Carson Summit Part 1: Joshua Schachter - Delicious

This looks like a really old post (I mean, it's from 2006 for goodness sake). Stay for the retro vibes but be aware any information in this post is likely way out of date.

Things we've learned.

The first talk of the day was a great one and set the tone for the rest of the day. Joshua gave us 45 minutes chock-full of useful information on how to get the most out your servers (from using a proxy to throttle performance, to server caching); building an API (the main point being make it easy for people to get in and out of it by using a simple solution such as XML and not requiring an API key); and URL-rewriting.

URL beautification became a common theme throughout the day as several of the talks urged for simple URLs. Joshua's reasoning was that they are the main marketing push for your site as they are copied and pasted everywhere.

Other points

  • Don't add features which replicate other well-used applications - eg personal messaging which is limited to a site - why not just use email instead?
  • Don't give people the features they want - find the underlying problem they want to solve and give them what they need.
  • Top 10 lists just encourage people to spam your site - people start focussing on the list position.

Tagging was one of the most useful things Joshua discussed as it is something I'm looking at for a couple of projects. Interestingly was built out of a txt file based system Joshua had for bookmarking websites, which eventually distilled his descriptions down into one-word meaningful 'tags'. He urged people not to be tempted to auto-add tags for users as this removes the attention required to give meaningful tags - the transaction cost of having to add tags each time makes the tag system stronger (although don't overdo this cost). I'm glad he spoke against a forced vocab list for tags. His reasoning was that a tag will mean different things to different people and there is no way you can add a 'description' to each tag.

He said make sure you measure useage of features over a period of time - if they are still using the feature long after it launched then it is a success. When testing for usability, don't give people tasks or goals as it makes them act totally different to how they would normally - they start reading everything on the screen.

For marketing he said they didn't really do any - most of it was evangalising users and he points out that isn't a community site, the community exists outside on AIM and email and other communication methods. doesn't own the community (along with all the pitfalls of hosting such a community).