Oren Michels is CEO of Mashery.
My closest cousin had cancer surgery yesterday. Based on what we (her close family and friends) have learned, this is the sort of cancer you really, really don’t want. Of course, we don’t know that for sure. After all, when we first got the preliminary diagnosis a couple weeks ago, everyone in the family began searching for answers. Searching for clues as to her course of treatment. And her prognosis. We all wanted to find out about that.
And we all found the same sites, the same obscure papers about this obscure cancer, papers that had somehow managed to rise to the top of the SEO heap. Why these ones? Were they “right?” Is the median survival rate 14 months? Says who? When? The prognosis for someone diagnosed ten years ago is likely diffenent than that of someone diagnosed more recently. And on and on.
Search as we know it today fails us at times like this. Despite what Google says, they have not organized the world’s information and made it universally accessible and useful. Far from it. A dozen of us found the same info on Krukenberg tumors, and yet we know little more now than we knew before we’d heard of them. Although this cancer is obscure, much has been written about it, and there are profound debates going on in medical circles about how it happens, where it comes from, and how to treat it. But little of that emerges on the first page of search results, or the second, or the tenth. But because Valerie is family, we persevere to the fifteenth, the twenty-fifth. Highly unusual search engine user behavior, I know. But totally reasonable in light of this situation. Yet we still see the same content dressed in slightly different skin.
How could search be better? Make it temporal – don’t just tell me what people are linking to; tell me when they linked to it, and how new and fresh the content is – blog search does this on some level, but non-blog content has as much need to be “fresh” as blog content. Allow us to search for a topic over time – searching for “Enron” today brings up a lot of information on the collapse and bankruptcy, but little on how it was perceived and written about at its peak. Hindsight is interesting, but we learn even more by looking at contemporary accounts of its success and gleaning understanding that will allow us to pattern match to current events.
Differentiate between original content and meta sites. Speaking of meta, focus on metadata. I’m much more interested in a site or blog dealing directly with Krukenberg than with a popular cancer site that happens to mention it; accurate metadata can help here. So can some sort of measure of authority…though that is a hard one to nail down. But we should still try. Get better at deduping, and weeding out similar or even identical content from multiple providers. Make it so the first page of ten results provides a broad look at the subject from a variety of categories of sources, and then drill down on “more like this”
Better results. More relevance. More timeliness. More focus on the desired subject.
Then what? Get people to use it. That is all about one thing – web service APIs. It is highly unlikely that anyone will dethrone the current search market leaders on virality alone, and few startups have the resources to compete using traditional marketing and promotional techniques. New search, and the new content that it finds, will gain traction if existing and newly emerging high-traffic sites incorporate that search in a seamless user experience. That takes APIs that are open, with easy and flexible integration into existing sites and applications.
I look forward to our sessions. I have no doubt that we will all learn a lot. Thanks for including me!