I noticed that TechCrunch picked up a story today about how FriendFeed launched it’s real-time search feature. Now, you can get search results based on constant stream of incoming data from various FriendFeed blogs.

I’m wrestling with just how useful this feature is. Given Twitter’s recent spot in the limelight, just about all of the other big names in technology have been adopting Twitter-like behaviour in order to stay relevant. So is this in fact like Twitter? Twitter’s main function is to serve out real-time updates. FriendFeed is all about aggregating social network information, so there’s potentially some overlap there.

BUT,… there’s still this question of usefulness. Nearly every successful search paradigm starts with envisioning a use case. “What’s the user trying to do, and how can we make that thing easier for them?” That’s where this whole real-time search results thing kinda falls flat on its face, as far as I can see it. I tried to noodle some scenarios that have a heavy search flavour to them, hopped over to FriendFeed, to see what happens. (http://blog.friendfeed.com/2009/07/real-time-search-we-have-it-its-here.html) These are the questions that I attempted to find answers for:

1) What kind of pizza is there around Toronto? (It’d be interesting to see what ‘Za joints people are talking about, no?)
2) How was Transformers 2?
3) What’s this whole thing about the Mythbusters and a Twitter account?

Now, the results are impossible to replicate (*ahem… real-time results and all*), but here’s what I discovered.

1) For the pizza scenario, on the first page of results, there was a small note about Pizza Libretto, and numerous other posts that contained the words “pizza” and “Toronto” without any real… relevance. Still searching for ‘Za.

2) A plethora of links to various Transformers 2 F.A.Q.s hosted on other sites, and a healthy smattering of “Transformers 2 sucks balls” comments. Alright, that was moderately helpful, but it doesn’t take much to get me to avoid the movie theatre these days.

3) The first time I entered “Mythbusters” and “Twitter”, I got jack. I got a few posts that had references to one or the other, but nothing that gave me any real answers. Then I figured I might do the same search over again. THIS time I got 5 posts in a row that all explained the issue.

The thing is, if I want to “search” for something, I’m not prone to sticking my head amongst a group of people so I listen for certain words, which is exactly what FF’s real time search attempts to do. It is sometimes interesting to see what people are chattering on about, and you can accomplish this quite well just by letting the stream update itself as new posts come in. But in that case you’re not really searching for anything, so much as you’re merely people-watching.

The first thought from many of you is going to be “Yup, he’s on the sauce again.”

10am Drambuies notwithstanding, hear me out on this. I was thinking about the way technology is moving forward and how we’re all collectively responding to our innate need to find stuff. The internet is built on a system of requests and responses. The very first “computerized” searches came in the form of relatively crude algorithms that were designed to brute force their way through reams of information to return the closest match to a given query. Present day search systems are much, MUCH more sophisticated. Predictive algorithms, organic search mechanisms, and multi-modal search engines (voice search for example) are just some of the ways that we’ve attempted to make “finding stuff” easier.

There’s just one problem. Finding stuff isn’t really easier because overall, people still don’t know how to search. Think about it. There’s still a massive disconnect between the way that the average person thinks, and the mindset that’s required to search for something “properly” online.

Take for example, a question that’s the de facto litmus test for a search scenario; “Where should I go for dinner tonight?” Now, if *I* were to start searching for the answer (and yes, I am somewhat snobbishly excluding myself from the general public in that I think differently when it comes to online searches), I would go to Google, make sure I’m cookied for Canada by signing into my Google account, and then I’d type something like the following into the query field:

The first result is a rather simple site, but it’s a list of lots of restaurants with links to their website. For a general query, it’s not that bad. However, if you type in a much more natural query like: “Where should I go for dinner tonight?” (still signed into my google account mind you), the results are markedly different. The first result is a Yahoo Answers page from Australia:

http://au.answers.yahoo.com/question/index?qid=20071130191037AAZmpb2

Now, there are of course some very valid reasons for the differences between the two sets of results. My question is a qualitative one, for starters. By asking a natural question, I wasn’t explicitly asking for a list of restaurants. How would the system know which restaurant I “should” go to? It doesn’t necessarily know what I like (yet). The system also doesn’t really know where I am unless I tell it (either through the query itself or via GPS if I’m using a mobile to execute the search). There are in fact a lot of variables that make “natural search queries” quite difficult to handle by a system that doesn’t know who you are.

A ha. Therein lies the reason why the current method of search engine optimization (from the search engine’s perspective) is flawed. It starts off by assuming that the system is completely disconnected from the person asking the question. The goal is to provide enough information to this blind user in order to make the search result more relevant. Queue meta-tags, content strategies, and any other number of mechanisms to make this happen, and make “possibly relevant” results float to the top.

Well, why bother with that assumption in the first place? The only other activity that rivals searching online, is social networking. Ok, fine. It’s actually porn, but we’ll just assume that’s a given and continue on with the story. You’d be hard pressed to find anyone with access to the internet that didn’t have some kind of profile or identity online. Between Facebook, Twitter, … and all of those other communities, nearly everyone is linked to one of them. Everyone’s got some kind of profile online, and most of those will continue to evolve as they accumulate history. Now, before I go on I’d like to clarify that I know the reasons why profile-related search results are perceived as a scary concept to some. Regardless of the fact that many are quick to enable things like search histories and information-sharing between websites, there are some that focus on the potential dangers of doing so. There is merit to these concerns, but it also represents a whole other series of potential material to cover. So for now we’ll move forward.

Think about it like talking to your Doctor or your Lawyer, your Accountant or even your friends. All of these people could be considered systems which are profile-aware. They know you in some specific context and as a result, they can provide you with answers to pretty vague questions (“I’ve got like…a thing on my arm, and it hurts. What’s wrong?”). With the exception of your friends, they’re legally bound to keep your information private but if you think that they all adhere to those regulations 100% of the time, I’m sorry but you’re fooling yourself.

Now think about how this translates to an online search. All of a sudden, “Where should I go for dinner tonight?” doesn’t seem all that unreasonable, does it? If there is a shift in focus from making disconnected information “possibly relevant”, to making relevant information accessible, the way we all “find stuff” could drastically change in the next few years. All of the pieces are already there. The same sophisticated search mechanisms I mentioned above wouldn’t have to change at all. Social networking is already set up to gather information about us. The only bit that’s left to do is to connect the two properly.

Though, I guess the whole thing is really going to get weird when you start seeing stuff like this in response to a dinner selection query:

Thoughts?