One of the joys for much of my early life was walking through a library, looking for a book to curl up in the corner with and read. As a young boy, I would walk to a local library and read at the wooden tables before I had my own library card to check out books. Later, I moved and had a newer library with large, comfortable chairs in which to sit and read a few pages. I've enjoyed the same thing as an adult in various bookstores.

I didn't often have a specific book I wanted to read, so I'd randomly walk around, looking at spines, covers, and choosing a book in a somewhat random fashion. My fellow founder at SQLServerCentral, Andy Warren, also appreciated the randomness of browsing in a library for bookstore, discovering some new author or story to enjoy. Across the years, we've discussed and debated whether or not there was a way to duplicate this experience with technology.

These days I tend to buy or borrow all my books electronically. The convenience and unparalleled and time is valuable, and I certainly don't miss the days of packing 4-5 large books for a week long conference trip. However, Amazon and my local library tend to use recommendation algorithms, or popular titles as the presentation method for their sites. I have lost quite a bit of the ability to enjoy the randomness that comes from wandering and happening upon new titles. Andy feels the same way, though none of our brainstorming has produced a way to duplicate the feeling of wandering through bookshelves in an electronic fashion.

I'm not sure if there is or isn't a way to deuplicate this electroncially, but certainly the feel isn't the same on a screen. All too often our focus when working with data is narrowed to a limited set of choices. And often when we build applications and provide data to users, we are trying to be exacting and relevant, not random. So much of what we choose to do in software is to remove much of the randomness from our systems. Event the "browse" features are often scoped or focused in a particular topic, subject, or area.

This filtering to a particular bubble of data is one of those areas where we have tremendous power in shaping the world. The code and queries we write, the organization of our data, this will have an impact on our users, and I'm not always sure this is for the best. Perhaps overall this is more helpful, but it also serves to prevent us from viewing the forest, only seeing the trees. If you doubt this, try browsing the internet in Private mode sometimes and run searches. You might be amazed how different the Internet can look.

Join the debate, and respond to today's editorial on the forums

Provided that ANSI_NULLS are ON, The only T-SQL statement above that evaluates to true is:

If ANSI_NULLS was OFF then all those statements would evaluate to true.

NULL values are not equal to anything in SQL Server, not even other NULL values. NULLs are also not unequal to anything. Think of comparing NULLs like looking at two bags with an unknown quantity of apples and asking if both bags have the same number of apples: both "yes" and "no" are incorrect.

The exception (forgive the pun) is for set operators such as EXCEPT, INTERSECT and UNION where NULLs are treated as equal. Here' a good article by Robert Sheldon about this topic:How to Get NULLs Horribly Wrong in SQL Server

» Discuss this question and answer on the forums