Steady, gradual change is easy to underestimate when you're up close to something.
While working on my GISWORX (Dubai) keynote for next week, I opened a few of my 2003-2004 PowerPoint presentations that examine the significance/future of web-based GIS technology on society and decision-making. Despite the fact that I don't feel like the fundamentals change all that much day-to-day, I was struck by the fact that most of my forward-looking presentation content from 2003-2004 is now happening. Core aspects of the "road ahead" sections from these presentations have arrived and I can replace my slides with real-world examples that are mainstream (if not yet ubiquitous).
Surprises as I look back? I didn't forsee the incredible shift from Java towards .NET that we've observed, I overestimated the probable future influence of WMS/WFS, and Google entering the spatial realm wasn't on our radar screen (it hadn't happened yet).
Though we sometimes make major strides forward overnight (e.g. leveraging new capabilities possible with ArcGIS Server), as I cobbled together examples of innovative developments from the last couple years I noticed that many of the "breakthroughs" happened incrementally through innovative pilot projects and the addition of relatively minor new capabilities. When examined together, it becomes clear the technology is actually evolving very quickly.