Posted on June 26, 2007 in Uncategorized by Rob Di MarcoNo Comments »

 

After reading Brad Feld’s Me.dium post, I took some time to play around with their Firefox plugin.  Short version, they track what page your browser is on and graphically display other people that are viewing that site or similar sites.  You can then chat (directly to friends or by “shoutout” to the public) with others that are looking at the same stuff.

The GUI is pretty slick.  Very responsive and fast with an interesting view on showing your network.  One of the designers talks about the thought that went into conveying the complex information.  Kudos to the developers and designers.

However, I just did not feel comfortable talking with other people or having other people see my name associated with what I was reading.  I felt it would be like if the librarian gave someone my reading list or if iTunes broadcast every song that I played (I hate to admit that I like the Dixie Chicks).

I use sites like del.icio.us to track my bookmarks and do not mind making things public, but it just felt a little too Big Brotherish for me.

Popularity: 6% [?]

Posted on June 26, 2007 in Uncategorized by Rob Di MarcoNo Comments »

CNN is planning on releasing a new and improved website.  Rather than just throwing a new site out there like USA Today attempted, they put a beta site out and really listened to the feedback from users before launching even asking if they are getting it and taking timeout to respond to the feedback.

Kudos to them for taking the time to listen to their users.

Popularity: 6% [?]

Posted on June 25, 2007 in Uncategorized by Rob Di MarcoNo Comments »

Saw this picture (ironically, off reddit.com) which illustrates the point of the previous entry.
Graphical vision of how Web 2.0 crowds work

Popularity: 6% [?]

Posted on June 21, 2007 in Software Development by Rob Di MarcoNo Comments »

The JRuby team just released the 1.0 version of the library.  Having played with Ruby before and Ruby On Rails before, this is an awesome step forward for using Ruby within the enterprise.  Now I can build Ruby On Rails applications that can easily interact with other JEE systems and services.  I can leverage some of rapid development properties of Ruby while leveraging the benefits of JEE for many server side processes.

As a contrast, at HMS, we have a legacy application that was written in Perl about five years ago that we would like to interact with a JEE application.  The application is the classic case of a prototype being built using a scripting language that made its way into production and never left.  Now, we are stuck maintaining it, but all of our other systems are in Java and we need to integrate this Perl application with them.

For performance reasons, protocols that involve XML over HTTP (e.g. XML-RPC or SOAP) are out of the question.  So our options are pretty limited.  We can use Inline::Java and connect over RMI (small community and some interesting JVM issues as we increase the number of Perl processes), we can embed CORBA into our Perl application (I love IDL), or we can write our own wire protocol (all of the pain of CORBA with none of the functionality).

Consider if we had written in JRuby.  Since it is running in a JVM, breaking out directly to Java is trivial.  So I can easily build out my prototype application and if it turns out that I need to extend the application lifetime, I can easily integrate it into my enterprise architecture.

Popularity: 6% [?]

Posted on June 21, 2007 in Software Development by Rob Di Marco2 Comments »

There has been much discussion around the “wisdom of crowds”, especially around sites like reddit and digg.  But there are a couple of flaws with this line of thinking:

  1. Half the people in the world are of below average intelligence.  When a crowd of idiots is making decisions, they are not always going to show great wisdom.  To extend this concept a little further, even if people in the crowd are not idiots, but are zealots, it can greatly influence the results that you see.  For instance, right now on the front page of reddit, there are three links to articles about RIAA injustices, five negative stories about George Bush, two stories about the alleged coverup of 9/11, and surprisingly, only one article on Ron Paul.  The reddit community, for whatever reason, tends to moderate up stories of these nature.  But what if a user DOESN’T CARE about these topics?  Sure they can hide that site from their results, but really, when I hide a link, I really want to do is muffle the community to be closer to MY views.  In other words, I wish reddit would not just show me what the most people had modded up (with minimal down mods); rather I wish that I could help define MY OWN crowd that would do the choosing for me.
  2. People tend to provide input only when they feel strongly about something.  Look at reviews on Amazon.com or ITunes.  Everything is either 4.5 stars or .5 stars.  There is no middle ground.  You cannot make rational decisions if the only voices being heard come from extreme positions (there probably is a follow on post about the presidential primaries for this point).
  3. You need a decent sized crowd for the crowd impact to really be felt.  Unless you have a very large set of preference data to pull from, or use very good methodologies to derive results (like how political polls work), finding the signal to noise ratio in the data is very, very low.

 

While it is interesting to look at where the masses are going, I think that we need to accept that accepting the wisdom of crowds over the wisdom of experts can only be applied in certain circumstances.  I do not see the crowd replacing the local food reporter, or for that matter, a good newspaper editor, anytime soon.

Popularity: 6% [?]

Next Page »