Notes: Designing the Next Generation of Web Apps

The following are my panel notes from SXSW. As I am not the fastest typer I have paraphrased what was said. Should you notice any mistakes please do point them out in the comments for corrections.

With hype around "Web 2.0" reaching an almost unbearable level, it can be tempting to dismiss it all. That would be a shame. For all the breathless marketing and ill-funded startups, there's some absolutely amazing design being done out there. We'll talk to those behind some of the most successful applications on the web and find out what the process was behind them, how they bring users into their designs, and what it's like to stay on the cutting edge of web tech without succumbing to it.

Moderator: Jeffrey Veen, Measure Map/Google

ER: I want talk about some of the map based work I've been doing, Moveon.org. It was a way of interacting with a panel this is a screen that tells you the results of an online poll; this is the result for California. After the move on project we created an app called Mappr to take Flickr photos and put them on the map. Mappr doesn't know about route 665 but when you put it together with tags you get more useful info.

Cabspotting records the paths of cabs in San Francisco. Animation of cab tracking data showing bright colours for cabs going faster. This project shows the ebbs and flows of data.

MT: One of the things discussed for this panel is iteration. At six apart we are working several iterations of our products. Comment is similar in terms of feel to livejournal. We want to get the next generation of people blogging. What does it take to get to that APIs aren't just external? We have APIs that are internal as well; for example systems like accounting systems you don�t want to rewrite for every project.

JV: how does the acquisition of Splashblog fit into the next generation of blogging?
MT: They have clients for mobile devices. We need to have something for blogging through mobile devices. You can post and you can receive comments.

GO: We've been working on some iterative designs. Flickr is a messy place but it's good that we can bring out things to talk to them about. When we release a new feature people give us their feedback. Flickr makes up the rules of what it is itself. We don't tell people what to do in Flickr. We demonstrate what you can do and then make it prettier through iteration. Slideshows are a good way to see beautiful photos. People suggested a black background so that's what we are going to do. Organzr is pretty heavy going; in the Current ui the workspace is too small. We changed it around. Someone quoted "Flickr is a giant chaos".

EW: We work at Odeo on audio stuff. For that last year or so we've been trying to bring audio to the web. We use flash only where we have to. It's been an interesting challenge we have this audio recorder where you can start recording things (flash). We show visual stuff for feedback. It's also possible to leave a voice message by clicking a button on someone�s website.

JV: People aren't going to be doing audio heavy lifting there are they?

EW: there's limitation with what you can do with audio recording in a web browser. People see us a podcasting company but we see our selves as an audio company.
JV: Is this the gateway drug??
EW: We�ll see!

JV: Measure Map was acquired by Google. At adaptive path our design practices consulting was only one of the things that we wanted to do we also wanted to publish what we did. We wanted to take enterprise level software and boil it down into something simpler. Where are the areas of incentives? There's a lot of power around parsing server logs, how are people participating in my blogging. So we're not focusing on page views we're interested in links in and links out. We used a tech in an agnostic way. Measuremap made their own maps using flash instead of using another product.

I developer the Measuremap application as a designer not as a development focused application. I wasn�t constrained by the technical feasibility of what we were doing. At writely they are Taking user feedback and feeding that back in.

MT: It's different from when we started. I started with my husband Ben and we could work in out own way. The problems came about when you have a team. "It doesn't have get much smaller than a husband and wife in a bedroom" (laughter)

60% of our team are engineers. How do you go from 2 people to a large team? Then iteration comes from learning what you are doing on a product basis and sharing within the team. You have to get the product out. Until we got comment out we couldn't wee what we needed to do. You�ve Moveable type which supports a large amount of clients then you have live journal which is in the tradition of getting things out. That's one of the things that has been good for live journal.

JV: We had the same pattern we had to have discipline to release Monday and Thursday nights instead of all of the time. We could build up features and then do a release and that worked better for us.

GO: the difference between our system was that we could send to CVS and then press a deploy button so any developer could make a small change and then deploy it.

JV: I consulting I found it frustrating that I could make those kind s of fast changes, but in the start-up helps that you can move faster.

ER: It's not crazy all over the place development. Your thinking about the problem and try something sot you can test solutions quicker and the solution you try may change the problem.

GO: We also had a live user base who were happy to feedback about problems

EW: A couple of months ago we adopted a scrum methodology which says you have a time period to denote your iterations like 30days. We are down to 2 weeks. When we first adopted it, it was like this is an eternity. What we did is keep the release cycle and decide what we are going to iterate on.

JV: I'm impressed by this tight integration with feedback building out apps with the assumption that users can create their own experience. This is change in the way that we create our apps.

How many talked to a human travel agent to create a booking to come here?
Audience: 3 hands shown

JV: Travel agents are gone now. This has also happened in publishing. In the same way the app that you are producing can be considered as publishing.
Consider how many web services these days that you pay for. It�s very different to 3 years ago. How's that affecting the decision making going forward?

MT: People using products are varied people; they don't wakeup saying I want a blog. People interact with blogs through reading them very few bloggers start a blog without participating in a community first. We've been lucky an unlucky people are blogging about blogging about blogging tools. It gets a bit much so how do you find a balance. The whole group is smarter than that individual. I don't want to be misquoted: "Mena Trott's bloggers are stupid"

Blogging to public is iffy. Seeing what people are doing and figuring it out.

JV: You do that with audio, trying to put audio into a page.
EW: I don't think we've been able to help people to work that out. With audio it's more intimidating but they love to receive the audio messages. If you record something by default it's private

JV: How did that change behavior?
EW: It helped to encourage people to participate. It's private and it has an unguessable URL.

JV: Designing in our applications. The editorial control in applications is coming from algorithms now. If you think of the front page of digg as your news page that's a set of algorithms that dictate what you are doing.

GO: We saw how people were adding photos and we had a Russian genius who saw patterns in the data. Interestingness is a very tricky and highly secret algorithm that works out activity around a signal photo.

GO: the explore page is an interesting photo from the last 7 days. Not editorial. Mixed with other editorial content
JV: I love that

MT: One of our decisions was we didn't want emoticons in our products.

AR: We just started working with digg to create some visualizations of their data. At the moment we are looking at the data to see what's interesting. There's an idea that you don't plan and you design for flow rather than for static experience.

JV: the hype machine aggregates audio blogs to give custom radio streams of the blogs based on the kind of music. This is almost what we are talking about with the semantic web. User generated content means that design just becomes a container. Chicago crime.org: users are generating content and turning that back out. This is an example of mashing up googlemaps and crime reports to show where robberies have taken place.

MT: The only thing I would add is that in terms of writing a post I will put stuff that I have written. What we do is all containers for other people's work. It's a design decision where templates have been a big part of what we do but now it doesn't seem that important for us. It doesn�t need to be so unique.

GO: Listening to Mena consider someone's experience of coming into this big messy place, it's a case of showing the options without necessarily describing them.

AR: Design is a container not "just" a container. We get to build these containers that are something that flow.
MT: What would a waterfall be without a container?

JV: The level of detail that went into wired but it had no influence on the content.
EW: Where you are designing containers for other people�s containers. On page in MT won't necessarily look like the next.

Q: Rapid iterations are wonderful I find myself more redesigning applications.
MV: I don't think it is something for fortune 500 companies. I don't like a project where we say we're going to redesign from scratch.
GO: If you iterate too quickly it's possible for the dev to get out of control. You should have to play a game to change your email address.

MT: It has to do with expectation as well. Typepad people pay a fee every month and if some one sees a bug they will call the office. If you have a small company then they want something to always work.

JV: Amazon does serious A/B testing to see which thing works but I think it's a mess you need some innovation.

Q: How do you guys build ethical standards into your projects?

MT: Export has always been big.

JV: Feedburner can take your feed. And that's a big commitment. When you go to delete a feed they help you to redirect feeds.

Q: You design based on commented feedback, what do you do about the minority of crazies?
GO: It depends on the crazy. The difficult voices are the loudest. If it's useful then great if it's ranty then say calm down.

JV: People wrote into Lycos to say they didn�t like the colour blue. Some feedback is astute but some is bouncing off the walls.
MT: Our support team are in the trenches to get it to product teams. If everyone has the same idea for a feature then maybe it should be implemented.

Q: In an iterative model of 9months it would be good to have 2 week cycles. If you do a release how do you unwind it you have made a mistake? Also people don't know what they want.

EW: We only roll back when it's broken. It doesn't come up that much.
GO: you want feel that you were going to put out is good.

MT: If we had only put things in that people requested then we wouldn't get so far.

JV: User research is a foundation from which you innovate and write good content.

Q: where are you getting the inspiration for the philosophies if not from trad design practices?
MT: I use books and magazines. Working out from print how things work. Everything around you is important.

AR...

JV: Flickr started out as a game where you could share photos as part of the game.
GO: I look at the way people interact using body language, first impressions.

Q: There's been more people talking about making money out of content. How can content creators monetise their content.
GO: Maybe there will be some digital market place.
MT: moveable type has ways for ads etc.
EW: We�ve talked about it but we're not sure what we'll do yet.

Show Comments