Incoming Links
Earlier today Vinod Khosla remarked that the hubris of the engineers at Excite caused them to dismiss the opportunity of buying Google’s technology for $1,000,000. Well, after lunch I met Joe Kraus who was president at Excite back then. He was pissed because the Vinod left off a couple of zeros from that figure. No love lost there…
This session is a showcase of what the various research labs are up to.
Gene Becker HP Labs. Google, Yahoo, HP makes me think of that old song “One of these things
is not like the other!”. [laughter] Here to talk about ubiquitous technology. The web is becoming: diverse, social, creative, mobile, contextual, physical. Here are some examples:
Finally, onto utility computing. Experiment with DreamWorks - a utility rendering engine. Put together a few thousand computers to be used as an overflow service by DreamWorks. DW don’t need to buy computers only to have them idle for months at a time. Can also enable access to small artists - renting time. [ID: like a recording studio?]
Next up are Usama Fayyad and Prabhakar Raghavan, chief data officer and head of research at Yahoo! respectively. Yahoo announced major increase in research recently. More innovation at a faster pace. Yahoo launches several hundred products a year.
Usama Fayyad: Creating the science of the web: merging product innovation with world-class science. Developing a new science: explaining communities, navigation of huge data sets. Yahoo Buzz uses market forces to predict success of new technologies. Better than monitoring searches - get much earlier indication.
Prabhakar Raghavan: Yahoo! mindset. no monolithic search engine. Includes a slider labelled shopping on the left and researching at the right. Sliding the bar dynamically changes rankings of results from shopping oriented to review oriented. Not limited to those two extremes, system could also work with family friendly or adult for example.
Next Prabhakar demos TagLine. Flickr has about 100million tags. About 1 million distinct tags. Mined data to find most bursty tags.These are tags that started off with low frequency and shot up in popularity very quickly. The flash demo shows a timeline and floating tags as the demo moves from 2003 to present day. Tags with images fly by indicating the bursts at that date, typically major events and incidents.
Alan Eustace,VP of engineering at Google and Jason Shellen. Google have a rule that you scale the number of things you do with the number of people. If you double the people you should do twice as many things. This drives small teams.
Sex Identification - can machines do it? Have 2 billion images in search and a huge number of personal images in Picasa. Uses a statistical model. Potential to identify same person in different pictures.
Information overload - how do I keep up with blogs? Google Reader.
A conversation with Vinod Khosla. He has switched from investing via Kleiner, Perkins and is focussing on investing his own money.
Vinod: a new class of utility and applications has emerged, beyond ecommerce and clickable sites. Some of the peer-to-peer phenoma such as Flickr and Skype. Don’t include Skype in Web 2.0 - it’s too linear. Kids live in a Media Mangle. They move between media very easily. Talking to Fox, they didn’t realise how much kids used IM while watching TV.
There is still plenty of money out there for startups. Remember though that just because you got funded doesn’t mean you’ll be successful. Advice is to experiment. Too much money is bad for a startup because they don’t examine their business plans in enough detail.
The older a company gets, the more traditional it gets. For example, with Excite, people became over confident. He convinced Sergey and Larry to sell their technology to Excite for about a million but Excite thought they could do so much better themselves so rejected it.
On blogging:much rather read the blogs than CNN. But, would rather use IMDB than any individual review. It’s about trust.
On mobile :Just like we have Web 2.0 we have Mobile 2.0, things that go beyond SMS. Hard to tell what it will be but mobile search and local search will be big. Media Mangle will be big here. Look past ringtones and simple games to what’s next.
Open textbooks in the style of wikipedia would be huge. Wants people to step up to do it. [ID: could he possibly be thinking of OpenCourseWare, or WikiBooks? Someone send him those links]
Question: all the talk is about PCs, have you seen any talk around simpler devices? Voice and text are important. For a device you always have with you it will not be the smartest or the largest. The most interesting thing will be the network proxy, so client can be thin and low power.
Question: about the future of search engine technology, what do you think will be coming? One thing that hasn’t been applied enough is peer or collaborative data.
Jason is from 37signals, and gave us an informal talk about Less. He suggested five ways to do less
Less money. Traditionally you write a business plan and get some money. This is not recommended any more. You end up with debt which is always bad. Hardware is so cheap it’s almost free. Money can’t buy you passion. All it buys you are salaries and people. Perhaps you need less people.
Less people. I think 3 are ideal. If you think you need more than perhaps your product is too complex. You just need a designer, a coder and a sweeper who goes between the two and has some product sense.
Less time. Work less, shorter hours. It’ll focus you on what’s important.
Less abstractions.Just build things, learning as you go. Get rid of the functional specification documents. Instead, build the product starting with the UI, the user experience. That’s what defines your product: customer experience.
Less software. Less features, less support. Do simple things like del.icio.us or upcoming.org. Don’t solve the complex problems. There are plenty of simple ones to cherry pick
There is one thing you need more of: more constraints. All of the above will constrain you to produce more quality
A panel run by Tim O’Reilly involving Danny Rimer (Index Ventures) , Jeff Barr (Amazon), Toni Schneider (Yahoo!) and Sam Schwartz (Comcast Interactive Capital)
Tim: clearly to make money you have to own somrthing. What is it? Jeff: we give out a lot of data for free. We own the process of people actually getting the items, distribution, fulfilment.
Toni: we own the user’s trust. There are lots of ways to get data out of Yahoo. Tim: you own the metadata too? Toni: there are types of data we can’t contractually open up.
Danny: openness is a direction all these companies want but it’s a matter of maturity. We’ve opened up Skype messaging through API. Tim: will you interoperate with VOIP from Google, MS or Yahoo? Danny: can’t answer that, perhaps eBay will decide to do that.
Sam: we’ve come from a traditionally closed environment but our broadband business is different and we’re being more open.
Toni: it’s important that when you open up you don’t just throw it out there, you need to support it. This is why Yahoo! set up the developer network first. Openness goes both ways, contributing toYahoo and allowing reuse.
Tim: as VCs, what are you looking for? Danny: innovation is happening outside of the valley, Israel for example, and look to China in the future. Most companies we back are free services. Free is important, mySQL is getting a third of the downloads of Skype and it’s an enterprise product!
Sam: split venture work into two buckets: infrastructure and applications to showcase them.
Toni: a lot of people are building without worrying about business models. They’re doing it for recognition and to get bought out by Yahoo!
Danny: I disagree. In my opinion these kinds of acquisitions are a cheap way to lock in talent and use it in-house.
Jeff: seen developers audition for jobs by building something cool and saying if you hire me I can do this for you.
Tim to Danny: why eBay? Danny: there were a number of options. I suspect that eBay want to be a bigger player in the Internet space.
Speaker from the floor says that 22% of all auctions on eBay are created by third-party tools enabled by eBay web services. Toni: important to remember that these kinds of services are primarily created to support an existing business model. We need to do more to support genuine innovation in these areas.
Hossein Eslambolchi, from AT&T spoke about the new and old worlds of telecommunications. Traditionalists say the intelligence should be in the network. IP bigots say we make the network dumb and have smart endpoints. But there needs to be a middle ground.
He characterises the two sides as the forces of control (old world, circuit switching) and the forces of freedom (new world, packet switching)
Contrasts between the old and new worlds:
Dave Sifry, from Technorati, is up on stage. Technorati now tracks 18.9 million blogs. 70,000 created daily, about one a second. 55% are still active 3 months later. 8% are spam blogs. Technorati can get them out of the index quickly. There are huge spikes during election, launch of MSN spaces and now new blogs from China.
Posting volumes are growing and has identifiable spikes for world events: london bombings, katrina etc. Blogs are encroaching on mainstream media when measured by incoming links. Different stages of dealing with blogs: denial, anger, now into acceptance phase.
Almost [ID: only!!] a third of posts are tagged or categorised.
More than search. It’s about exposing community and context.
Launching today: use authorative bloggers to scope your searching. Demonstrates search on bush and filters by tagged sets of blogs. Let’s you ask what the Katrina bloggers are saying about Bush, or the gardening blogs!
Mary has a reputation for going fast… so here goes! (slides here)
Key data points:
Conclusion: first ten years (1995-2005) of commercial
Internet were a warm up act for what is about to happen
Tim: Jonathan you’re probably one of the highest placed bloggers around. How do you find the time and why do you do it? Jonathan: it’s key to community, transparency is good, how can a community form if you wall yourself off?
Tim: it’s hard sometimes, I think twice before posting stuff that gives away ideas to competitors. Jonathan: I’m never unauthentic on my blog, but I act as I should as an executive officer. I cause palpitations within Sun occasionally!
Tim jumps randomly again and asks Mitchell about the relationship with Firefox developers that now work for Google. Google have a strong information firewall. Mitchell: if it’s in the browser then we share the information. We’ve always had a good flow of information between developers, but the marketing information is harder.
Tim calls for questions from the floor and Mitchell wryly suggests that questions to the both of them would get extra marks. She, like me, obviously feels that Tim’s style has been to conduct two interviews at once rather than interviewing two people together.
Dare again: the rise of AJAX spurred by Firefox copying key features from IE. Will you copy in the future? Mitchell: not everything. We got a lot of heat for not supporting ActiveX but that turned out to be the correct decision.
Dare: why do you dismiss a web based version of office? Jonathan: I’m not dismissing web based office but there are many hammers for different nails. What would the advantages be? No one technology solves every problem. Would you implement Firefox in AJAX? Where do you stop?
Question: how do you make Web 2.0 and giving everything away sustainable? Jonathan: OSS forces you to figure out whether you’re going to make money from services. A few things will be true in 5 years as now. Content will be valuable still. Intellectual Property is key. If you’re Dell, just distributing other people’s IP, then you’re in trouble from WallMart.
Next up we have a panel on Open Source futures run by Tim O’Reilly: Can Open Source Stay Open? The participants are Mitchell Baker from Mozilla and OSAF and Jonathan Schwartz, COO of Sun who seems to be delayed somewhere so we just have Mitchell for now.
Tim O’Reilly notes that the killer apps for Linux are not things like the Gimp but are web applications like Google and Amazon. [Jonathan turns up]
Tim: some say Sun is late to the OSS party and yet you were founded on those principles. Jonathan: one way to think about OSS is as intellectual property distribution. Don’t focus on the code, price (free) is more important. Code has no value inof itself. Combine distribution of code with services around it. Tim: Sun is a hardware vendor. Jonathan: no we’re not, didn’t you see the announcement with Google, that wasn’t about hardware.
Tim gets flustered after some rather awkward exchanges with Jonathan regarding Sun’s declining position as the company the Internet runs on.
Tim: one of the benefits of OSS is that the developer can “just do it” without any barriers to entry, what’s your take? Mitchell: Mozilla’s trump card is the power of the web. People can change firefox, but the average person won’t.
Tim: Greasemonkey raises questions about who is in charge of the user experience. Will that take hold? Mitchell: part of it will take hold but the average consumer is confused and doesn’t know the difference between the browser, the internet and the search box. So Greasemonkey won’t take hold in general but we can extract the germ of that idea and put it in the core.
Tim: Jonathan, what was in between the lines of the Google/Sun deal. Jonathan: power is in the community that can contribute to the future success of the product. Also power of distribution, i.e. how many are out there. Sun have a lot of technology and Google have a lot of distribution on the web. Sun have distribution to the desktop via Java. OpenOffice will get ability to cross search other documents but imagine if you could save your documents on the network. [hint, hint!]
