I just saw this new experimental site called AllYourWords which allows you to associate a word with a website. This is exactly what what RealNames did way back when before they crashed and burned with the bubble. This new site itself may not be much, but it's neat to see the idea come back because honestly, I have quite a few thoughts on this topic.
First, RealNames was an awesome concept: You'd just type in a regular word or phrase into a box and you'd go to an associated web page. Simple and powerful - I'm pretty sure there's a patent on the idea. It was full of potential at one point, and RealNames was even integrated into the Internet Explorer address bar for several years. It was like AOL Keywords for the Internet. AOL's Keywords are the same sort of thing implemented years earlier on their proprietary system too: you type in a keyword and go to a specific section of the service. I think it started out as just a shortcut for members, but quickly turned into a marketing tool for AOL and it's partners/advertisers.
As a business, both systems operated on same sort of principle: Since words are a scarce resource, companies could buy words and guarantee that their site always appears. RealNames operated as a sort of alternative Domain Registry, and though AOL had a bunch of non-commercial keywords (maybe they still do) which they used to help their members navigate their service, they also sold keywords about commercial topics as well. "Madonna" or something like that would bring you to a Modonna-themed mini-site or promotion or something.
Putting the business issues aside for a moment, I think this is How It Should All Work. There is no good reason on earth why human beings need to understand the need for "http" or "www" or "com" (or any other TLD) and dots, slashes or colons put just in the right places. And for god's sake give us back the rest of our punctuation and (most importantly) our spaces!!! Are you with me on this? I should be able to type in real words and arrive at a web site without having to remember whether the domain is a .net or a .com or god knows what else. I'm hardly alone on this - everyone uses Search Engine text boxes as their principle web navigation UI nowadays. "What's that domain name for American Airlines again? I don't know, I'll just type it into a search box and get to it that way..." Same thing for apps on the desktop like Quicksilver on the Mac: Type an app or document name, and it *poof* it appears. We aren't "searching" per se, we're simply navigating using what is more commonly seen as a search interface - but in this case we know what we want already and how to find it, we just want to get to it in the fastest and easiest way possible.
This is why it's a shame that RealNames didn't take off. We shouldn't need to use Search this way, it should just be a part of how the Internet works.
Just as an aside: Personally (and I've said this before), I think that Microsoft blew it in 1995/96 when they re-focused on the Internet and launched Internet Explorer 3.0. If instead of relying on the DNS system as it exists now, they instead implemented their own MS-DNS system, they would own the entire internet as we know it. They sort of did this with RealNames in the browser later on, but they didn't go far enough. If Microsoft had embraced the RealNames system at a system level and services level - i.e., you could FTP or SSH to a server by using its RealName - and opened up the system to be used by other OSes and systems, it would be game over for everyone else. We'd all be using RealNames right now instead of the DNS system and Keith Teare would be a bazillionaire.
But it didn't work out that way, and instead Microsoft eventually removed RealNames integration from IE and is now instead funnelling words in the toolbar to MSN Search (where they are of course, selling keywords like RealNames did...). Funnily enough, Firefox actually routes non-URLs in the address bar to the first result of a web search, which can lead to some wackiness. The other day, on my Mac using FF, I arrived at Microsoft's homepage and was completely bewildered. How did *that* happen? [I thought that] Microsoft had registered http.com, [but it was just FireFox doing a search for me and going to the first result]. Try it for yourself. It took me a few minutes to backtrack, but I eventually figured it out. [Updated: I got mixed up about what happened here. Thanks to Waldo for setting me straight.]
Anyways, IMHO the fatal flaw in the RealNames business was actually the commercial nature of the system. It's not particularly fair that anyone with a big enough wallet can come along and just buy a word, is it? It ends up generating distrust among the users of the system, and I'm sure causes considerable problems for trademark owners. Coke could buy "pepsi" and vice versa. It works for a closed system like AOL, but for an open internet system it breeds suspicion. Also, RealNames wasn't a web service (that I know of... this was probably before that was a big thing). It would've been really cool if any service out there could've queried the RealNames system with a term, and received a valid URL (or better, an actual IP address) in return.
So back to AllTheWords. It's inspiring in the sense that they're tackling the same idea again, but I would do it completely differently. First, I would make it a web service and let any system out there use the back end db, so it wasn't just a web page and a redirect. Then I would also make the registration system open, but not first-come, first serve like they're doing it. That's dumb. It'd be much better to let registered users vote on which URL a term is supposed to point in a democratic fashion. Sort of like a Digg.com for keywords. That way we'd all have more confidence in what's going on behind the scenes. Yes, it seems that PageRank does this sort of thing in an automagic way, but I'm thinking about what a crowd of people could do if they chose keyword sites deliberately (as opposed to as a by-product of linking). There's only a few hundred thousand commonly used words out there, and maybe a million or so expressions, I can't imagine it'd be that hard.
There would definitely have to be systems in place to control wild voting changes, though. The keywords need to be reliable. I never use Google's "I'm Feeling Lucky" button for example - which sends you directly to the first link in the search results - because I always have a nagging doubt about whether or not the order has changed since the last index update. That said, I do like how AllTheWords has expiration dates on their keywords. That's an interesting idea to keep terms fresh... Maybe not once a month, but once a year, all the keywords go up for a vote again... ;-)
Obviously a system like this would not be without conflicts. Look at another similar system which has to deal with the same sort of name space conflicts: WikiPedia. There's this massive ongoing battle, for example, over whether it's called Petrol or Gasoline. (Everyone besides Americans uses the former.) Still the system works. I type in a word, and I get a generally thorough explanation of the topic in return. Now, what if I wanted to type in a word or phrase and go to the best website concerning that keyword instead? For generic terms, there would obviously be issues: where does "car" go, for example? But we can all agree that Coke should go to Coca-cola's site, right? And unlike in past schemes, they really shouldn't have to pay for it. I mean there's a international trademark system already in place (as flawed as it may be) for a reason: we figured a lot of this stuff out a hundred years ago. If you own a trademark, you should get the keyword. I guess generic keywords would just be forwarded to a general search about that subject, and let the web decide what's best by links, as per usual.
And now, if you're still following this long stream of conciousness, you must be thinking of one word by now: Tags. Of course! It's all related! I am one with the Web 2.0 Universe! I've been Flickrized! Just kidding... Tags are very similar in that I'm trying to associate a real word with a web page somewhere. But most of the time, they're generic words describing the content, not necessily a page itself (unless it's a system like Del.icio.us or Yahoo's MyWeb). But still, you could count those tags up in a sort of "TagRank" fashion and use them as votes for keywords... (I'm sure someone, somewhere is already doing this.) That might work as well for generic terms.
But remember, I'm not talking about creating Yet Another Search Engine, I'm talking about a *navigation* engine: A way of using the web that your Mom would feel at home with. I have no idea how something like this would make money (charging for words is out of the question), but I think there's a real need for a service like this. International, Open, Democratic, etc.