David Precopio recently blogged here about Portals, wikis and other transitory technical phenomena. In 2000, portals represented a vision for how a new, user-driven web was going to develop - one which paradoxically required lots of new proprietary software and teams of developers.
Shortly after 2000, application server vendors looking for a way to boost sagging license revenues (what do you mean we can't charge a premium for a commodity J2EE product?!) glommed onto portals as a handy revenue extender/floor wax. They immediately shifted into high gear, selling the whitening and brightening qualities of their "collaborative" web products.
Into the middle of all this crept the real collaborative web products. Using social media as their proving ground, they pioneered user generated content, interactive UIs (remember those nice client/server UIs - they're back as AJAX) and web services (remember those nice objects - they're back as services).
We met a company recently that referred to their 20 person portal project as "our latest boat anchor." Portals are just the latest example that technology vendors often have the right idea about what the future needs but too often try to implement the tools of the future from the worn out components of their past successes.