This editorial was originally published on April 23, 2009. It is being re-run as Steve is on vacation.
Data quality is an issue in many of our systems that rely on humans for data entry. Even if you only import data from other sources, how reliable is the data that exists in those systems?
I found an interesting article on computer interfaces and it highlights some of the ways that have been designed to work with computers and put data into, as well as get data out of, computer systems. Most of these are probably familiar to you and there's nothing really groundbreaking in the article.
It had me thinking, however, of whether or not there's a better way to input and extract data from systems. I've been watching Star Trek: The Next Generation on DVD lately, and they primarily use voice recognition, but I see all kinds of flaws in that. My wife has worked with a lot of speech technology, and her current company makes the Dragon speech recognition software. I asked her if she thought it was good. She did, but it paled when compared to a good, old-fashioned keyboard. The rate of input and the few mistakes she makes typing far outweigh the benefits of using speech. Perhaps that will improve in the future, but I wonder if it will ever really supplant the keyboard.
Multi-touch has gained a lot of notoriety lately, especially in presentations, but I'm not sure it's a great way to get data out of a system for a single user. There are other possibilities, but in anything we develop, the rate of input as well as mistakes, are things on my mind. As a data guy, I am entirely too aware of how much work it is to clean data later.
Managing, storing, and securing data all is great, and you can do the best job in the world, but if users can't easily access and use the data, is it valuable? If it's not correct, is it useful? The way we work with computers will likely evolve, and while I'm not sure what will work best, I do know that us DBAs will always be in demand to ensure the data is as correct as it can be.