The Pace is Quickening
A major premise of the Kurzweil book is that technological change accelerates exponentially, and that even the exponentials are accelerating. Each epoch builds on the last. Epochs are getting shorter.
Kurzweil focuses on complexity, noting that evolution produces increasing order, and that technology can extend evolution by building ever-more-efficiently on this order. Very quick feedback loops are all around us, pushing the rate of technological change along and producing faster and smaller devices. Meanwhile, biological evolution continues, but at such a slow rate that it hardly matters.
He boldly predicts that computers as we know them will disappear by the end of this decade, to be replaced by virtual reality environments. No more offices by 2020. He suggests it's time to invest in tiny sensors and natural language search engines that can topple Google.
He's probably right, although it may take more than 20 years for all of this to happen.
[Really — why can't we ask a question of all of this online information and get back something meaningful? Search engines right now are great at finding things we are already confident exist, but not so great at helping us put information in order so that we can increase our own intelligences. Training in pattern recognition — that's what we need search engines for.]
Even if he's only half right (or even less than half right), Kurzweil's work suggests that it's a good time to be alive and interested in the effect of technology on human beings.
Up with Up
Someone told me yesterday that the battle over internet access was “cable's to lose.” In other words, cable companies have such a great position in the marketplace that all they have to do is persuade content companies to trust them. (Apparently cable royalties to content guys are a bit black-box-like — so the people with content are predisposed not to trust. I'm sure there are other reasons as well.)
Cable wants to provide the internet “triple play”: voice, data, and video.
But here's a problem with this scenario: Cable doesn't care about uplinks. Cable companies provide asymmetrical connections, with download speeds that are much higher than upload speeds. Much, much higher. (Telephone companies also do this, but the differential is higher, generally, for cable.) As CNET reported earlier this month, this is a huge problem for users that actually want to build and share things online, as more and more people want to do.
Citizen journalism? Forget it. Can't get there from here. Video blogging? Can't be done.
If users come to expect slow uploads, they'll be content with … slow uploads. They'll be swallowing “triple play” material managed for them by cable/content behemoths.
So I'm proposing some slogans (send in suggestions, please).
Up With Up!
Up Counts
They Better Care About Up
We Are Up
Up.
New book excitement
In my mailbox today was an enormous package with an enormous black-covered book inside it: The Singularity is Near, by Ray Kurzweil. I'm pledging to you here, right now, that I will read this right away and report back. I almost missed getting off the subway today (the urban equivalent of sitting in the driveway listening to a show on your car radio that you just can't bear to leave) because I was diving into the prologue. Note: not “introduction.” “Prologue.” This book is a theatrical event.
The subtitle of Kurzweil's new book is “When Humans Transcend Biology.” His premise is that humans have the ability to understand, model, and extend their own intelligence, using the computational power of machines to help.
We're trying to model e coli, so why not model our own minds? It's a big big book, so part of my plan in telling you about its dramatic arrival in my life is to set the stage for some later posts. The chapters go off on genetics, nanotechnology, robotics, diet (we knew he'd go there), and many many pages of responses to critics.
Deep breath. Onward.
Who's Protecting Whom?
USIIA today released a white paper that outlines ten principles for broadband policy and suggests that there should be an industry self-regulatory organization for broadband questions (that would handle conflict resolution and drive consensus on policy, among other things). USIIA's white paper does not contain good news for internet users, as I'll explain in a moment.
But I'm conflicted about the essential move that USIIA is making, which is to assert that “the industry” can manage its own house. In general, I'm all for industry self-management. Here, though, it looks as if providers of transport are ganging up with providers of content to ensure that all internet experiences are carefully managed. Who gave transport providers the right to cut across layers of the protocol stack and dictate what applications run on their networks? Answer: telephony and cable history. But that's not the same as the history of the internet, which is full of aspirations to not discriminate against other layers.
This self-regulatory move is different from, say, the ICANN model. ICANN is organized to manage consensus on global domain name policy, and to coordinate entries into the root. But ICANN carefully does not get into other layers, and nor do domain name registries. Com doesn't try to manage the services that .com registrants can use or see.
So although I'm all for industry taking care of itself through contracts, as long as there is adequate competition and real choice, there are at least two problems with the USIIA model: First, it's masking an aggressive push towards fully managed networks in which transport and content are fused. It's not so much “self regulation” as “cartel creation,” and it's based on magnificent layer-crossing. Second, the provider principles are founded on the assumption that adequate competition exists in the broadband world in the US, and I'm not sure that's true. Take a look at this quote from p.8:
Consumer advocates have for the last decade pressed for government mandates and subsidies that would eliminate barriers to entry and bring an unlimited stream of new competitors into the broadband markets. This is done for two reasons, both incorrect. The first is the belief that there is in some way a dearth of competition in broadband when in fact most population concentrations have a choice of four or more platforms with at least one competitor in each.
Is that true?
USIIA is suggesting that “the broadband industry” needs to transition to a “consumer-focused” marketplace in which external legislation and regulation isn't needed. There's a coy suggestion n.8 on p.12 that USIIA might itself be the right self-regulatory body for broadband issues:
It may be assumed that the US Internet Industry Association would serve as the industry body, since it already has non-profit incorporated status, a strong reputation for service to the Internet industry and a recognized presence in the industry.
USIIA's principles add five provider-oriented bullet points to the four that were foreshadowed by the FCC earlier this month (plus one about spyware services, just to balance the slate between “consumers” and “providers”). Here they are, together with questions they raise in my mind:
1. Network providers are entitled to interconnect with broadband transport services and other network providers through agreements and commercial contracts based on volume, terms, points of connection and other established market parameters;
Is this a principle that would cut against the idea of providing commodity transport? It's a moot point for the moment, based on Brand X and the FCC's declaration on DSL treatment, but why is this a principle that is good for users?
2. Network providers are entitled to offer vertically-integrated broadband access and applications;
How does this principle relate to end-to-end? If something is successfully vertically integrated, users don't get to choose whether or not to use it. It's the only choice.
3. Network providers are entitled to offer proprietary services to their customers, so long as the provision of such services does not interfere with the customer's right to access content and run applications of their choice.
Is the “so long as” meaningful? If a provider's terms of service say “you can't do X,” won't that trump whatever “rights” the user has in the abstract? If contracts are to be respected, surely the EULA is the most respected of all.
4. Network providers are entitled to offer managed services to their customers;
What's the difference between this and No. 3? And what happened to the “so long as” proviso in No. 3?
5. Network providers are entitled to operate on the same basis as other network providers in the same market without interference from local governments or political subdivisions.
This seems to be a shot at derailing municipal broadband efforts. Why shouldn't cities get a chance to do this if they want to?
These principles, together with the FCC's four, seem to point in just one direction: a new internet, carefully managed and monetized for the benefit of network managers. Why is this better?
The USIIA work deserves a close look, but it doesn't seem like good news for anyone other than large incumbents who want to make deals without interference.
Sense and sensibility
Jerry Falwell's sensibilities were injured by a site calling itself www.fallwell.com. The site clearly said that it had nothing to do with Falwell, and was critical of Falwell's views — it also received only 200 hits a day.
After a district court found that the name was infringing Falwell's trademark rights, and ordered the site to transfer its name to Rev. Falwell, the site owner bravely appealed. Today the Fourth Circuit acted with enormous sense, and issued an important opinion [pdf] striking down Rev. Falwell's claims.
Among other things, the opinion makes clear that you need to consider domain names in context — here, there was no question, once you got to the content of the site, that you weren't dealing with the “real” Falwell. And the opinion is rightfully skeptical of “initial interest confusion.”
(The idea behind “initial interest confusion” is that if you persuade a consumer to follow your signage [domain name] thinking that they're going somewhere in particular that isn't your site, even if by the time of purchase from you they're absolutely clear about what they're buying, something has gone terribly legally wrong. Right — it doesn't really make sense.)
Congratulations to Paul Levy of Public Citizen, and all the lawyers and law professors who weighed in on this case.
Paul Graham essays
One of the joys of this past weekend that I didn't write about (because I was enjoying it too much to type) was a session by Paul Graham about writing essays. I came into the room as he was urging people to have the courage to delete large chunks of text, and I stayed.
Someone asked, “You always get slashdotted. What's up with that?” Graham had a graceful response, but he could also have said, “Well, I write well-crafted essays full of new ideas. And my essays aren't too long.”
He talked a good deal about controversial subjects that he doesn't want to write about, many of which are gender-related. His style in the room was relaxed and conversational, and there were people leaning up against the walls all around him. It was a great session, and I'm looking forward to reading Hackers and Painters.
Foo morning cont.
Danese Cooper talks about women in open source. Why are only 2% of open source project people women? She and her colleagues are looking for concrete information about what kinds of people self-select (and are happy) into open source projects. She talks about the Debian women project, which affirmatively welcomes women without letting them lurk, and finds them mentors. Someone quips: “This is Women 2.0!”
Danese says there's a fear that bringing in women will make open source projects less fun. There are lots of trolls on women-oriented development lists. People discuss what to do with lists like these. It takes leadership to change the list environment. (Danese notes that she was the first woman interviewed by Slashdot and many of the questions were inappropriate. Men don't seem to notice this.)
Someone suggests that Apache could change its practices to be more acceptable to both men and women. Someone says that these same issues will hold back developing country newbies. Several people agree that mentoring is the real issue.
Someone talks about Maria Klawe, the dean of engineering at Princeton, who has done studies about getting women to stay longer in computer science. Forced mentoring doesn't work, but when freshman curricula is all labs in programming pairs, both women and men stay on. There's a discussion about fathers teaching daughters.
Danese wants O'Reilly to do an entire track on this subject at some future conference. She suggests that actually working on the issue, rather than just talking for an hour or two, could be helpful. She's interested in the summer of code, and what will happen there. Danese is starting a women's effort connected with opensource.org, and urges people to subscribe.
Foo morning cont.
Joi Ito and Geir Magnusson talk about making money with open source. Joi says you can compete with numbers of customers, not code, when you have an open source business.
Geir works for IBM because he came from a company providing support for people using Apache Geronimo. IBM bought the company, and is keeping the same model. Later he says that enterprises want the support services, because “they want one neck to choke.”
Joi says he knows a Japanese bank going to Linux and all open sourceware. They've cut the cost of the bank to 10% of what it was, and that investment has paid out. It's easier to save money than make money, someone notes, with open source businesses.
Joi asks: is it really true that big companies re-release the modifications they've made to open source? The response: big companies like interactions between professionals, so they'll work professionally with the open source community; IBM does a good job as being a steward; neither big companies nor open source people are idiots.
Joi says St. Gallen did a course on management of software and published all the results as case-studies. The companies didn't mind having their competitors see this information, because they weren't competing on this basis. The lesson is that sharing information (like open source software) often has no competitive impact on a company that isn't primarily in the business of developing software.
Sometimes people release things under open source just to get market share and put competitors out of business.
Joi notes that there's an interesting relationship between open source and open standards. Ability to influence a standard can have enormous effects. Bram Cohen notes that he's reluctant to add new elements to the BitTorrent protocol that don't make sense from an engineering standpoint. HTTP 1.0 was straightforward; HTTP 1.1 went into a huge list of features and was a disaster from a software engineering standpoint. Protocols aren't being worked on well.
We drift into talking about open source and protocols. There are lots of BitTorrent clients that don't use Bram's software. Bram stewards the protocol and provides technological leadership.
Chris DiBona notes that standards bodies take enormous resources to follow, much less dominate. He talks about a process that can fix particular problems by just having Yahoo!, Google, MSN engineers talk to each other (eg, “nofollow”). Things can emerge that don't have the blessing of standards bodies but that act like standards. Broken/expensive standards don't get chosen.
Joi says that the rest of the world is just at a tipping point with open source. Brazil is getting it, but other countries aren't there; they're still looking for certificates and authenticated things. Peope get fixated on proprietary material. Similarly, companies like to set rules and open source often gets filtered out. Not paying is psychologically scary (”you get what you pay for”).
Having the best product isn't the answer in software sales, says Joi. Often it's having the sales channel or consulting services — and bragging customers.
The open source companies in the room care a great deal about the terms of the license associated with their software.
There are companies that are Windows shops just so they can use Windows calendar. Why is there no open source replacement for this? Chris DiBona: Exchange is easy to install, easy to use, lots of devices only talk to this program. This battle will go on for another 30 years.
Calendar sharing, contact book management - all this is a big issue with open source. It's not happening. Pulling something together to solve these problems is hard. There are so many independent entities.
Joi says the customer can drive this. It's harder to do a Chandler than for a bank CEO (or the army) to fix this himself. A bank CEO can assign 150 people to work on the problem. But the open source community isn't working effectively on synching yet. And MSN is giving away product to avoid the open source threat (this is just a rumor).
Salesforce.com did a tool that manages 75% of customer relationship management — sometimes doing just enough is…enough. They made a fortune. We need open source calendaring that is “good enough.”
Foo morning
Saul Griffith shows some howtoons, along with some video of kids enjoying building things. Howtoons are coming out in books soon. He says 10-year-old boys all want to build flying skateboards. Ten-year-old girls all want machines that will allow them to see their friends' dreams. He hopes some of those girls will grow up to be neuroscientists. He talks a lot about liability risks. In the question period, people talk about how to scale Saul's operation (he does evenings at science museums for kids). Someone points us to sas.org.
In general, everything SquidLabs is cool. Take a look at instructables.
Foo afternoon
Two very interesting discussions: open source VoIP technology, and Identity 2.0.
1. Surj Patel talked about mergers of online applications with telephony. Some examples: amabuddy — amazon over voice (great); dodgeball — social app with hacked free sms; google sms — over wireless voice next?, and more. He predicts that software will get better and punks will start writing cgi scripts for their own phone-network applications.
Surj himself is building an open source linux cellphone. It's made of inexpensive components. He predicts he'll have a prototype in 3-4 weeks.
2. Nivi introduced us to “email for voice”: slawesome.com. He just put it up a week ago. You can also use it to make podcasts. He wants to take text and make it into interactive audio and video.
3. Brian McConnell talked about providing on-demand access to any audio resource, using Asterisk as the telephone part. He's created a telecast system. Call 415-368-3022. You'll hear an English radio station that's coming from somewhere in Switzerland.
4. Blaine Cook talked about the slowness of the telephone companies. For example, someone discovered that TMobile voicemail uses caller-ID for authentication, and caller-ID can be spoofed. TMobile offered to hire the person who pointed this out (but TMobile didn't fix the problem). Everything can be spoofed — ANI databases (central to E911) — and so the rules by which the telephone network works are crumbling.
Last year during the Republican National Convention, Blaine and his colleagues did radio streaming so people with cell phones could listen to radio news. They did text-to-speech, so that people could get voice messages about what was happening. Same during the election — working with four people for 20 days, they built a system that allowed a voicemail message to be delivered to 10,000 people in 20 minutes.
2. Identity 2.0: Dick Hardt gave his celebrated presentation. His view is that Sxip will be widely adopted because it's simple, clear, open, and allows claim-based mechanisms to determine identity. He said that talking about identity now is like talking about the net in 1993 — everyone needs really basic information to grasp the concept.
Brian Fitzgerald talked about OpenID, and a lively conversation ensued.
It's very hard to blog this conference — there are so many things going on at once, and everything is interesting.