Under the right conditions, large groups can be very intelligent, often more so than the smartest expert in the room.

What do intelligent crowds look like? Look at the racetrack. A crowd at a racetrack can give a near perfect forecast of the future. Crowd is a mix of experts, novices and cranks. Similar is the stock market. Over time it is very difficult for even the best money managers to outperform the stock market. (Or look at newsfutures.com.)

(Lots of stuff here that it would be good to see graphs and examples.)

Aggregation can happen in many different ways.

Diverse crowds are far more likely to be intelligent than non-diverse crowds.

I know most of this already. Time to skip to another panel…

Fried: Obscurity is a good thing. If you can fail in obscurity, you’ll learn a lot more that way without being berrated by the public. Removes the fear of failure.

Less comes for free with a side business — people should be embracing. Conventional wisdom is you need to outdo competitors. “They have five features, so we need six.” Instead deliver products that are small, clear and simple.

Functional specs are political documents about covering your ass, not about creating a good product. Illusions of agreement — everyone interprets it differently.

Three types of less:

Less time is good. The more time you have, the more time you waste on things like overplanning and overthinking. Abstract work isn’t real work (functional specs), etc. More time encourages procrastination. Don’t worry so much about planning, stay small and get at work.

Less money is good. The only thing money really buys these days are salaries. Spending your own money makes you use it more wisely.

Less software. Build less software, build less well. There’s endless time to add later, but you can’t remove after you add. Less software is more manageable. Less things you have, the less that go wrong. Too many people focus on solving complex problems when the simple ones haven’t been solved: (Don’t build software that’s clever and tries to capitalize sentences, etc.)

Embrace constraints: “We need more people to finish the product.” Maybe the product has too many features. Easier to improve products when there’s less to change.

Use products or software as you build them. You’ll learn that you have things you don’t need, and need things you don’t have.

The companies that take VC money always seem to need more of it, but those who don’t take it always seem to have enough.

Don’t plan too far in advance, as you don’t know what’s going to happen. If you have to grow, you don’t have to get big.

Curiosity is much more important than bullet lists of skills on job descriptions. What people did in the past doesn’t matter.

Coudal: It’s easy to get lost in thinking what you’re doing is the only thing you can do.

Fried: When you make a change to a product, you’ll get a lot of noise from people. Do not react to squaking to change something back. It may go away as users get used to changes. Explain convictions, and don’t freak out when people get upset. Build the interface first and then start to use it as you build. You can look and agree on real stuff, but you can’t agree on abstract (functional) ideas.

Adding features to functional specs is easy, but not easy to implement in real life.

Coudal: If we get an RFP that’s more than two pages, we won’t do it. If they spent that much time on it, imagine how big of a pain they’re going to be when we get them as a customer.

Making it up as you go along is a good thing, and the whole deal for Fried’s and Coudal’s companies.

Fried: Write stories instead of functional specs. Write a story about how something should work, and then have someone design it. Usually works better than spec.

Fix budget and timeframe, and that’s version #1 of the product. Prevents scope creep. makes you focus on the most important things.

Beta tags are kind of ridiculous.

We make mistakes when we compare the past with the possible. (Bush Sr. looked bad against Clinton, but Bush Sr. looks good compared wth Bush Jr. Or a Big Mac may not be appetizing, but it is when you’re on a plane with no other food.)

Good presentation, but not much to take away from with notes. Or to practically apply to day-to-day life, for that matter.

I’m sure I’ll be more concious of bad decisions all the time now, though. After I make them.

Panel: J Wynia (wynia.org) / David Swedlow (metastorming.com) / Mary Hodder (napsterization.org) / Liz Lawley (mamamusings.net)

Blog: http://www.beyondfolksonomies.com/

Wynia: What is a folksonomy?

Swedlow: It’s a bottom-up way of organizing information. More informal than taxonomy — “dog” instead of some long latin name.

Lawley: They are a way that people classify things — language and classification are not the same things. In folksonomies, people get to define their own organization, not have it defined upon them by some kind of authority.

Hodder: Usability of current folksonomy interfaces are pretty bad. Flirckr and delicious can have useful elements, but they still have a way to go. Users can be overwhelmed in both the collection and entry phases. Fear of more slicing and dicing as it won’t actually get us anywhere.

Wynia: Recently saw article on tagging problems with delicious: Five months you may have tagged something web design, but now tag it web design. Article suggested implementing rigid taxonomies, which kind of misses the point.

Wynia: Traditional way of handling problems with tagging is to go back and adjust them, but most in the population will be lazy about it and just leave outdated tags. But the fact that these things are already related is already evident in delicious. Maybe build tools that implicity tie specific tags together — web design and webdev, for example.

Swedlow: It’s not folksonomies vs. taxonomies — you really need both. Folksonomies get you to the general area, taxonomies get you to the specifics.

Swedlow: Maybe use implicit tagging.

Check attentiontrust.org. (Hodder is on the board.)
Hodder: Not a big fan of automated tagging. Systems just need to be in place to allow tagging — video site she working on right now pulls from both tagged and untagged sites — sites that allow tagging have almost 100% use. The fact that it’s a little bit hard and that someone had to think what to put forces better information — someone has to care to do it. You’d lose that context if automated tagging was implemented. Quality of information wouldn’t be as good.

Swedlow: Distinction between implicit tagging and automated tagging. Implicit makes use of work already being done. (What? Offer suggestions instead of automatically placing information.

Lawley: Asks for show of hands for who here uses tagging. Most do. “You are not most people.” Most people, even in IT, don’t want to use these sites and features. But it’s giong to start to happen in different contexts — Vista will have some tagging componenets in it. We don’t spend time thinking thourhg use scenarios. How are we going to make users’ lives better? The wisdome of crowds can be overstated (If you look at what democracy has given us it doesn’t always give us the best options).

Audience: Users already use folksonomies with folder structures on their computers — just to need to make that kind of work easier elsewhere.

Audience: Opositie positions or opinions can be just as useful. Ex: Movie reviewer who always endorses movies you hate, but hates movies you’ll love. Also, folksonomies need to be effortless.

Wynia created a tool to search the content of his own bookmarks. Lots of held breath during his description.

Clustering technology in Flickr resolves some of the issues of confusing tags — sxsw, sxswi, sxsw2006, sxswi2006, etc.

Lebowsky / Peter Rojas (endgadget) / Marc Canter (broadbandmechanics)

Canter: Beware the person who says all the solutions will be on one device. Consumer Electonrics, mobile, video games, etc, are all separate and have the potential to continue to do so.

Need open APIs so users can move information between platforms and systems. Can create an archipeligo of functionality, which can then compete with Microsoft, Yahoo, etc.

Rojas: Islands are developing on the hardware side. All hardware is trying to lock in users, not let them use other services and companies. For end users, makes trying to move towards a convergent lifestyle less attractive, as they have to worry about what’s compatible. Average users will have trouble understanding how all of this works together. “Why can’t I use this iRiver player with iTunes?”

Companies are so focused on controlling ecosystem of convergence that they aren’t really interested.

Canter: First companies in usually try to prevent others from getting involved.

Rojas: One of the biggest reasons mobile devices are held back as platforms is because there isn’t a single or leading platform to develop for. Consumers don’t want to worry about interoperability, but companies would have to give up some control to do so. One of the great things Microsoft did was give, mostly, a common development platform for PC development.

Developing for apple is a lose-lose scenario. If you’re successful, they’ll rip you off. If you’re not, you’re not. But watch them for R&D.

(Look up Oragami.)

Q: What’s the financial incentive for a company to open their standards?

Canter: Apple will be the last one, followed by Verizon. But everyone else will be open, forced by customers. As Yahoo asks, how do you get the next 30 billion clicks?

Rojas: Companies have to ask, do we want to control a small closed market, or have a slice with a huge open one?

Canter: Three percent of fiber is turned on. Other companies that aren’t in it right now will eventually run around closed big guys in the company.

Statement from Yahoo guy: Open APIs is their big area right now, as they see much of what they do as commodities.

Q: What about customers scared of the bill from their convergent, wireless functionality.

Canter: Keep supporting open in way shape or form. One of the big boys will eventually figure out it’s to their competitive advantage to work with the open people, and then they’ll jump on it.

Rojas: I think consumers are much more intelligent than we give them credit for. They refuse to spend their money until they’re offered a solution that works. Slow uptake of 3G. It’s too expensive and complicated to use.

Canter: The battle for convergence is for the wire into the home. Is the telephone wire, or the cable wire? There is no endgame — it’s what’s next.

Q: What about PVRs becoming obsolete because of broadcast flags or other DRM on broadcasts.

Rojas: It’s a broadcaster’s super fantasy to control what people watch, and lots of efforts to do so. Again, though, there’s a competitive advantage to a company that goes a different way.

Canter: Best features/Brand is what gets customers, not closed devices.

(Look up songbird. Open source that wants to support Microsoft DRM.)

Well, we’re in Austin. It would be pretty generous to call the Park Plaza hotel three stars, but for $40 a night, I’m not complaining. Well, almost: I can’t get online tonight–the network connection on the desk doesn’t actually plug into the wall–so this will have to be posted tomorrow. (Other hotels call maintenance for network problems. The Park Plaza calls security.)

The trip down went well. It sucked that we had to stop for a layover, but American ran on time and I guess it wasn’t that long. It’s in the 80s down here, and while the weather hasn’t bothered me too much yet, I’m glad I brought sandals. Ye dogs would be stinky without them.

Dinner came at Stubb’s near downtown Austin. Great briscuit, and mashed yams that might as well have been dinner. Beyond that, there was the requisite walk down Sixth, and a quick mental remapping of the city.

Just over 10 hours until the first panels begin. It feels good to be back in Austin, and at SXSW.

More later.