The Big Stonking Post continues for WDS07, with day two...

In this post:

Scott Berkun - The myths of innovation

We are very bad in our industry for forgetting history and repeating mistakes.

The moment of epiphany is a myth - Newton wasn't hit on the head with an apple, but we like to think it's true.

Innovation myths play on the belief that ideas are outside of us, that we might just get lucky to be visited by inspiration but it's not our fault if it doesn't happen. They discount the fact that really Newton spent his whole life asking questions.

We like to think ideas are discrete units, that you can pull out an idea when you need it. However research shows that it's more about process, having a mode of thinking and working that leads to getting results.

Many really successful people/groups don't talk about innovation, they just solve problems.

"The less you use the word 'innovation', the less like you are to actually do it."

When people want to pose as innovators, they tend to use the word a lot.

Challenge - stop using the word "innovation"!

Discussing the mythology of greek builders - we revere their engineering skill because so many things they built survive today; forgetting that all the crappy things fell down. The truth was that only the well-built, marble structures survived. The terrible wooden slums burned down or collapsed.

If we think we're so great at innovation, why is it still so hard to send a nice looking email? Why do we need reference guides to what works? Why is it so hard?

When we mythologise innovation, we forget all the irrelevant factors - politics, business, timing, luck, etc. They massively effect what happens but we have a rosy-eyed view that none of that mattered.

Modern exploration myth - Star Trek - doesn't show a real explorer's life of long periods of wandering around finding nothing. We get *blip* new planet *blip* new planet...

Challenge - spend a half hour looking up the origin of any piece of technology you use. You'll probably discover quite humble origins and many failures before the end success.

Our historical view of events and timelines lets us think that somehow there was a plan, but really the people who really created stuff didn't know where it would go. They didn't know what the life of their invention might be.

Technologists use the word "luddite" as a slur. But we would probably react the same way given the luddites' situation - their livelihoods taken away suddenly by a new tool. If we were replaced with a little black box and sacked, we'd probably want to smash that little black box.

People have an emotional response to change - this is a big challenge to overcome.

Consider warfare technology - Japanese culture felt that gunpowder warfare was dishonorable, so they stayed with sword warfare. This was a culturally-based decision - they chose honour over innovation. The technology alone wasn't enough to change they way they acted.

So, if you want to do something innovative you have to understand how it will play against cultural values. You have to consider the social aspects of what you're doing.

Innovation distilled:

  1. Delegate
  2. Take risks / make mistakes
  3. Reward initiative

Consider the story of 3M. Minnesota Mining and Manufacturing ("one of these things is not like the other"). Entrepreneurial venture, their first effort was a mistake; so that had to go back and take what they learned and try something new. They ultimately became a sandpaper manufacturer.

Then a car company had a painting problem; and one of their engineers (Richard Drew) kept working on a solution, despite the boss saying no "that's not what we do". He ultimately invented masking tape, which made tons of money off it.

So 3M had to work out how to support that sort of innovation.

(truly awesome three paragraph quote that is worth looking up the slides to get ;))

Managers need to empower employees to be innovative, they need to create an environment where people can try things, make mistakes and keep going.

The chairman of 3M bluntly states "Mistakes will be made", it's a truth of innovation. Yet nobody budgets for mistakes. Business likes to think they can avoid mistakes to the extent that they simply won't happen.

[Only about 1/3 of the audience had heard of Google's 20% rule - one day a week is your own to work on what you want.]

The 20% rule actually came from 3M. It was a way to delegate authority to people to try new ideas.


Recommendation - work out the non-technical threats to a project, then assign people to work on those threats. They're just as important as the technical issues/threats.

George Oates - Human Traffic

Chaos leads to cooperation - roads narrowed and signage removed led to people having to pay attention and negotiate passage through the shared space. People have since developed their own systems.

What is Flickr? "A great place to be a photo."

But before it was a photo site, it was a game! Online MMORPG, an offbeat and fun world. It was a deeply social space. People formed their own contact networks. This pattern of making friends transferred to Flickr.

There were "magnetic centre" people, who had large bubbles of people forming around them.

So, what next? Photo sharing in 2003? With the rise of digital cameras people wanted to host lots of photos. People wanted to get them off their hard drives and share them with people. But the available services weren't community based. There was an opportunity to mix it up.

The first generation was the same interface as the original game - photos plugged in.

Working on the site, they "didn't sweat the technique". They made mistakes and kept going. [Imagine that, there are strong correlations to Scott Berkun's talk!]

So the early userbase of Flickr had a lot of nerds, bloggers and nerdy bloggers. Also snarfed in the userbase of the original game. With that userbase they had a shared understanding of netiquette. They were understanding and even amused by "the show" as things were changing, breaking, evolving...

There was an appreciation for experimentation.

They moved away from their original interface since people didn't really get it, and it was synchronous - two users had to both be online at the same time to interact.

They had an open "ideas forum", open to everyone but ultimately with about 50 really vocal users. Although a minority they were important to the process.

The "gamma" release was the first instance of shock in the user community. They were taken aback by the changes.

Removing rules from a space can make it more creative and collaborative. People will find their comfort zone within that space.

Initial concepts

Initial concepts for flickr:

  1. Mirror the offline experience of looking at photos
  2. The "photostream"
  3. Individual as publisher
  4. Layers (onion concept):
    People you know
    Things you like
    The Universe

Sudden switch... instead of the mundane daily life stuff, with the Jakarta embassy bombing people posted photos within minutes. So it was an unintended news source. But at the same time, there were lots of photos of lovely daily life as well as the

Design strategies

  • User personas with care. You don't really know everything about everyone.
  • Design for zero data. When people sign up they don't have anything in it yet. Assume naivety (not as a pejorative). Offer some guides, hints, show some options.
  • Sensible defaults. Too many options does not mean more choice. "Simple simple simple". Lots of people don't change any settings.
  • Experience "gates" (eg. first time you upload a photo, post to a group, tag a photo). Choose your own adventure.
  • Show activity. We are social creatures.
  • Go UTF-8!

Support structures

Flickr today? It's enormous! 1.5 billion photos, 1 million per day. 10 million members "that's not a community, it's a freakin' society!"

Organic growth. You can't say "let's add some community". Be hands-on at the start, be involved, be human.

Community guidelines. "Don't be creepy." Guidelines set the tone.

Provide excellent customer care. "Thanks! That worked!" Have humans answering emails and participating in forums. Inject some personality into support documentation! Let your brand show (eg. playfulness).

Have neutral point of view. With great power comes great responsibility.

Observing community


  1. anonymity
  2. recognition of you
  3. community
  4. leadership
  5. celebrity (fame)

rinse. repeat.


Your newest member is a delicate little seedling. Nurture them.


show yourself - be visible

Beware competition - we're all winners.

Don't become obsessed with interestingness! It's run by the magic donkey :P

Be open - nobody likes surprises. Bad ones, anyway.

Aaron Gustafson - learning to love forms

Aaron's preference for wrapping items is to use a list, ordered or unordered depending on situation. Semantically correct AND useful.

Naming structure for form elements based on form ID - keeps things specific.

Button element - has problems with IE submitting all buttons; but they have a big advantage in terms of how they're rendered. You have more opportunity for control with the button element.

Aaron adds extras like colons to labels using content: in the CSS, so it can be changed easily. About the only browser which doesn't support it is IE, but if the colon doesn't show the form is still fine.

Similarly, uses :focus to highlight current input, even though IE6 won't get it, it's a progressive enhancement.

Width of a select element is usually input width + input padding + 4px

Aaron has created a dropdown alternative which is styleable and accessible, however it is hard and requires javascript.

You can use a hidden input to set a default value for a checkbox, that can be passed to the backend instead of null. The same name is submitted twice, but the second one is picked up by the back end. Aaron has never had to actually write code to ignore the first value; the system just automatically picks up the second/replacement value.

Wrapping an extra <label> around related values - not strictly allowed, but Aaron's testing shows it works ok in screen readers (according to Fangs).

[more stuff about messaging, didn't have time, will post to slideshare]

Lisa Herrod - Usability: more than just skin deep

Not only do "disabled" people benefit, there are groups of people who consider themselves "normal" but benefit from accessibility.

Personas tend to be shallow. They may cover one type of user, but don't cover the broad range of people who end up using the sites.

True user experience design: user profiles are inclusive, not exclusive; accessible design from the outset; ground-up approach.

Example: The Australian's search results page, which is accessible and valid but not very usable for screen reader users. Of course it's outside scope of the talk to mention, but some of the issues have a contractual basis ;)

But the real point: valid, compliant code doesn't make a good user experience.

Looking at the accessibility guidelines, there are check points for roles other than frontend. IA and design need to be involved.

Expertise is essential. You need people who actually understand it to work on it.

User testing usually misses people who actually use assistive technology.

It's not all about blindness! Accessibility is a much broader issue. People still focus on blindness.

User Focussed Development == User Centred Design. UFD is the technical equivalent of UCD.

It's not just about the code (it's about the peeps)

Role Specific Checklists - we need accessibility checklists for specific roles in development. Not all the checklists relate to frontend!

Six steps to recovery:

  1. Define primary user group
  2. Site build
  3. Code review
  4. User testing
  5. Rework code
  6. Final compliance review

User profiles can relatively easily incorporate disabilities. Social profile + disability = more complete than just the usual social profile.

The meat of the talk... the checklist! WCAG guidelines, but with items assigned to frontend, scripter, content producer, IA, designer, etc.

Frontend still ends up with the majority of check points, but not all.

Plus it's useful to look at who needs to work with each other to achieve compliance.


(spent too much time talking to write)

Slides at

Adrian Holovaty - Being smart about your data

Adrian as a web developer is interested in making a site which is a joy to use and explore, full of serendipitous moments of discovery.

This is becoming a must-have; people are starting to expect it. They're used to sites like IMDb which let you browse deeper or laterally from your starting point.

You could ask yourself: how useful is this site to researchers? They are going to be hard-core users of your data.

Serendipity is good because it increases your site's stickiness (keeps people clicking around) and it increases usefulness.

To make this happen, you have to be smart about your data.

It all starts with the structure.

For large scale sites, you will need really good structure - you're not going to be doing it all manually.

Getting structure is half the battle. It's human stuff, and harnessing that is hard.

Adrian as a background in journalism...

Journos are fundamental data people. They collect, analyse and publish data. But it tends to be fairly flat, not linked enough.

The tragedy is that journos collect all the data, but they don't allow things like rich linking because they don't put any structure into their data. They go to a source, get a bunch of information; they write structured notes; but then they publish plain text.

Lesson one: structure your data.

Lesson two: all data has structure.

Lesson three: give your data "the treatment"

Once you have some structured data, how do you get to a well-done work of hypertext.

Hence, tips for efficient hypertext.

Example: crime data. The original data for Chicago was being published in a boring table, no links.

Step one: list fields. eg. date, time, crime type, address, location type, arrests made?, case number, etc.

Step two: identify which ones are interesting enough that someone would want to browse by them. eg. yes I would browse by the day it happened, yes I want to look up crimes in my suburb. Arrests made? Well perhaps not for browsing - the information is good but you might not browse by it. Similarly probably won't browse a list of case numbers.

Step three: make breakdowns. Crime by date, by time, by type, by location.

Step four: make list pages. All the crimes on a particular date, listed on the page.

Step five: make detail pages for concepts. In this case, a page for the crime listing.


Things to note

  • You don't know why a user is looking at a particular bit of information
  • Permalinks for concepts are important. eg. You should be able to link to a particular crime.
  • Also, linkability/bookmarkability
  • SEO - advantages to having the most granular page about a given subject.
  • Serendipity


"By the way, Chicago's nice, you should visit..."


Another example is "Faces of the Fallen", showing US marine casualties in Iraq. Done for the Washington Post. It's compelling.

However the concept applies to any information. eg. video game reviews, political advertising


So the next logical step, since the output is formulaic... build a tool: Django Databrowse.

Not just for externally available systems, it's an excellent tool for evaluating your own data.

The tool is sometimes useful because it is "dumb" and will show you ways of looking at your data that you yourself might not have tried/thought to try.


  • Find the structure in your data
  • Design for serendipity

Q: How did he encourage journos to enter the extra data?

A: One way is to choose where you want to do it. The more niche/granular you got, the more passionate people are about the data. eg. he did a database for a small paper's movie reviewer, the reviewer was blown away. So if you can find champions it becomes an easy sell.

The Wash Post had a culture of competition which meant people wouldn't share their notes, they were pitted against each other. In those cases there's no way the reporters are going to share it with a web geek.

"Culture stuff... that's for smart people, not geeks..."

Mark Pesce - Mob rules

[Listen to the podcast. The intro was too good to retype. You have to hear it.]


We don't especially need mobile phones. They're bling. The way we usually use them is relatively trivial.

But to the third world, they're essential. The mobile is far more necessary for their lives, since it determines whether they find a good market for their produce (or whatever it is they do to survive).

Grameen Bank founded Grameenphone, because so many poor people were taking out loans to buy mobiles.

None of this was predicted.

The first world, the rich, don't know about this. But the poor do. They have known for a while and they've been buying phones.

Sometime in 2008, half of humanity will have a mobile phone.

It's not slowing down. In fact it's getting faster.

Prediction - campaigns by charities, to buy a mobile for the poor.

Nokia created a phone for this market and it is the most successful device in history, far more successful than the ipod.

Pervasive wireless communication is of far more value to the poor than the rich.

They don't need ebay. They can work it out for themselves. They just need connectivity.

People are the network.

The network in every form is anathema to hierarchy.

The net interprets hierarchy. as damage and routes around it.

The network is quietly kicking the legs out from under hierarchy. It's not an epic pitched battle.

The same mob rules apply everywhere, to everyone. It applies just as well to a kenyan farmer with a mobile as it applies to a movie producer in queensland.

"Enough of philosophy, let's play."

One more thing... there's still one hierarchy. which is stubborn and resistant to change. It's Telcos. The firms which created the network are somehow immune to the effects of the network.

BUT this could change.

Meraki mini router... creates a mesh network. They cost <$50.

This means there's the ability to have a decentralised network.

Meraki minis have been given away in San Francisco and is doing better than any official attempt to provide free wifi.

Internet access in Australia has always been about bending over and taking it........

But no more. Mesh networks could change this.

It's all software. Any machine can join the network.

Terranet AB - swedish company turning phones into wifi mesh networks. Telcos didn't fear VOIP on mobiles, they feared mesh networks.

  1. The mob is everywhere.
  2. The mob is faster, smarter and stronger than you are. "The street finds its own use for things, uses the manufacturers never intended." - William Gibson
  3. Mob rule is not about sites, it's about services.
  4. The mob does not need a business model.
  5. Make networks happen.

Advertising is a form of censorship.

Everything that's online is built on sand. Advertising is a demand that you pay attention. And it's a demand that can no longer be enforced.

If the mob doesn't need a business model, how are you going to make money? The mob doesn't care.

Once networks are created, they can not be destroyed.

Embrace your obsessions. You will be rewarded.

The mob gets whatever it wants. Fortunately they generally want the same stuff as we do - better lives for themselves and their families.

"Wisdom consists of the anticipation of consequences." - Norman Cousins

Have courage and keep moving. Standing still is not an option. / Mob Rules (The Law of Fives)