| |
Live Blogging The Future of Web Apps
The place is packed. 400 people were expected. 800+ are here. Insane. Free WiFi and it looks like 1/2 the room is blogging the event.
Joshua Schachter
del.icio.us
Joshua is giving a relatively disorganized stream of consciousness on his trials and tribulations of a bunch of things that happened while working on del.icio.us.
- Getting it right the first time: Wait for some stuff to break instead of trying to fix everything first.
- Data: Splitting content into areas. Text. Images & other. Use different servers for different stuff.
- API: The easier you make your API to get into and out of, the more people will use it.
- IDs: Do not expose unique IDs to the public. Some bonehead will iterate over all of them and crash your system. They see tens of thousands of hits from people trying to scrape the entire site.
- Features I: What features to add? What you put in is as important as what you leave out. I don't add features that exist elsewhere. Like email/communication devices. Let people what
- Features II: Build what people will use rather than what they ask for. Get to the bottom of why people are asking for something. Solve the problem.
- RSS: The native way to represent a list of links. Another set of tools. Put them everywhere you can. The doorway in and out. Understand the headers/ if.not modify. Stash the time stamp, and then you don't have to recalculate the entire feed. 60% of the traffic is due to repeat calls to the database.
- URLs: Build a system so that the URLs are understood. No session keys. It is ugly and should be avoided.
- Surprises are a good thing. Watch for behaviour in the system. Need to decide: Amplify. Ignore. Kill.
- Passion: Solve problems that you see now. Don't go look for others. Passionate people who see them will point them out and help solve them.
- Release: Closed betas/ Limited releases are a horrible Every day that you are not live is another day you are not live and doing something. People cannot be passionate about your service. Domn't wait. Get it done. Get it out there.
- Attention: Any sort of aggregation of attention is a good thing. As the group gets bigger, the bias drifts. You need to microchunk things after a while. Allow the system to gracefully fragment the conversations.
- Spam: People will try to get into any aggregation of attentioin. Top 10 lists create a nuissance. A distraction. Don't provide feedback to a spammer. Let them think that everything is ok.
- Tags I: Not about classification. About user interface. Storing a state. A working context. Not all metadata is tags. The value is in the attention. Not just that you saw this, but that you saw this and thought it was worth taking the time to do it. Less value to automated tagging.
- Tags II: If you make it too easy (low transaction cost) it hurts the value. Make people do a minimum amount of work.
- Tags III: Beware of librarians.
- Motivation: Understand why people are there. Most of the time, it is of selfish interest. If there is no value, then why would they be using it.
- Effort: Spen time building things that people will use. Don't do it just because you think it is cool.
- Intuition: Guesswork backed by numbers. Monitor things. Survivorship analysis. Look at usage over time by specific users.
- Measurement: Measure behaviour, not claims. They measure what people do (bookmarking) not what they claim (quality ranking).
- Testing: 3 days of user testing. Worked with Creative Good (NYC) and liked it.
- Language: Speak the user's language. Uses "bookmarks"
- Registration: Don't make people register before they use your site. Show them value first. People want to see what they will get before they register. Email addresses are valuable. Fears of spyware are real. Let them wonder around first. Present gestures/verbs first, and then make it short, sweet and then get them back to where you were.
- Design Grammar: When you break from normal behaviour, stick to as much as you can that is well-known and understood so as not to confuse people too much. Innovation is ok, but confusion is probably not worth it.
- Morals: It is their data. You just get to use it. Have some respect.
- Infection: Convey every kind of communication vehicle possible to get the word out. Turn RSS feeds into apps. Look for viral vectors.
- Communities: It is not a simgle community (unless you) It is a tool, but I do not want to own the conversation. Problems like flame wars and stuff. Enable communities to use your system, but don't necessarily own them.
Wednesday, February 08, 2006
 Comments:
|
|
|
 |
Subscribe:
Postcards by email |
|
|
|
 |
 |
Rewind: In case
you missed it |
|
|
|
|
| Parlez-vous Deutsch? |
|
|
|
| In Heavy Rotation |
|
|
|
|
Previously
Archive
Author.ity
|
|
|
