Moving from it to them
I just had an interesting conversation with my 22 year-old daughter, who is having problems buying some tickets for Jerry Seinfeld online. At first the entire theater's website was down, now it is working, but the link for buying tickets is not responding. She doesn't understand why it isn't working when the page loads fine. I tried to explain that you can't think about "it", but rather "one of them" being broken. The theater is probably on a different server from the ticketing system, so one can be working while the other is down. She has been using the Web for 10 years, so she certainly understands that sites can go down, but when distributed portions of a site fail, this is often hidden by the page design. I know this is going to be a major confusion for users when sites become dependent on external web services. The dirty little secret of Web 2.0 is that a loosely joined architecture is also more error prone.
I remember that early users of the Web in about 1996-7 were often confused about which site they had actually purchased things from. I ran a software downloading site called DaveCentral, and people where constantly emailing us thinking that they had bought software from our site, when they had actually followed a link to the product's publisher. They also were confused about whether they were still on Yahoo, if they followed a link from there. This may sound silly to today's experienced Web users, but I'm sure that the type of aggregated application that is the basis of much of Web 2.0 will be the cause of lots of user confusion. This isn't a fatal flaw, but it will be an obstacle to overcome. I can imagine the first time a major protion of Google glitches, like the maps, and applications all over the world crash.
When I added the code to this blog to notify blog pingservers with new posts, I found that about 40% of the servers failed for one reason or another. I didn't bother debugging each one, since I had a lot of pings that got through, so I was assured of good search engine coverage anyway. This type of "percent yield" on RSS feeds and APIs should be part of the design, user interface, and user expectations of Web 2.0.
