Friday, November 23, 2007

Speaking of local expertise - Google Maps

According to the Washington Post this morning, Google Maps is now embracing the concept of local expertise and allowing users to fix things that are wrong on the the maps, "Wikipedia style". This is great, shows the power of web-2 technology, and illustrates exactly what my last two posts in the nature of stable, locally-driven adaptive change is about. (See Adaptation and Control and the FM 3-24, and Unity and Adaptation, continued. )

First, millions of people are recruited, effectively, at zero cost to Google, to review the maps and their findings are harvested by the maps themselves. This covers the "adaptive" part. Second, the changes are rate-limited, so that no major catastrophe could result even from a malicious user who simply wanted to cause trouble. No item can be moved more than 200 yards, and schools, hospitals, and police stations can't be moved without approval at all.

This is great! I love it!

This is the kind of change that, say, medical record systems, or for that matter, any record keeping system really needs: users who spot an error can somehow, in some way, either change the error and fix it, or at least register a question-mark on the suspect data point, and after enough question-marks are recorded, some authorized human can look at it and agree to the change and make it.

The system has another property I discussed, as does Wikipedia -- it is self-correcting and user-enforced norms provide the control stability in which this dynamic change takes place. So long as all the recent changes are clearly visible, everyone knows that some of them are in error or malicious, and need to be taken "with a grain of salt" --i.e. skeptically. But program code or data that has stood the test of time, and been there for years without anyone questioning it, can be shown in "gold" instead of newbie "green" and relied on more.

With this caveat, the ability to be a little skeptical of news, and open eyes, the social norms control mechanism works as well as a staff of highly-paid, highly-trained editors, but at much less cost, i.e, free. But, unlike the bottleneck created by a central staff of experts, regardless how expert, the system can also adapt and change if enough people on the front lines see something wrong that would surprise even the expert.

Thank you Google for illustrating how a massive-parallel, rate-governed, social-norm enforced system can beat centrally-managed, authority-enforced architecture for both reliability and adaptability. And, every time a thousand new areas of expertise are needed, the system automatically adjusts and accepts input from the crowd, way faster than a thousand new experts could be located and hired and put on staff. And, when the expert's expertise turns out to be stale, or outmoded, the system can keep on changing much more easily than trying to upgrade or fire an entrenched "expert". ( Similar to the problems with tenure at universities these days. The long time constant and depth of expertise is great, unless the world is changing so fast that those turn into liabililities, not assets.)

Here's the cite and an few quotes.

Google lets users fine tune its maps
by Catherine Rampell
Washington Post, Nov 23, 2007

Tired of getting bad directions? Google Maps, Google's popular mapping and driving directions tool, is now letting users correct its maps, Wikipedia-style.

Google, the company with the most frequently used search engine, has been beefing up its other features over the last several years in an effort to develop more user-generated, interactive content. ... The new map feature is part of this larger effort to exploit the collective knowledge of Google users.

[Said] Google spokeswoman Kate Hurowitz. "Accurate, detailed local information about neighborhoods, towns and the world around us is important to our users, and yet no one knows a town better than the people that live there."

Starting this week, users with Google accounts can move the small green arrows that pinpoint addresses on its maps. To avoid "monkeying with markers," Google said, there are restrictions on which markers may be edited: Moving a place marker more than 200 yards requires Google's approval. Users also cannot edit the locations of schools, hospitals, police stations or businesses that are registered with Google's Local Business Center. Visitors to Google Maps can choose to see the original location for all markers that have been changed.

This is the same sort of thing I've advocated Wiki's for in terms of help manuals for many computer systems. We're not sure what alien race writes the manuals, but they are not in customary English and not helpful. They really need actual local users to be able at least to stick on sticky notes, and comment on what some paragraph means, or that you need to ignore this instruction and see page 285 instead. Then, they're actually useful.

I suggested such a thing back in 1994 or so, at an American Association for Artificial Intelligence conference in Washington -- suggesting that this was a good task for the operating system to handle, and not force individual applications to figure out how to collect and share user notes and responses. At the time, it got a round of applause from a room of 1000 people who agreed with the idea. We seem to be finally getting there, with Web-2 technology and Google, not Microsoft, leading the way.

As my prior post emphasized, the interaction term can be way more important than the "self" term. Everything we do in life is "stone soup", that can be opened up for small improvements that local villagers it turns out are eager and willing to make for free. The result can be a fantastic broth starting with simply a pot of water and a "magic stone" -- and a willingness to accept input from others about small changes that might improve things a little.

Then, the power of numbers and wisdom of crowds can take over, and finish creating soup from a stone.

We have lots of stones and water -- now what we need is more players like Google to make it easy for us to help each other out just a little bit, and the social norms that hold the plan together, and reasonable rate-regulators to protect us against ill-advised rapid change or some person's private agenda. It works! It's a great model! We should do more of it!

No comments: