Wednesday, August 03, 2005

The Google Search Game

A fair while back, Darius and I riffed for a short while on creating a game using Google's rather unique "OS" - primarily, the searching capabilities of Google.

My idea was that the gamers would have to be adding to Google's value while playing - probably by some kind of sorting/analysis of other sites. I was thinking of something which played similarly to the fictional "alternate reality" games such as ILoveBees: you have clues or goals, and then you go around trying to understand what there is to understand and get the next clues and goals.

The problem was, of course, that the real world is rather more boring. Trying to do something like this based on the existing internet would be, almost all the time, so easy as to be boring. You want to know the population of Botswana? It's thirty seconds away.

So the game would have to be specifically about the REALLY HARD information. It can't be a game of organizing and filing generic internet pages. That would get boring in a matter of hours, and you'd have to pay to keep people playing. You'll need to have some kind of reward, regardless, but so long as you're giving out rewards, they should be limited and for services of actual value. Bots can file most web pages fairly accurately.

What KIND of "really hard information" are we talking about? What kind of reward? What kind of chain of patterns? How can you make searching into a game?

The key is to search OUTSIDE of Google. Searching INSIDE of Google is largely pointless. Of course, Google is so large that searching outside of Google probably includes the millions of pages which Google knows about but ignores.

What sort of searching? What kind of system could be set up to keep people continuously busy and interested without requiring an absurd amount of employee work and SIMULTANEOUSLY enhancing Google's knowledge base?

How about Google: Adventure? (Googventure? Advent Google? Exploogle? Gooxplore?)

The basis is simple: the users submit locations they know about, such as restaurants, neat shops, nice overlooks - whatever strikes their fancy. Other people post reviews. A point is given for reviewing, and a point is given to the originator if the review is good.

"Terrible idea!" you should be shouting. "Scripts will make thousands of reviews."

You have Google Maps, right?

This is where the data dissection comes in.

The system automatically reviews the locations of the reviews plus the time they say they visited (you have to post a review within 24 hours). From this, you can easily determine a "flight path" for the reviewer.

Now, I'm not likely to do much reviewing or posting, but I'd love to spend a few minutes a day looking at people's "flight paths". A review of this place on Tuesday, that place on Thursday. It would be pretty clear to a human eye whether someone was a fake or not. If they get a false positive, then the real person will complain, and Google can investigate more closely.

But people will make scripts to judge that, too! In fact, they can make scripts which judge their own scripts!

Not if you require a gmail account for those people. And, if they script or bash, you can ban them. Even if you don't require gmail, their success depends mostly on their accumulation of points, which vanishes if they are banned.

Oh, they'll certainly be some scripting going on, and you'll probably need an employee to scan the flight-analysers' patterns... but you won't need to scan the umpteen thousand people entering reviews. Only the people reviewing the reviewers.

Creepily big-brothery? Maybe. It sounds like it would be a very interesting experience, and I know a lot of people would be fascinated by the chance to look at the habits of other people. A clear warning about the nature of the data review should cover the danger.

Moreover, this would be an excellent opportunity to gather data about how people act, react, think... a grand gathering of sociological and psychological data.

Mmmm. I like the idea.

The only thing which is shaky is what points are used for. I think points should weight your entries - the more points you have, the more "weight" you carry in the averaging programs. It would also not be amiss to have some small stuff available for purchase by points expenditure - such as extra gmail space, or maybe a 10% off a book at Amazon. Something of that nature.

However, just having more weight would probably be enough for most people.

I think it's a neat idea! As an infoholic, I think it's a GREAT idea!

It's probably only useful for urban centers, though: you need to have a Google-friendly population ready and willing to review locations. I'm not so sure how many there would be.

Of course, you could apply the same pattern anywhere. The basic idea is that you make a layer of people who essentially do the dirty work for you, finding all the problems on the public layer. Then you just have to scan the scanners, instead of all the people. You could do it with books, with restaurants, with web sites - anything that people enter a lot of data for and you want to keep from being script-bombed.

No comments: