Off to etech

I'm going to San Diego for the O'Reilly etech conference later today. Not that I post frequently, but I'll be posting even less.

I used to enjoy going to IBM's internal WebAhead/Get Connected conferences, less so for the presentations but for the conversations with people. They were a witches brew of ideas out of which a number of products and solutions were developed. IBM doesn't hold WebAhead any more (cost cuts, travel restrictions, repercussions from Todd's rendition of "Time Warp" at WebAhead '99, etc.), which is a shame. As much as I believe in online technologies as a way to connect people, nothing replaces face to face get togethers.

I have absolutely no goals for etech other than to meet some people and stir some creative juices again.

etech Tuesday morning session

Rael Dornfest

• "remixing" -- that "prosumers" and general consumers are opening their boxes and exploring
• Going behind the curtain.
• Syndicated e-commerce means you don't need to own every component of your application.
• "All progress depends on the unreasonable man" --George Bernard Shaw via Rael Dornfest

Tim O'Reilly

• Tim o'reilly: pattern observation. Design patterns and internet applications.
• Pattern language -- Christopher Alexander.
• "Users add value to shared data"
• "make participation the default aggregating user data as a aide effect of their using your application (shirky via o'reilly)"
• Don't design for single device anymore
• Capture and share the social fabric underlying the application, rather than artificially constructing another
• Why is it that enterprise software has to be continually reinvented while tcp/ip is 30+ years old. That it has to do with the core packet idea of TCPIP.
• Ora into data visualization (look at treemap)

Stewart Butterfield -- flickr

• Web services as startup strategy
• Issue: loss of control over pace of things happen by open web services, impact on scalability
• ~250,000 api requests / day 3m pageviews / day

Danny Hillis -- Applied Minds

• They build stuff to 1.0 then hand off to licensees or other developers
• Demo of graphical map table and new 3d map table
• He echoed the ideas of building services ased on user contributed content. The value of the service increases as the amount of content is contributed.
• Metaweb -- sharing and rendering of public data in public space

Jeff Bezos -- Amazon / a9

• Syndicated search via a9
• Modified rss to add three tags to indicate # of items returned, current index and next index
• "Channels" of search results.
• Can select specific data sources for search and rank them in order of preference
• Can turn on/off displays of results from data sources
• Notion of "consuming search" -- connecting to 3rd party search services and getting results (via RSS with above modifications)
• Aggregated search results
• Aside: He clicked "Yes" when asked by Windows Update if he wanted to shutdown.
• Make it possible to integrate 3rd party search into their interface via RSS
• OpenSearch RSS

Todd says throw the book at Bernie E.

Turboville: The Rise and (Hard) Fall of Bernie Ebbers: Not only did they throw the book at former WorldCom CEO Bernie Ebbers, they reinvented what the book looked like. [...] This verdict has teeth, and it demonstrates that when chief executives and other high level management conduct themselves in a manner unbecoming to business professionals -- unethical or illegal – there is a price to be paid.

etech Tuesday morning session pt 2

The second half of the morning covered in the labs briefings from Microsoft, Yahoo!, Google, and AT&T as well as a fun bit on John von Neumann by George Dyson.

Rick Rashid talked for Microsoft...focussed on devices they are developing, one of which a person wears in order to capture (and digitize) as much information as possible about the person's environment for further use. Another, like the map table at Applied Minds, was a desktop surface which combined real time image analysis with a projector so you could use direct physical manipulation of objects which may have other images projected onto them from the projector. Eg: a sheet of paper is turned into a viewport for a set of images.

Gary Flake talked about Yahoo! Labs and the work they are doing there. He introduced a new game called the Tech Buzz Game which Yahoo is using to tap the collective wisdom of the web which I initially thought was comparable to Slashdot's moderation and meta-moderation, but is instead more comparable to the delphi coracle of John Brunner's Shockwave Rider. There's a couple things going on, first they've defined a new market auction concept: Dynamic Pari-mutuel Auction. That this works by having people predict the popularity (or obscurity) of various terms in search land, in the specific case a Buzz index. If your prediction is correct you get more "dollars" to spend and bet on further predictions. They developed this in tandem with a company named NewsFutures. That it's an electronic market in outcome prediction. The market is infinitely liquid and does not require a market maker (these are my notes, not necessarily my opinions).

Peter Norvig talked about Google Labs. I...I didn't take many notes, which generally means I wasn't interested or didn't find anything to note down. He demoed a personalized version of search but it wasn't clear if it's live yet. Basically you slide a scale ranging from no personalization to (I guess) high personalization based on a profile you've completed. The search results are then filtered based on the intersection between your profile, your personalization level, and the actual results.

The only bit I'll note from George Dyson's talk was this quote which I'm certainly mangling: Real Progress is crossing phenomena which he may have attributed to Nils A. Barricelli though I wasn't too clear on that.

Kevin Kealey presented from AT&T. Mostly about spam, and the realization from AT&T's perspective that all this crap is weighing down their networks and costing them real money. So they have developed technology (which he's going into further in a session I just missed) to filter spam/spim/spit and bots and whatnot at the network layer.

Afternoon etech: content free notes

Nothing much to post for the afternoon. I missed the first half of the sessions dealing with the hotel.

I managed to get bac in time for Sam Ruby's session "Just" Use HTTP. The crux of the session is that there are many, many specs, and that they all build on the base protocols and standards like HTTP and Unicode. HTTP itself is vague on various matters like character encodings for content, which is less a problem for people reading web pages than for applications consuming web services. HTTP also doesn't specify the character set for URIs (as an aside: I tried to use URI inside IBM for years and was frequently flamed on the various WWW* fora...it's nice to see if finally used).

I'm also sitting in on Amazon.com: E-Commerce at Interplanetary Scale which....which I'm not getting anything from.

Good idea: trackback on session pages

O'Reilly, sponsors of the etech conference, have set up trackback pings for the various session pages. Good idea (personally, I think any page which can be linked to should have trackback and/or pingback enabled, except for the stupid trackback spam problem). Only problem: it doesn't work. The previous post included links to two session pages. The trackbacks to those session pages returned errors:

Ping 'http://www.oreillynet.com/cgi-bin/tb/tb.cgi/e_sess_5974' failed: HTTP error: 302 Found

Which is just the result in MovableType, the trackback doesn't show up on the page so I'm not sure if it was recorded and I just can't see it, or if something else is going on.

(Yes, I could try debugging it, but not from the conference floor)

Remix

The theme of etech is Remix: the notion that there's all this stuff out there (data, services, applications, even physical products like Tivo) that people are taking and remixing, creating something new, possibly useful, certainly interesting derivations. Much of the content industry is aghast at this sort of thing, the RIAA being the classic example in going after mash-up DJs like The Kleptones or Beatallica. This week's The Economist has an article fortunately timed to this conference: Economist.com | The future of innovation The rise of the creative consumer. How and why smart companies are harnessing the creativity of their customers:

How does innovation happen? The familiar story involves boffins in academic institutes and R&D labs. But lately, corporate practice has begun to challenge this old-fashioned notion. Open-source software development is already well-known. Less so is the fact that Bell, an American bicycle-helmet maker, has collected hundreds of ideas for new products from its customers, and is putting several of them into production. Or that Electronic Arts (EA), a maker of computer games, ships programming tools to its customers, posts their modifications online and works their creations into new games. And so on. Not only is the customer king: now he is market-research head, R&D chief and product-development manager, too.

At the heart of most thinking about innovation is the belief that people expect to be paid for their creative work: hence the need to protect and reward the creation of intellectual property. One really exciting thing about user-led innovation is that customers seem willing to donate their creativity freely, says Mr Von Hippel. This may be because it is their only practical option: patents are costly to get and often provide only weak protection. Some people may value the enhanced reputation and network effects of freely revealing their work more than any money they could make by patenting it. Either way, some firms are starting to believe that there really is such a thing as a free lunch.

Comments off, and no ey.

Comments are hosed again. I'm in the process of junking the MT cgis entirely and switching to PHP for the mt-comment and mt-trackback scripts but it's not a priority. If you need to comment, send me a note via the handy dandy contact form or email if you know one of my email addresses.

Also, my eyboard seems o be having some problems with cerain eys. his sared some ime his morning and as a resul i loos lie I can' ype a all. I'll fix i in pubs when I return to New Yor.

etech day two

Let me start by admitting that choosing to stay at the Hilton San Diego by the airport was a huge mistake. First, on Monday when I arrived they didn't have a room for me. Seems they lost several rooms due to an unspecified problem. So they shipped me off to the nearby Sheraton for the night.

Yesterday morning I was able to switch back before heading into town for the conference. Now, taxis to and from downtown San Diego cost $10. Not bad, but it takes forever to get a taxi (somewhat surprising given the massive queue of taxis at the airport, I guess that when you call they dispatch someone from downtown or further, like L.A.).

Today I called for a taxi and it took nearly 30 minutes...tomorrow I'm toying with walking to the conference.

I saved some money by staying at the Hilton over the conference hotel, but not much when considering the inconvenience.

Anyway, notes from this morning's session in the next post

etech day two, morning sessions

Neil Gershenfeld

The morning opened with Neil Gershenfeld from MIT, talking about manufacturing technology using computational assemblers. That is, getting the systems to manufacture components directly, without the manual intervention that is normally required. So, you hook a bunch of CAD/CAM systems together with various devices to automate manufacturing.

He also talked about spreading micro-fabrication labs to places in India, Ghana, the Samis of northern Finland. That there's value in teaching people how to build the technology directly, not just the abstract technology concepts. Ie, give people the ability to build computers, don't just send them the computers as-is.

After Neil there was a discussion about this hardware type of hacking. I didn't take many notes but captured this quote from someone on the panel: Put creativity in the critical path.

Cory Doctorow

Cory Doctorow gave a great speech starting with the problems we see with technology today like spam, and possibly logical solutions like charging for email and how that only creates further problems because it raises the cost of email for everyone. He then went into a riff on DRM and how DRM is simply a system of control, which collapses if any one component is compromised. In the interim it serves to constrict technology development.

Complex ecosystems are influenced, not controlled

He's put the text of the speech online: All Complex Ecosystems Have Parasites.

Justin Chapweske

Justin Chapweske presented on Swarmcasting. Unfortunately I think his session got picked for trimming (because the others had run long) so he sort of had to rush through. Basically swarmcasting is some technology to sit on top of HTTP and enable distributed downloads alá BitTorrent, except that you replace your HTTP stack with the Onion Networks stack so your applications don't need to be modified to use the technology.

Jimmy Wales — Wikipedia

Jimmy took the audience through Wikipedia. I don't have much to note about the talk, however it lead into a great group discussion on folksonomies.

Panel discussion: folksonomies

The panel consisted of Jimmy Wales from Wikipedia, Stewart Butterfield from Flickr, Joshua Schacter from del.icio.us, and Clay Shirky from NYU's ITP and other places.

This is just a collection of notes and quotes:

• When wikipedia introduced its taxonomy it was chaotic for several weeks in the English articles but settled down eventually.
• Tags are for noting aspects of things and not necessarily a replacement for taxonomy/classification
• Flickr tags started as retrieval method for people but became a way for people to group content
• Joshua Schacter noted that while all three groups use tags, they're tags for different purposes:
  • Wikipedia has one or more people who categorize articles into their hierarchical taxonomy. Since it's a wiki, anyone can change the categorization but the group as a whole will correct it if the consensus is that the categorization is incorrect.
  • Flickr's tagging consists of people tagging content they created as a way to recall (and possibly as a way to group content with similar photos by others)
  • del.icio.us' tagging consists of people tagging pages created by others
  • Technorati's tagging consists of authors/content creators tagging their own content.
• A questioner asked if there was a way to link all the tags together, but would that have any value since the tags may be the same word but have different meanings depending on the context they were tagged in.
• Joshua Schacter: When you categorize something it's your instinct that's the most reliable and reproducible thing.
• Question: that these systems would be perfect for RDF…why don't they use it? Joshua Schacter: that tags and RDF can work together, it's not either or. That RDF is complex and the delicious version on top of RDF was measurably slower.
• Question: wouldn't tags be more useful with various sorts of meta data (eg language). Answer: Tags are lower barrier to entry. If you encumber them and make them more complex it lowers the usability and utility of tagging.

13 things that do not make sense

New Scientist 13 things that do not make sense - Features. Here's the 13 things, read the article to learn the details (and perhaps cause your head to spin a few times):

1. The placebo effect
2. The horizon problem
3. Ultra-energetic cosmic rays
4. Belfast homeopathy results
5. Dark matter
6. Viking's methane
7. Tetraneutrons
8. The Pioneer anomaly
9. Dark energy
10. The Kuiper cliff
11. The Wow signal
12. Not-so-constant constants
13. Cold fusion

etech day two, afternoon sessions

I don't have much to report from the afternoon sessions today, generally I'm disappointed in the sessions I sat in on, two were just total fluff though they got lots of applause.

Jason Fried -- 37 Signals

One high point was Jason Fried speaking about lessons learned at 37 Signals from building Basecamp. He organized his speech around four themes:

1. Reducing mass
2. Embracing constraints
3. Getting Real
4. Managing debt

Reducing mass — keep things small and controllable. Implement changes incrementally, don't pack a lot of features into one big release.

Embracing constraints — that many organizations fear constraints and go to great lengths to avoid them, and are unprepared when they encounter them eventually. Constraints create creativity, they're where creativity happens. He cited an example that when they launched Basecamp, they had not figured out how to implement billing. They had 30 days to resolve that little problem since they had a 30 day free trial. The tradeoff was that they got to focus their time on developing the base product, rather than get distracted by the development and implementation of billing.

Getting real — they started by designing the UI and focussed on "real" prototypes, rather than churning code for awhile and then looking at the prototype.

Managing debt — covering both financial debt as well as project debt. That debt in projects is the payment you have to do sooner or later when you make tradeoffs in decisions. If you implement a hack to get around a certain problem, some time down the road it may (will) resurface and you'll have to deal with it...that's debt. It's not necessarily something to avoid, but you need to be aware of it.

He also talked about decisions...that people spend a lot of time trying to make the right decision, make it, and then thrash afterwards. Decisions are transitory events. You mae the decision you make based on the information you have, and move on. If the context changes (and it always does), the decision may no longer seem wise, or may even be proven incorrect. You can't hold off on making the decision on the fear that it may or may not be correct.

Back from etech, more later

I returned home around 6:30 this morning...I have lots of notes which I'll post after I get some more sleep (I managed to sleep on the flight but it's only a five hour flight).

etech was fun though, I'm definitely going back.