Web 2.0 Conf Day 2

(Raw notes, updated throughout the day, reverse-chronological)

Session: Dojo Offline Toolkit

Brad Neuberg, discussing Dojo Offline Toolkit. Interesting premise. They have built a 300k runtime to handle offline support, “super cookies” and “browser-agnostic cache”. I’m having trouble understanding the relationship between this and Dojo. It adds a namespace dojo.off, and changes dojo.storage. But is it intended to be bundled with Dojo? Saying things like “Here’s what Dojo recommends”, so I assume he’s speaking on behalf of Dojo. (Clarified – will be in Dojo distribution at end of this week).

Making a good point. Why do you want a merge UI? Merge/resolve UIs suck, and just confuse people.

Pretty sweet, “replay log” architecture. Implemented as local proxy, only caches GETs, doesn’t queue post. You need to enable etag support, HTTP 1.1 if you want the pages to run offline.

Session: Large Datacenters with Shipping Container Model, James Hamilton

Amazing talk. He’s going into excruciating detail about the economics that drive Goog, Yhoo, and Live. This is really deep internal information that most people don’t know about our companies. Enterprises are fundamentally different from big datacenter at scale. Enterprises consolidate to reduce admin costs. Big datacenters one or two orders of magnitude cheaper in admin costs – admin costs become negligible, so economics drive to have more systems of smaller size.

Sun shipping 242 systems in 20′ x 8′ x 8′. Rackable systems – 1152 systems in 40′; 2 million dollars. They are both serviceable.

James making the case for using that extra space for computing hardware. Don’t worry about servicing each system – same as hard drive sectors. As long as you can depower dead systems, you are better off. Lots of good Q&A. James puts another very interesting stake in the ground – not only should the details of these systems be public, they should be available for others to use just like you buy an OS today. That goes for rack/container design, but also for storage service and ops management.

Microformats: John Allsop

Telling people to try Operator on Firefox. Now he’s doing an RDF vs. Microformats (googlefight) smackdown. He is off in the weeds, talking about “wisdom of the crowds”, as if there is a relation – it is very obtuse and unsubstantiated. He seems to be arguing that Microformats put small guy more on level playing field with Google, since you don’t need Google PhDs to extract semantics. This ignores Schmidt’s comments about “we don’t have better algorithms, we just have more data”. The data center and 80% market share are the moat – not the PhDs.

Now showing XFN. Rips on IE6, admits that IE7 works with the rel selectors. Showing hCard (a bit more practical). This session is really remedial. Talking about descendent selectors now.

Keynotes: Eric Schmidt

(heavy paraphrasing) Schmidt, announcing that Google is launching presentations app to go with rest of their Office suite. He repeats, “Like I told Scoble, we don’t compete with MSFT”. Battelle calls B.S. on him, “Come on!” Eric well-rehearsed, “this is about a more natural fit with the web”. Battelle: Why would you pay 3.1 billion for a company that MSFT thought was worth 2 billion max? Eric: Mumble, mumble. Battelle: are you going to kill the SEO business you bought with Doubleclick? Eric: dunno yet. Battelle: you distribute an app that deletes doublelick cookies. Battelle: Doubleclick has tons of information about my business. So does Google. You know my business better than I do; am I a sissy to be terrified? Eric: You could always go use one of the zero remaining companies we haven’t bought. Comfort, comfort, mumble, mumble. Battelle: ironic, MSFT and AT&T claiming antitrust against Google. Now Eric seems to be doing an about-face, saying that YouTube is going to be the one video platform to rule them all. Battelle: Are you going to compete with S3, EC2? Eric is getting annoying – says basically, “yes we will compete, but we’ll say that we aren’t and pretend it is something different”. Doing same thing with “net neutrality” (we don’t compete with telcos, honest). Eric: mobile is the new plastic. We will suck the profits from telcos investment in 3G, 4G. Eric: we’re good people, we helped Darfur. Battelle: You guys are so cool because you pay lip service to data portability Eric: Yes, we do pay lip service; thank God nobody ever examined that promise deeply (we would just say it’s too hard technologically)

Keynotes: Hitwise and Sifry

Participatory web, pretty stereotyped in terms of demographics segmenting. Men lek. Has some theories about what it takes to cross chasm. Predicts Yelp and Stumbleupon poised to blow up. www.ilovedata.com. www.hitwise.com.

Sifry: most people don’t realize they are reading blogs. Blog growth is slowing, but top link targets are twice as likely to be blogs now. NYT still dominates; for some reason Dave says NYT is not a blog. Mentions Doc Searls – high linkage, low traffic versus porn – high traffic, low linkage. Says Iran has a repressive regime? 88 of top 100 bloggers are new each year.

Session: Accessibility 2.0: Matt May

Really important session. Web 2.0 changed a lot of things, accessibility standards were developed in a different time.

Starts with (paraphrased) “Accessibility is right not privilege” (I forgot to bring my “When?” T-shirt). It is also the law (WCAG/508). If someone cannot browse, find, login, register, or transact – it’s not accessible. User testing is the only way to know for sure. No silver bullet.

Valid, semantic HTML is the best start. Next, use unobtrusive JavaScript (content, behavior, and style should be three totally separate pieces). Make sure your CMS and tools don’t get in the way.

1.3m legally blind, 421,000 deaf in both ears. More than 10x increase for people who are just impaired. Cognitive. Makes the point that this is simply what HCI is about anyway. Accessibility is related to SEO (”Google is a blind user with millions of friends”).

Evaluates YouTube and Twitter. YouTube is OK, Twitter is better – but CAPTCHA (visual only) is fatal flaw. Recommends www.webaim.org, www.webstandards.org, www.juicystudio.com.

3 Responses to “Web 2.0 Conf Day 2”

  1. Mark MacLeod Says:

    Maybe I just missed the link, but where are the webcasts of these sessions?

  2. allenjs Says:

    AFAIK, they aren’t webcasting them. All of our sessions from MIX07 will be on the web, though, within ~ 24 hours of the session (and keynotes streamed live).

  3. Mark MacLeod Says:

    It’s a Web 2.0 Expo, yet they don’t have webcasting? How ironic.

Leave a Reply

OPTION: 1