Monday, October 14, 2013

Crash Scene #1

Check out these photo search images:

Crash Scene #1
photo search
Image by Roger Smith
On November 28, 2007, our Civil Air Patrol Squadron was called to search for the wreckages of a crash of two T-6 Texan II trainer aircraft based at Columbus Air Force Base in Columbus, Mississippi. The aircraft had collided in mid-air near an auxilliary air strip near Shuqualak, Mississippi. We dispatched an air crew including Mike Hainsey as mission pilot, Trey Breckenridge as mission scanner, and me as mission observer and photographer. Our mission was to locate the aircraft and photograph the crash sites. When we arrived on this crash site, USAF ground personnel had already located it.

After the Steve Fossett crash, I heard a lot of people who have never done aerial searches say things like "how hard can it be to find a crashed airplane?" So, how hard is it? Can you find the crash site? Identifying the red fire truck in the center of the photo doesn't count.

I guess I should also mention here that both crew members in both aircraft safely ejected, and were uninjured except for a few cuts and bruises.

Get Wowd
photo search
Image by jurvetson
Wowd has entered private beta testing, and I got some invite keys for my flickr friends if you want to give it a whirl. Feedback on the concept and implementation are welcomed.

In short, Wowd has built a fully distributed search engine. No server farm. No web crawlers. It’s like the Skype of real-time search.

Besides changing the cost, and data center energy equation for a new entrant, we think this architecture will have some interesting implications. Instead of relying on content or link analysis for page rank, which is architected – and sometimes gamed – by the publishers on the web, Wowd relies on the real-time surfing behavior of real people. It’s hard to game the behavior of the crowd in the cloud as this scales. So, you get the real-time web, the social web, and over time, the deep web. Think of all those sites that block web crawlers or are hopelessly out of date in search results today – from social networks, to eBay and Craigslist to various niches, like used car sites.

It has been quite exciting to see the product in development over the years. When the founder, Boris, first came to us, he had a bold vision and a smart team of young developers in Serbia. It reminded me of the Skype team from Estonia in so many ways. Boris and I spoke with the founders or original programmers at most of the major search companies, and they opined that it couldn’t be done (with reasonable system latency). That made it all the more interesting, and we seed financed the company. In full disclosure, I am also on the Board.

We thought that we would need many thousands of users before the distributed search engine would “work” and that early search results would be skewed by small numbers, biased by a disproportionate number of Serbs =). It is fascinating to watch in the early days, as many of the properties of the system are emergent, and the quality and performance of the system should improve with scale.

So, if you want to play with it, the link below should take you past the gates to the Wowd site. Please do so only if you are curious and willing to play with early code. It is a browser app that works across Windows, Mac and Linux. If you like it, great, if not, please keep it open and check in from time to time to see if it has improved, and of course, constructive feedback is even better. OK, so here’s the private beta Invite Link. Click on the "Get Wowd" button there to join the Wowd cloud.

• Recent article and video in readwriteweb (their invite keys ran out, fyi)
• Blogs: Wowd, Boris’ Distributed Search