Nick: It’s about show, not tell.

Tim: It used to be that you had to provide a working model to get a patent. Think about all the junk sw patents. A working prototype ought to be a requirement.

Marina: And it’s not just about building apps. It’s about understanding the users and getting the incentives right.

Q: Everyone has a different definition of “agile.”

Tim: There’s always more than one way to do it, to cite the PERL slogan. There’s a family of things you can call “agile” : iterative, small pieces, feedback loops. Any version of agile is better than any version of waterfall development.

Nick: If it delivers in weeks, then I don’t care what we call it.

Q: You’re competing with Silicon Valley for people. Have you thought about offering H1 B visas for USDS?

A: That’s very interesting to us.

Q: How about the politics involved? There are forces that have spent decades dismantling in place systems that let Congressfolk know what they’re doing, etc. What are you doing to make sure that USDS is robust?

A: I’ve been surprised at how easy it’s been to identify the common causes you share whether you’re a Dem or Repub. No one wants veterans to have bad service from the VA. TBD is how the contracting community will respond. We’ve tried to be clear that this is not about taking away business from contractors. We’re not on a witch hunt against them. We want to make them more efficient. In many instances they’re delivering the systems we asked for but we got the requirements wrong.

Q: What skills should students develop to be attractive to USDS, etc.? Especially for Kennedy School students?

Marina: I hired a Kennedy School grad who’s been amazing. He has good dev skills,but more important he’s able to understand and ask questions and navigate through problems.

Haley: Show results.

Tim: It’s not about being a rockstar coder. Rather, solve user problems, and have a fundamental facility with tech that lets you say, “Oh yeah, that’s easy to do. Here’s the tool you use. The consumerization of IT means that you often don’t need to go to someone else to get something done.” “Tools like GitHub should be in your reportoire.” Young people can come into govt and help it see what things are easy so we spend money on what’s hard.

Nick: Go to a hackathon and work on some project together.

Q: As a designer and architect, how much of it is govt interacting with architects, city planners, and people who care about design? Lawyers often implement laws without regard to design.

Haley: We hire product designers, visual designers, user researchers. We’re in desperate need of more of them.

Tim: Code for America uses a lot of designers as well. And designed should be tested and iterated on as well. “It’d be awesome to have the equivalent of agile dev in city design.”

Nick: You could argue that The Constitution is a design document.

Q: I haven’t heard much from you about saving money.

Marina: I have to lead with the impact on veterans’ lives. Cost-savings is important but it isn’t enough of a driver of change. Even if we save $2B of the $80B, it barely dings the chart.

Haley: It’s an amazing secondary outcome of building better services for users. Also, cost-cutting inadvertently puts you in adversarial stance with some of the folks there. It’s easier to focus on who we are serving.

Q: What are you doing to create a sense of urgency?

Marina: We’re not going to be around forever.

Haley: We are living in a services delivery crisis. Veterans are dying because of that.

Nick: “Practice radical empathy.” You can’t just drop in as a hot shit technologist. You have to have empathy not just for veterans but for the person who’s been in this job for thirty years. You have to ask how you can make them the hero.

Tim: When immigration reform was on the table, that created urgency to get ahead of the topic so that there would’t be another sort of meltdown. And people want to be remembered for doing something good. Listen to what they’re trying to accomplish and how you can help them. There’s a human element.

Q: How to bring the human voice into policy circles? USDS asked me to report on what it’s like to get an immigrant visa. They sent me to the Dominican Republic where I talked with people about their experience, and shadowed them when they went to the local immigration office while they waited for 5 hours for their 10-min interview. The deliverable was a memo to the President. Having a background in policy was helpful in writing that in a way that made user needs and experience understandable to policy makers.

Haley: It coupled the policy discussion with implementation suggests. That’s rare and can be transformative.

[What a fantastic panel. And a completely awesome set of people — more examples of what true patriotism can look like.]

1 Comment »

July 13, 2015

What open APIs could do for the news

In 2008-9, NPR, the NY Times, and The Guardian opened up public APIs, hoping that it would spur developers around the world to create wonderful and weird apps that would make use of their metadata and spread the availability of news.

Very few little happened. By any normal measure, the experiment would have to be deemed a failure.

These three news organizations are nevertheless fervid evangelists for the same APIs—for internal use. They provide an abstraction layer that makes the news media’s back ends far easier to maintain without disrupting their availability to users, they enable these organizations to adapt to new devices and workflows insanely quickly, they facilitate strategic partnerships, they lower the risk of experimentation, and more.

This was the topic of the paper I wrote during my fellowship at The Shorenstein Center. The paper then looks at ways we might still get to the open ecosystem for news that was first envisioned.

The full paper is available freely at the Shorenstein site.

There’s an op-ed length version at Nieman Reports.

1 Comment »

January 7, 2015

Harvard Library adopts LibraryCloud

According to a post by the Harvard Library, LibraryCloud is now officially a part of the Library toolset. It doesn’t even have the word “pilot” next to it. I’m very happy and a little proud about this.

LibraryCloud is two things at once. Internal to Harvard Library, it’s a metadata hub that lets lots of different data inputs be normalized, enriched, and distributed. As those inputs change, you can change LibraryCloud’s workflow process once, and all the apps and services that depend upon those data can continue to work without making any changes. That’s because LibraryCloud makes the data that’s been input available through an API which provides a stable interface to that data. (I am overstating the smoothness here. But that’s the idea.)

To the Harvard community and beyond, LibraryCloud provides open APIs to access tons of metadata gathered by Harvard Library. LibraryCloud already has metadata about 18M items in the Harvard Library collection — one of the great collections — including virtually all the books and other items in the catalog (nearly 13M), a couple of million of images in the VIA collection, and archives at the folder level in Harvard OASIS. New data can be added relatively easily, and because LibraryCloud is workflow based, that data can be updated, normalized and enriched automatically. (Note that we’re talking about metadata here, not the content. That’s a different kettle of copyrighted fish.)

LibraryCloud began as an idea of mine (yes, this is me taking credit for the idea) about 4.5 years ago. With the help of the Harvard Library Innovation Lab, which I co-directed until a few months ago, we invited in local libraries and had a great conversation about what could be done if there were an open API to metadata from multiple libraries. Over time, the Lab built an initial version of LibraryCloud primarily with Harvard data, but with scads of data from non-Harvard sources. (Paul Deschner, take many many bows. Matt Phillips, too.) This version of LibraryCloud — now called lilCloud — is still available and is still awesome.

With the help of the Library Lab, a Harvard internal grant-giving group, we began a new version based on a workflow engine and hosted in the Amazon cloud. (Jeffrey Licht, Michael Vandermillen, Randy Stern, Paul Deschner, Tracey Robinson, Robin Wendler, Scott Wicks, Jim Borron, Mary Lee Kennedy, and many more, take bows as well. And we couldn’t have done it without you, Arcardia Foundation!) (Note that I suffer from Never Gets a List Right Syndrome, so if I left you out, blame my brain and let me know. Don’t be shy. I’m ashamed already.)

The Harvard version of LibraryCloud is a one-library implementation, although that one library comprises 73 libraries. Thus the LibraryCloud Harvard has adopted is a good distance from the initial vision of a single API for accessing multiple libraries. But it’s a big first step. It’s open source code [documentation]. Who knows?

I think it’s impressive that Harvard Library has taken this step toward adopting a platform architecture, and it’s cool beyond cool that this architecture is further opening up Harvard Library’s metadata riches to any developer or site that wants to use it. (This also would not have happened without Harvard Library’s enlightened Open Metadata policy.)

1 Comment »

November 26, 2014

Welcome to the open Net!

I wanted to play Tim Berners-Lee’s 1999 interview with Terry Gross on WHYY’s Fresh Air. Here’s how that experience went:

  • I find a link to it on a SlashDot discussion page.

  • The link goes to a text page that has links to Real Audio files encoded either for 28.8 or ISBN.

  • I download the ISBN version.

  • It’s a RAM (Real Audio) file that my Mac (Yosemite) cannot play.

  • I look for an updated version on the Fresh Air site. It has no way of searching, so I click through the archives to get to the Sept. 16, 1999 page.

  • It’s a 404 page-not-found page.

  • I search for a way to play an old RAM file.

  • The top hit takes me to Real Audio’s cloud service, which offers me 2 gigabytes of free storage. I decline.

  • I pause for ten silent seconds in amazement that the Real Audio company still exists. Plus it owns the domain “”

  • I download a copy of RealPlayerSP from CNET, thus probably also downloading a copy of MacKeeper. Thanks, CNET!

  • I open the Real Player converter and Apple tells me I don’t have permission because I didn’t buy it through Apple’s TSA clearance center. Thanks, Apple!

  • I do the control-click thang to open it anyway. It gives me a warning about unsupported file formats that I don’t understand.

  • Set System Preferences > Security so that I am allowed to open any software I want. Apple tells me I am degrading the security of my system by not giving Apple a cut of every software purchase. Thanks, Apple!

  • I drag in the RAM file. It has no visible effect.

  • I use the converter’s upload menu, but this converter produced by Real doesn’t recognize Real Audio files. Thanks, Real Audio!

  • I download and install the Real Audio Cloud app. When I open it, it immediately scours my disk looking for video files. I didn’t ask it to do that and I don’t know what it’s doing with that info. A quick check shows that it too can’t play a RAM file. I uninstall it as quickly as I can.

  • I download VLC, my favorite audio player. (It’s a new Mac and I’m still loading it with my preferred software.)

  • Apple lets me open it, but only after warning me that I shouldn’t trust it because it comes from [dum dum dum] The Internet. The scary scary Internet. Come to the warm, white plastic bosom of the App Store, it murmurs.

  • I drag the file in to VLC. It fails, but it does me the favor of tellling me why: It’s unable to connect to WHYY’s Real Audio server. Yup, this isn’t a media file, but a tiny file that sets up a connection between my computer and a server WHYY abandoned years ago. I should have remembered that that’s how Real worked. Actually, no, I shouldn’t have had to remember that. I’m just embarrassed that I did not. Also, I should have checked the size of the original Fresh Air file that I downloaded.

  • A search for “Time Berners-Lee Fresh Air 1999” immediately turns up an NPR page that says the audio is no longer available.

    It’s no longer available because in 1999 Real Audio solved a problem for media companies: install a RA server and it’ll handle the messy details of sending audio to RA players across the Net. It seemed like a reasonable approach. But it was proprietary and so it failed, taking Fresh Air’s archives with it. Could and should have Fresh Air converted its files before it pulled the plug on the Real Audio server? Yeah, probably, but who knows what the contractual and technical situation was.

    By not following the example set by Tim Berners-Lee — open protocols, open standards, open hearts — this bit of history has been lost. In this case, it was an interview about TBL’s invention, thus confirming that irony remains the strongest force in the universe.

    1 Comment »

  • Next Page »