One of the great benefits of being a fellow at the Berkman Center is the chance to participate in our Tuesday lunches. Fun folks from the worlds of law, development or the Internet are waylaid while passing through the greater Boston area, dragged to our secret lair in darkest Cambridge, and compelled to give presentations while our gang of inquisitors pose thorny questions. Oh, and box lunches are served.
It’s very nice, really, and open to the general public. If you’re in Cambridge and interested in these issues, or even just in free lunch (because there is such a thing as a free lunch), check out the Berkman luncheon page and let Wendy know you’re coming.
Our lastest victim was my friend Jim Fruchterman, social entrepreneur, big geek and generally brilliant software for social change dude. Jim is the founder of Benetech, a remarkably cool technology company focused on building products that bigger tech companies won’t build (for market reasons) but which have huge social impact. Rather than give the full intro to his work, I’ll offer my notes of his talk:
Jim got his start working on computer vision in a lab at Stanford. His lab was working for the Deparment of Defense on automatic feature recognition and smart bombs – Jim found himself wondering if this technology could be used to help the blind read rather than to kill lots of people. His advisors were less interested in this idea, and he found himself helping run a company that cleaned up in the character recognition space, pioneering tools that did omnifont character recognition (i.e., that were able to recognize text irrespective of whatever font the document used.) While Jim continues to work on the financial side of that business, he mostly runs Benetech.
Benetech comes out of a recognition that most big companies can’t address small markets. Machines that read for the blind might be a $1-5 million dollar market per year. That’s a size not especially interesting to most tech companies, which are looking for markets that are $50m a year or more. But when you’ve got a company that attempts to be profitable, but owned by a not-for-profit foundation, you can choose to take on projects that are marginally profitable and have huge social impact. Hence, reading machines for the blind.
Benetech uses a character recognition card (developed by the other company he works with, I think…), a flatbed scanner, and some text to speech libraries from IBM and offers a reading machine at a price point closing in on $1000, much cheaper than competitive machines designed by Ray Kurzweil. The business broke even in year one and is now so profitable, Jim was able to sell it to a for-profit company for $5 million.
That $5 million has financed the next generation of Benetech projects. One of them is bookshare.org. Jim describes it as “Amazon plus Napster plus talking books, but legal.” Basically, Bookshare takes advantage of the fact that there are hundreds of reading machines located around the world, where blind people can scan a book and turn it into a digital file, which can be turned into speech. But that process is time-consuming (about three hours a book) and it seems logical that a book scanned for one reader should be usable by another blind reader.
So Bookshare allows blind readers to download all they can read for $50 a year. They can upload books that they’ve scanned and share them with other readers, ala Napster. The $50 a year fee doesn’t yet cover the costs of the operation, but does cover about 25%, reducing the project’s dependence on charitable support.
Remarkably, this is completely in line with US copyright laws. There’s a provision, known as the Chafee Amendment, which allows redistribution of copyrighted works for people with certain verifiable disabilities. (No, being too cheap to buy music from the local record store does not qualify as a verifiable disability, you crafty Napster users.) Jim approached major publishers before starting to redistribute their works and, once they understood what the project intended to do, they were generally supportive.
The project uses the weakest possible digital rights management technology, because Jim recognizes that users need to convert the content into various forms: large-type print, braille, MP3, and so on. So the content is unlocked and marked with a warning which reads, Jim says, “If you post this book on the Internet, you screw the blind community. Don’t do it”. Benetech has developed credibility with the publishing community by removing cracked e-books from the site and generally ensuring publishers that they’re not trying to put them out of business, but trying to help the blind.
Another successful Benetech project is Martus, a package of “basic IT tools for human rights groups”. Jim asserts that most testimonies taken by human rights groups get lost: files are destroyed when people get arrested, computers get seized, paper files get eaten by termites. (Really.) As Patrick Ball, the leading stats guy in the human rights field – and now a Benetech staffer says, “Human rights is an information processing industry where no one is developing tools.”
Benetech tries to make storing files easy and safe for human rights organizations. Jim contends that 99% of the groups that organizations like Privaterra train on PGP don’t end up using it (I doubt the figure is that high, but I accept the general point – PGP’s hard to use, and it doesn’t help anyone to know PGP and not use it.) So Martus is the combination of a client, which looks a lot like an email client, which allows users to publish documents locally or to a webserver. Documents are encrypted when saved and copies are uploaded to a server run by a trusted party: Open Society Institute for some groups, the Catholic Church for others.
New Benetech projects include a low-cost landmine detector, and a series of cellphone-based applications for the blind. Jim shares my conviction that PCs are the wrong platform to deliver IT services to the poor – he sees cellphones, which have already become indispensible to blind people, working as reading machines and navigation engines for blind people at low costs.
The common theme to Jim’s work is an attempt to respond to market failures in bringing intellectual property to the market. He’s interested in figuring out how to “gently coerce” IP-holders to part with their rights for limited, specialized markets. For instance, he convinced Intel to donate $250k of mismarked, unsaleable chips to build reading machines – eventually, they decided just to donate $1m of properly marked chips. IBM has donated ViaVoice software, HP donates scanners to keep them from the low end of the market. Are there ways to convince contentholders to give their content under similar terms with similar PR benefits?
Jim sees great opportunities for business students, mid-career professional trying to put their skills to work in the social sector. As he points out, a $1m a year business would be a total failure in the high-tech sector; “It’s a barnburner in the social sector”.
Benetech isn’t the only group pursuing this model. David Green’s Aurolab, based in India, makes programmable hearing aids and sells them at $5 – 100, rather than the $2000 these devices sell for in the US. How? They invested $2m in designing a $5 chip that’s the heart of the device, which now is able to be manufactured for $45 a unit. The hearing aid is sold at higher prices to those who can afford it and subsidizes the cost of hearing aids for those who can’t. He’s evidently planning on expanding into the US, selling the devices at $200-400, cross-subsidizing the cost for users in India.
The Aurolab example raises two important points. One is the importance of being able to do basic R&D in social technology. It’s easy for funders to get excited about a project once you’re selling $45 hearing aids to poor people; it’s far harder to get a funder to put up huge sums to fund the research that makes the device possible.
Second, it forces us to wrestle with the idea of differential pricing. Jim points out that social activists folks tend to talk out of both sides of their mouth on this issue. We pressure big pharma to offer low cost drugs to people in developing nations – to address issues like AIDS, we’re going to need differential pricing. On the other hand, differential pricing tends to be politically unpopular. It offends a basic sense of fairness that an item costs one price for one person and another for a poorer person. It also opens up the very real problem of reimportation – goods sold at one price in a poor nation can be reimported and sold at a higher price in the US or Europe.
Jim generated the most controversy when he talked about comparative approaches to getting intellectual property into the service of the social sector. He offered two options: ask the licenseholder to do the right thing and allow use to take place, or change regulations to allow expections for the disabled. I suggested a third: compulsory licensing.
Jim admits that the solutions he’s currently using are imperfect. The Chafee exception doesn’t apply outside the US, so Bookshare is currently a US-only institution. While other countries like Canada have similar exemptions, there’s no mechanism to allow for exchange, permitting a work created for the blind in Canada to be used in the US.
Of great interest to me, Jim is talking about “public interest exceptions” which might permit people in very poor nations access to certain content without explicitly seeking permission. This is similar to an idea being talked about in Creative Commons circles – a license that would allow rights to be reserved in most nations, but permit widespread use and derivative products in certain poor nations. The holy grail is a library of content that people in poor countries could print on demand for $1 a copy, a project that Internet Bookmobile is already working on, though with texts that are already out of copyright.
(Jim points out that Creative Commons could also use some small license changes to make content more accessible to the blind and disabled. CC folks have generally been open to the idea that content should be available to the disabled without undue restrictions, but changes in the license code might make the process easier.)
Urs Gasser, a fellow Fellow and one of the sharpest legal minds I’ve ever encountered, had a pointed question for Jim regarding creating multiple exemptions to copyright law – isn’t this just patching a broken system? Aren’t there other rights – like fair use – that DRMs prevent people from accessing? If we carve out special exceptions for the blind, are we giving up on the bigger fight of challenging all restrictive DRM regimes? (No, we didn’t come up with a good answer, either.)
A dense, fascinating, important talk. Thanks for joining us, Jim.